PIXEL CLASSIFICATION A FUZZY-GENETIC APPROACH
光敏色素互作因子的英文
光敏色素互作因子的英文The English term for 光敏色素互作因子 is "Photoreceptor Interacting Factor" (PIF). In this article, we will delve into the details of PIF, its function, and mechanisms in more than 1200 words.Introduction to Photoreceptor Interacting Factor (PIF):PIF Family Proteins:Role of PIFs in Photomorphogenesis:Photomorphogenesis refers to the light-induced developmental changes that occur throughout a plant's life cycle. PIFs act as key integrators of light signaling pathways and positively or negatively regulate the morphological changes associated with photomorphogenesis. In the dark, PIFs, especially PIF3 and PIF4, are highly accumulated and inhibit seedling photomorphogenesis by repressing the expression of light-responsive genes. Upon exposure to light, PIFs are rapidly degraded, allowing photomorphogenesis to proceed. This degradation is mediated by the light-activated photoreceptors, resulting in the activation of target genes involved in chlorophyll biosynthesis, leaf expansion, and other light-dependent processes.PIFs and the Phytochrome Signaling Pathway:Phytochromes are a class of photoreceptors that detect red and far-red light. They play a vital role in the regulation ofseed germination, de-etiolation, flowering, and shade avoidance responses. PIFs interact with phytochromes to modulate their activity. Red light activates phytochromes, resulting in their conversion from the inactive Pr form to the active Pfr form. PIFs, specifically PIF3 and PIF4, bind to the Pfr form and prevent its entry into the nucleus. This interaction sequesters phytochromes in the cytoplasm, impeding their ability to regulate gene expression associated with light responses. Consequently, PIF repression is relieved upon degradation or inactivation of PIFs, allowing downstream gene expression and photomorphogenesis to occur.PIFs and the Cryptochrome Signaling Pathway:PIFs, Flowering Time, and The Circadian Clock:Conclusion:。
医学图像分割方法综述
医学图像分割方法综述林瑶，田捷1北京,中国科学院自动化研究所人工智能实验室,100080摘要: 图像分割是一个经典难题，随着影像医学的发展，图像分割在医学应用中具有特殊的重要意义。
本文从医学应用的角度出发，对医学图像分割方法，特别是近几年来图像分割领域中出现的新思路、新方法或对原有方法的新的改进给出了一个比较全面的综述，最后总结了医学图像分割方法的研究特点。
关键词：医学图像分割 综述1．背景介绍医学图像包括CT 、正电子放射层析成像技术（PET ）、单光子辐射断层摄像（SPECT ）、MRI （磁共振成像技术）、Ultrasound （超声）及其它医学影像设备所获得的图像。
随着影像医学在临床医学的成功应用，图像分割在影像医学中发挥着越来越大的作用[1]。
图像分割是提取影像图像中特殊组织的定量信息的不可缺少的手段，同时也是可视化实现的预处理步骤和前提。
分割后的图像正被广泛应用于各种场合，如组织容积的定量分析，诊断，病变组织的定位，解剖结构的学习，治疗规划，功能成像数据的局部体效应校正和计算机指导手术[2]。
所谓图像分割是指将图像中具有特殊涵义的不同区域区分开来，这些区域是互相不交叉的，每一个区域都满足特定区域的一致性。
定义 将一幅图像，其中g x y (,)0≤≤x Max x _,0≤≤y Max y _，进行分割就是将图像划分为满足如下条件的子区域...：g 1g 2g 3 (a) ，即所有子区域组成了整幅图像。
(b) 是连通的区域。
g k (c) ，即任意两个子区域不存在公共元素。
(d) 区域满足一定的均一性条件。
均一性（或相似性）一般指同一区域内的像素点之间的灰度值差异较小或灰度值的变化较缓慢。
g k 如果连通性的约束被取消，那么对像素集的划分就称为分类（pixel classification），每一个像素集称为类（class）。
在下面的叙述中，为了简单，我们将经典的分割和像素分类通称为分割。
函数功能解释（英文版）
tifs2seq
Create a MATLAB sequence from a multi-frame TIFF file.
Geometric Transformations
imtransf orm2
2-D image transformation with fixed output location.
Computes and displays the error between two matrices. Decodes a TIFS2CV compressed image sequence. Decodes a Huffman encoded matrix. Builds a variable-length Huffman code for a symbol source. Compresses an image using a JPEG approximation. Compresses an image using a JPEG 2000 approximation. Computes the ratio of the bytes in two images/variables. Decodes an IM2JPEG compressed image. Decodes an IM2JPEG2K compressed image. Decompresses a 1-D lossless predictive encoded matrix. Huffman encodes a matrix. Compresses a matrix using 1-D lossles predictive coding. Computes a first-order estimate of the entropy of a matrix. Quantizes the elements of a UINT8 matrix. Displays the motion vectors of a compressed image sequence. Compresses a multi-frame TIFF image sequence. Decodes a variable-length bit stream.
地统计与遥感---专业英语词汇
地统计以及遥感英文词汇300个：gray level co-occurrence matrix algorithm灰度共生矩阵算法characteristic of atmospheric transmission 大气传输特性earth resources technology satellite,ERTS 地球资源卫星Land-use and land-over change 土地利用土地覆盖变化Multi-stage stratified random sample 多级分层随机采样Normalized Difference Vegetation Index归一化植被指数Soil-Adjusted Vegetation Index土壤调整植被指数Modified Soil-Adjusted Vegetation Index修正土壤调整植被指数image resolution ,ground resolution影象分辨力(又称“象元地面分辨力”。
指象元地面尺寸。
) remote sensing information transmission遥感信息传输remote sensing information acquisition遥感信息获取multi- spectral remote sensing technology多光谱遥感技术Availability and accessibility 可用性和可获取性Association of Geographic Information (AGI) 地理信息协会Difference Vegetation Index差值植被指数image quality 影象质量Enhanced Vegetation Index增强型植被指数Ratio Vegetation Index比值植被指数Spatial autocorrelation 空间自相关Lag Size 滞后尺寸Ordinary kriging 普通克里金Indicator kriging 指示克里金Disjunctive kriging 析取克里金Simple kriging 简单克里金Bivariate normal distributions 双变量正态分布Universal kriging 通用克里金conditional simulation 条件模拟image filtering 图像滤波optimal sampling strategy 最优采样策略temporal and spatial patterns 时空格局Instantaneous field-of-view瞬时视场角azimuth 方位角wavelet transform method 小波变换算法priori probability 先验概率geometric distortion 几何畸变active remote sensing主动式遥感passive remote sensing 被动式遥感multispectral remote sensing多谱段遥感multitemporal remote sensing 多时相遥感infrared remote sensing 红外遥感microwave remote sensing微波遥感quantizing，quantization量化sampling interval 采样间隔digital mapping数字测图digital elevation model,DEM 数字高程模型digital surface model,DSM 数字表面模型solar radiation spectrum太阳辐射波谱atmospheric window 大气窗atmospheric transmissivity大气透过率atmospheric noise 大气噪声atmospheric refraction 大气折射atmospheric attenuation 大气衰减back scattering 后向散射annotation 注解spectrum character curve 波谱特征曲线spectrum response curve 波谱响应曲线spectrum feature space波谱特征空间spectrum cluster 波谱集群infrared spectrum 红外波谱reflectance spectrum反射波谱electro-magnetic spectrum 电磁波谱object spectrum characteristic地物波谱特性thermal radiation 热辐射microwave radiation微波辐射data acquisition数据获取data transmission数据传输data processing 数据处理ground receiving station地面接收站environmental survey satellite环境探测卫星geo-synchronous satellite地球同步卫星sun-synchronous satellite太阳同步卫星satellite attitude卫星姿态remote sensing platform 遥感平台static sensor 静态传感器dynamic sensor动态传感器optical sensor光学传感器microwave remote sensor微波传感器photoelectric sensor光电传感器radiation sensor辐射传感器satellite-borne sensor星载传感器airborne sensor机载传感器attitude-measuring sensor 姿态测量传感器image mosai图象镶嵌c image digitisation图象数字化ratio transformation比值变换biomass index transformation生物量指标变换tesseled cap transformation 穗帽变换reference data 参照数据image enhancement 图象增强edge enhanceme边缘增强ntedge detection边缘检测contrast enhancement反差增强texture enhancement 纹理增强ratio enhancement 比例增强texture analysis 纹理分析color enhancement 彩色增强pattern recognition 模式识别classifier 分类器supervised classification监督分类unsupervised classification非监督分类box classifier method 盒式分类法fuzzy classifier method 模糊分类法maximum likelihood classification最大似然分类minimum distance classification最小距离分类Bayesian classification 贝叶斯分类Computer-assisted classification机助分类illumination 照度principal component analysis 主成分分析spectral mixture analysis 混合像元分解fuzzy sets 模糊数据集topographic correction 地形校正ground truth data 地面真实数据Tasselled cap 缨帽变换Artificial neural networks 人工神经网络Visual interpretation 目视解译accuracy assessment 精度评价Omission error漏分误差commission error 错分误差Multi-source data 多源数据heterogeneous 非均质的Training sample 训练样本ancillary data 辅助数据dark-object subtraction 暗目标相减法discriminant analysis 判别分析‘salt and pepper’ effects 椒盐效应spectral confusion光谱混淆Cluster sampling 聚簇采样systematic sampling 系统采样Error matrix误差矩阵hard classification 硬分类Soft classification 软分类decision tree classifier 决策树分类器Spectral angle classifier 光谱角分类器support vector machine支持向量机Fuzzy expert system 模糊专家系统endmember spectral端元光谱Future extraction 特征提取image mosaic图像镶嵌density slicing密度分割least squares correlation 最小二乘相关data fusion 数据融合Image segmentation图像分割urban remote sensing 城市遥感atmospheric remote sensing大气遥感geomorphological remote sensing地貌遥感ground resolution地面分辨率ground date processing system地面数据处理系统ground remote sensing地面遥感object spectrum characteristic地物波谱特性space characteristic of object地物空间特性geological remote sensing地质遥感multispectral remote sensing多光谱遥感optical remote sensing technology光学遥感技术ocean remote sensing海洋遥感marine resource remote sensing海洋资源遥感aerial remote sensing航空遥感space photography航天摄影space remote sensing航天遥感infrared remote sensing红外遥感infrared remote sensing technology红外遥感技术environmental remote sensing环境遥感laser remote sensing激光遥感polar region remote sensing极地遥感visible light remote sensing可见光遥感range resolution空间分辨率radar remote sensing雷达遥感forestry remote sensing林业遥感agricultural remote sensing农业遥感forest remote sensing森林遥感water resources remote sensing水资源遥感land resource remote sensing土地资源遥感microwave emission微波辐射microwave remote sensing微波遥感microwave remote sensing technology微波遥感技术remote sensing sounding system遥感测深系统remote sensing estimation遥感估产remote sensing platform遥感平台satellite of remote sensing遥感卫星remote sensing instrument遥感仪器remote sensing image遥感影像remote sensing cartography遥感制图remote sensing expert system遥感专家系统active remote sensing主动式遥感passive remote sensing被动式遥感resource remote sensing资源遥感ultraviolet remote sensing紫外遥感attributive geographic data 属性地理数据attributes, types 属性，类型Geographic database types 地理数据库类型attribute data 属性数据Geographic individual 地理个体Geographic information (GI) 地理信息Exponential transform指数变换false colour composite 假彩色合成Image recognition 图像识别image scale 图像比例尺Spatial frequency 空间频率spectral resolution 光谱分辨率Logarithmic transform对数变换mechanism of remote sensing 遥感机理adret 阳坡beam width波束宽度biosphere生物圈curve fitting 曲线拟合geostationary satellite对地静止卫星glacis缓坡Field check 野外检查grating 光栅gray scale 灰阶Interactive 交互式interference干涉inversion 反演Irradiance 辐照度landsatscape 景观isoline 等值线Lidar激光雷达landform analysis地形分析legend 图例Map projection地图投影map revision地图更新Middle infrared中红外Mie scattering 米氏散射opaco 阴坡orbital period 轨道周期Overlap重叠parallax 视差polarization 极化Phase 相位pattern 图案quadtree象限四分树Radar returns雷达回波rayleigh scattering 瑞利散射reflectance 反射率Ridge山脊saturation 饱和度solar elevation太阳高度角Subset 子集telemetry遥测surface roughness表面粗糙度Thematic map专题制图thermal infrared热红外uniformity均匀性Upland 高地vegetal cover 植被覆盖watershed流域White plate白板zenith angle天顶角radiant flux 辐射通量Aerosol 气溶胶all weather 全天候angle of field 视场角Aspect 坡向atmospheric widow大气窗口atmospheric 大气圈Path radiance 路径辐射binary code二进制码black body 黑体Cloud cover云覆盖confluence 汇流点diffuse reflection漫反射Distortion畸变divide分水岭entropy熵meteosat气象卫星bulk processing粗处理precision processing精处理Bad lines 坏带single-date image单时相影像Decompose 分解threshold 阈值relative calibration 相对校正post-classification 分类后处理Aerophotograph 航片Base map 底图muti-temporal datasets 多时相数据集detector 探测器spectrograph 摄谱仪spectrometer 波谱测定仪Geostatistics 地统计Semivariogram 半方差sill 基台Nugget 块金Range 变程Kriging 克里金CoKriging 共协克里金Anisotropic 各向异性Isotropic 各向同性scale 尺度regional variable 区域变量transect 横断面Interpolation 插值heterogeneity 异质性texture 纹理digital rectification数字纠正digital mosaic 数字镶嵌image matching影像匹配density 密度grey level灰度pixel,picture element 象元target area目标区searching area 搜索区Spacelab 空间实验室space shuttle航天飞机Landsat陆地卫星Seasat 海洋卫星Mapsat测图卫星Stereosat 立体卫星aspatial data 非空间数据。
易康培训教程
单一的规则中的主要功能
算法 算法作用的影像对象域 算法参数
在影像中一个单一的规则能使一个具体的算法应用到一 个具体特定的区域，条件信息为选择特定区域的分类或合并 提供了很好的语义信息。
如图：规则窗口中显示了一个规则流程
当您用眼睛观察一个区域时，您会从特定区域周围再观察到局部， 您可能通过观察该特定区域的特殊尺寸、形状、颜色等，所有这些使 您把它和一个具体的事物联系到一起最终判定出是什么物体。
例如：
您看到的是两个圆形物体，您只能分类出是两个圆形、蓝色的物体
您会立刻把它们三者联系到一起来，左图两侧的物体为刀子和叉子 ，而中间的蓝色圆形物体为盘子。右图圆形则为车轮。
1、从菜单/文件选择Open projects 或点击Open Project按钮
2、浏览QB_Yokosuka文件夹，选择我们练习要用到的样例工程 QB_Yokosuka.dpr Yokosuka工程文件是一整景Quickbird影像的 一部分，工程过程中已经被应用在规则树中。 这个工程中包含了一个影像对象层，同时该影像对象已经被分了 类。在默认情况下一个显示影像数据的窗口以及规则树和影像对象 信息的窗口被打开。
2 .2.1 显示影像对象
浏览设置（View Settings）工具条
1、 确定浏览的图层（View Layer）是否选中按该按钮 2、 选择显示或隐藏形状（轮廓）按钮 来显示该层中所有 影像对象的形状（轮廓），从而可以判断是否与影像的 地物外形一致。 3、 显示对象的均值图，可选择对象均值图和像素图按钮
一个特征是目标对象的相关信息的表述。
遗传学第十二章遗传与发育课件.ppt
（二）细胞谱系示意图（lineage diagram）表明 每个体细胞的生活史
◇ 雌雄同体的C.elegans的完整的细胞谱系
（三） 卵孔（vulva）形成的遗传分析
1、 C.elegans的生殖系统
2、卵孔的发育方式： 细胞通过信号分子来改变其它细胞基因组
1、母体效应基因（maternal-effect genes)
◇ 母性效应基因编码转录因子、受体和调 节翻译的蛋白，在卵子发生（oogenesis） 中转录，产物由滋养层细胞合成并输送入 卵母细胞中，由细胞骨架瞄定在细胞质的 不同区域，沿前-后轴呈梯度（gradient） 分布。
◇ 母性效应基因产物的梯度起始胚胎发育, 突变研究指出，调节果蝇发育的母性效应 基因约40个。如：bicoid, nanos.
complex（ANTP-C）和 双胸复合体 Bithorax complex（BX-C）。
胚胎体节的划分确定后，同源异形基因负责确定每个体节的特征 结构。若发生突变，会使一个体节上长出另一个体节的特征结构，如 Pb基因的突变使触须变成腿和UbX突变成四翅果蝇。
同源异形基因的结构特点：
❖具有同源异形框。 ❖有多个启动子和转录起始点。 ❖有多个内含子。 ❖同源异形基因之间的相互排斥。
（二）早期胚胎发育 P376
1、配子发育与受精 精卵形成、受精信号
2、卵裂与囊胚形成 除哺乳动物以外，多数动物合子基因组暂时不表
达，即转录处于抑制状态。卵裂所需的物质主要来源 于受精前储存在卵细胞质中的母源性分子，包括 mRNA和蛋白质等，及其受精后的翻译产物。这些母 源性物质能够支持受精卵发育到囊胚期（Blastocyst )。 3、胚轴建立与图式形成 背腹轴(D-V)、前后轴(A-P)、左右轴。
什么是像素如何计算（Whatisthepixelcount）
什么是像素如何计算（What is the pixel count）What is a pixel? How do you calculate it?"Pixel" (Pixel) by Picture (image) and Element (elements) of the two word letters, is a unit used to calculate the digital image, like photography picture, digital imaging also has dense gradation continuity, if we put the image on several times, will find these are small points of continuous tone by many similar color composition, the smallest unit of these small point is the image pixel structure "(Pixel). The smallest graphic unit can display a single coloring point on the screen. The higher the pixels, the more colorful they have, the more they can express the sense of color.A pixel is usually considered as the smallest complete sampling of an image. This definition is very relevant to the context. For example, we can say that the pixels in a visible in the image (e.g. a printed page) or pixel represented by electronic signals, or digital pixel representation, or display pixels, or digital camera (photosensitive element) in pixels. This list can also add many other examples, according to the context, there will be some more precise synonyms, such as pixel, sampling points, byte, bit, spot, three points, a superset of stripe set, windows, etc.. We can also discuss pixels abstractly, especially when using pixels as resolutions, such as 2400 pixels per inch (PPI) or 640 pixels per line. Dots are sometimes used to represent pixels, especially computer marketers, so PPI is sometimes written as DPI (dots per inch).The more pixels are used to represent an image, the closer the result is to the original image. The number of pixels in an imageis sometimes referred to as the image resolution, although the resolution has a more specific definition. The pixel can be represented by a number, such as a 3 megapixel digital camera ", it has rated three million pixels, or represented by a pair of numbers, such as" 640 x 480 display ", it has 640 pixels and 480 pixels horizontal vertical (like VGA display, so that) its total number is 640 * 480 = 307200 pixels.Color sampling points of digital images (such as JPG files commonly used in web pages) are also called pixels. Depending on the computer monitor, these may not be in one-to-one correspondence with the screen pixels. In areas where this distinction is obvious, the points in the image file are closer to the texture element.In computer programming, pixels are called bitmaps or raster images. The technology of grating coming from analog TV at one time. Bitmap images can be used to encode digital images and some types of computer generated art.Primitive and logical pixelsBecause the resolution of most computer monitors can be adjusted by the computer's operating system, the pixel resolution of the display may not be an absolute standard.Modern LCDs have an original resolution according to the design, which represents the perfect match between pixels and the three element group. The cathode ray tube also uses the red green blue fluorescence three element group, but they do not coincide with the image pixels, so they can not be compared with the pixel.For this display, the original resolution produces the most detailed image. But because the user can adjust the resolution, the display must be able to display other resolutions. Non original resolution must be achieved by fitting resampling on the LCD screen, using interpolation algorithm. This often makes the screen look broken or fuzzy. For example, the original resolution is 1280 * 1024 display resolution is 1280 * 1024 looks the best, also by using several physical elements to represent a group of three to 800 x 600 pixel display, but may not be able to fully display 1600 * 1200 resolution, because the three physical elements were not enough.The pixels can be rectangular or square. There is a number called length width ratio, which is used to express pixels in many ways. For example, the aspect ratio of 1.25:1 means that the width of each pixel is 1.25 times its height. The pixels on a computer display are usually square,But the pixels used for digital images have rectangular aspect ratios, such as variants of PAL and NTSC for CCIR 601 digital image standards, as well as the corresponding Widescreen format.Each pixel of a monochrome image has its own brightness. 0 is usually black, and the maximum is usually white. For example, in a 8 bit image, the maximum unsigned number is 255, so this is the white value.In color images, each pixel can use its hue, saturation, and brightness to represent, but usually using RGB intensity torepresent (see RGB).Bit per pixelThe number of colors that a pixel can represent depends on the bit per pixel (BPP). This maximum number can be obtained by taking the power of two color depth. For example, the common values are:8 BPP [28=256; (256 color)];16 BPP [216=65536; (65536 color, called high color)];24 BPP [224=16777216; (16777216 colors, called true colors)];48 BPP [248=281474976710656; (281474976710656 color, for many professional scanners).256 colors or less color graphics often to block or plane format stored in memory, where each pixel memory is called to a color array index value. These patterns are sometimes referred to as index patterns. Although there are only 256 colors each time, the 256 colors are selected from a palette with a large selection, usually 16 megabytes. Changing the color values in the palette gives you an animation effect. Windows 95 and windows 98 are probably the most famous examples of this type of animation.For more than 8 of the depth of the digital three component (RGB) the sum of the respective digit. A depth of 16 bits is usually divided into 5 bits, red and 5 bit blue, 6 bit green (eyes aremore sensitive to green). The depth of 24 bits is generally 8 bits per component. In some systems, the 32 bit depth is also optional: this means that 24 bit pixels have 8 bits of extra digits to describe transparency. In older systems, 4bpp (16 colors) is also very common.When an image file is displayed on the screen, the number of pixels per pixel for the raster text and for the display can be different. Some raster image file formats have more color depth than other formats. For example, the GIF format has the maximum depth of 8 bits, while the TIFF file can handle 48 bit pixels. No display can display 48 bit color, so this depth is usually used for special professional applications, such as film scanners and printers. This file uses 24 bit depth rendering on the screen.Sub-pixelMany display and image acquisition systems cannot display or perceive different color channels at the same point for different reasons. This problem is usually solved by multiple sub pixels, each sub-pixel processing a color channel. For example, a LCD display usually decomposes 3 pixels per pixel horizontally. Most LED monitors decompose each pixel into 4 sub pixels, one red, one green, and two blue. Most digital camera sensors also use sub-pixel, which is realized by color filter. (CRT display using RGB fluorescent spots, but they are not aligned and the pixels of the image, it is not known as sub pixels).For a system of sub pixels, there are two different treatments:sub pixels can be ignored, the image pixels as the minimum elements can be accessed, or sub pixels are included in the rendering, which requires analysis and more processing time, but can provide better image in some cases.The latter is used to improve the appearance resolution of color displays. This technique, known as subpixel rendering, use pixel geometry respectively for operon pixels, for planar display original resolution in terms of the most effective (because the display pixel geometry is usually fixed and known). This is a form of anti aliasing,It is mainly used to improve text display. Microsoft's ClearType, available on Windows XP, is an example of this technology.megapixelA megapixel is one million pixels, which is usually used to express the resolution of a digital camera. For example, a camera can use a resolution of 2048 x 1536 pixels, usually referred to as "3 million 100 thousand pixels" (2048 x 1536 = 3145728).Digitally used photosensitive electronic devices, or coupled charge devices (CCDs) or CMOS sensors, which record the brightness level of each pixel. In most digital cameras, CCD uses some sort of colored filter, with red, green, blue regions in the Bayer filter, so that the photosensitive pixel can record the brightness of a single primary color. The camera interpolates the color information of adjacent pixels. Thisprocess is called de-mosaic, and then the final image is created. In this way, the final color resolution of a x megapixel image in a digital camera may last only 1/4 of the resolution of the same image in the scanner. In this way, a picture of a blue or red object tends to blur over a gray object. Green objects don't seem blurry, because green is assigned more pixels (because the eyes are green). See the detailed discussion of [1].As a new development, Foveon X3 CCD uses three layers of image sensor in each pixel RGB intensity detection. This structure eliminates the need for understanding and eliminates the associated aliasing of images, such as high contrast color blurring of edges.Similar conceptsSeveral other types of concepts, such as volume elements (voxel), texture elements (texel) and curved surface elements (surfel), are derived from the idea of pixels, which are used in other computer graphics and image processing applications.Pixel of digital cameraPixels are the most important indicators of digital cameras. Pixels refer to the resolution of a digital camera. It is determined by the number of photosensitive elements on the photoelectric sensor in the camera, and a photosensitive element corresponds to one pixel. So the bigger the pixel, the more the light-sensitive element, the greater the cost.The image quality of digital camera is determined by pixels.The bigger the pixel, the greater the resolution of the picture, and the size of the print does not decrease the print quality. Early digital cameras were less than 1 million pixels. From the second half of 1999, 2 million pixel products have gradually become the mainstream of the market. The current development trend of digital cameras, pixels like PC CPU main frequency, there is a growing momentum.In fact, from the point of view of market classification, for the popularization of products, considering the factors of price performance, the pixel is not bigger, the better. After all, 2 million pixel products have been able to meet most of the current consumer applications. Therefore, most manufacturers in high-end digital camera pursuit of high pixel at the same time, the current output is still the most popular oriented million pixel products. Professional digital cameras, there are more than 100 million pixel level products. And 3 million pixel level products, with the CCD (imaging chip) manufacturing technology progress and cost of further decline, will soon become the mainstream of the consumer market.In addition, it is worth noting that the current digital camera products, in terms of pixel nominal divided into CCD pixels and software optimized pixels, the latter is much higher than the former. For example, a brand digital camera currently popular, its CCD pixel is 2 million 300 thousand, and the software optimized pixel can reach 3 million 300 thousand.Pixel drawingPixels are actually made up of many dots.We are here to say "pixel" is not and vector corresponding to the dot matrix image, an image icon style but refers to, this style emphasizes the image clear outline, bright colors, and the other is the cartoon image, so many friends love.Manufacturing method of a pixel map almost no aliasing method to draw smooth lines, so often used.Gif format, and the picture is often in a dynamic form. But due to its special production process, if you change the size of the picture, it is difficult to guarantee the style.The application range of pixel painting is quite extensive, from childhood to play NES FC home screen until today's GBA palm machine; black and white pictures from mobile phone until today full-color handheld computer; even if we, in the face of the computer is everywhere filled with all kinds of software standard pixel map. Now pixel painting is becoming an art, deeply shocked you and me.Effective pixel valueFirst of all, we should make it clear that the actual pixel value of a digital photograph is different from the pixel value of the sensor. Take a general inductor, for example, each pixel has a photodiode, which represents a pixel in the picture. For example, a digital camera with 5 million pixels, whose sensor can output an image with a resolution of 2560 x 1920 - in fact, this value is only equivalent to 4 million 900 thousand effective pixels. Other pixels around the effective pixel are responsible for the other work, such as deciding what black is".Most of the time, not all pixels on the sensor can be used. SONY F505V is one of the classic cases. SONY F505V sensor has 3 million 340 thousand pixels, but it most intelligent output 1856 x 1392, that is, 2 million 600 thousand pixels of the image. The reason is that SONY put new sensors larger than the old into the old digital camera, resulting in the size of the sensor is too large, the original lens completely covered each pixel in the inductor.Therefore, digital cameras use digital sensors to produce digital images with the pixel values of the inductor pixels larger than the effective pixels. In today's market, which is constantly pursuing high pixel environments, digital camera manufacturers often object to higher numerical inductor pixels rather than effective pixels that reflect the actual image sharpness.Sensor pixel interpolationIn general, each pixel in the different positions of the sensor constitutes each pixel in the picture. For example, a picture of 5 million pixels is measured and processed by 5 million pixels in the sensor to enter the shutter light (other pixels outside the valid pixels are only responsible for the calculation). But sometimes we can see a digital camera with only 3 million pixels, but output 6 million pixels! In fact, there is no false place, but the camera in the sensor 3 million pixel measurement based on the calculation and interpolation, increase the picture pixels.When a photographer shooting photos in JPEG format, the imagingquality of this "camera expansion" will expand on the computer than we good, because "the camera is in the picture is not expanded to be compressed into JPEG format before. Those who have the experience of digital photo processing know clearly that expanding the JPEG picture in the computer will make the picture delicate and smooth, and decline rapidly. Although the digital camera image sensor than the interpolated pixel normal output picture quality is good, but the interpolated picture file size than the normal output picture is much larger (such as 3 million sensor pixel interpolation is 6 million pixels, the final input memory card pictures 6 million pixels). So the interpolation of high pixels doesn't seem to have much merit, and interpolation is like using digital zoom - it doesn't create details that the original pixel cannot record.CCD total pixelCCD pixel is a very important index, because the manufacturers use different technologies, so the manufacturers nominal CCD pixels do not directly correspond to the actual pixel camera, so when you buy a digital camera to look at the camera with the actual total number of pixels. Generally speaking, the total pixel level reached 3 million or so you can meet the general application, generally 2 million pixels, 100 VIENTIANE PLAIN products can also meet the low-end use of high pixel digital camera can get higher quality photos,Now some companies have launched 6 million pixel level ordinary digital cameras.。
基于注意力和多标签分类的图像实时语义分割
第33卷第1期计算机辅助设计与图形学学报Vol.33No.1 2021年1月Journal of Computer-Aided Design & Computer Graphics Jan. 2021基于注意力和多标签分类的图像实时语义分割高翔, 李春庚*, 安居白(大连海事大学信息科学技术学院大连116026)(********************.cn)摘要: 针对现阶段很多实时语义分割算法分割精度低, 尤其对边界像素分割模糊的问题, 提出一种基于跨级注意力机制和多标签分类的高精度实时语义分割算法. 首先基于DeepLabv3进行优化, 使其达到实时运算速度. 然后在此网络基础上增加跨级注意力模块, 使深层特征为浅层特征提供像素级注意力, 以抑制浅层特征中不准确语义信息的输出; 并在训练阶段引入多标签分类损失函数辅助监督训练. 在Cityscapes数据集和CamVid数据集上的实验结果表明, 该算法的分割精度分别为68.1%和74.1%, 分割速度分别为42帧/s和89帧/s, 在实时性与准确性之间达到较好的平衡, 能够优化边缘分割, 在复杂场景分割中具有较好的鲁棒性.关键词: 卷积神经网络; 实时语义分割; 多标签分类; 跨级注意力机制中图法分类号: TP391.4 DOI: 10.3724/SP.J.1089.2021.18233Real-Time Image Semantic Segmentation Based on Attention Mechanism and Multi-Label ClassificationGao Xiang, Li Chungeng*, and An Jubai(College of Information Sciences and Technology,Dalian Maritime University,Dalian 116026)Abstract:Improving the accuracy is the goal in real-time semantic segmentation, especially for fuzzy boundary pixel segmentation. We proposed a high-precision and real-time semantic segmentation algorithm based on cross-level attention mechanism and multi-label classification. The procedure started with an optimi-zation of DeepLabv3 to achieve real-time segmentation speed. Then, a cross-level attention module was added, so that the high-level features provided pixel-level attention for the low-level features, so as to inhibit the out-put of inaccurate semantic information in the low-level features. In the training phase, the multi-label classifi-cation loss function was introduced to assist the supervised training. The experimental results on Cityscapes dataset and CamVid dataset show that the segmentation accuracy is 68.1% and 74.1% respectively, and the segmentation speed is 42frames/s and 89frames/s respectively. It achieves a good balance between segmenta-tion speed and accuracy, can optimize edge segmentation, and has strong robustness in complex scene seg-mentation.Key words: convolutional neural networks; real-time semantic segmentation; multi-label classification; cross-level attention mechanism收稿日期: 2020-02-16; 修回日期: 2020-05-23. 基金项目: 国家自然科学基金(61471079). 高翔(1994—), 女, 硕士研究生, 主要研究方向为深度学习图像语义分割; 李春庚(1969—), 男, 博士, 副教授, 硕士生导师, 论文通讯作者, 主要研究方向为数字图像处理、基于视频的运动目标追踪; 安居白(1958—), 男, 博士, 教授, 博士生导师, 主要研究方向为模式识别、海上遥感图像分析.60 计算机辅助设计与图形学学报第33卷图像语义分割是计算机视觉的一项重要技术, 相比图像分类和目标检测, 它是一种更细粒度的像素级分类技术[1], 该技术在生产环境中具有实现成本低、部署方便的优点, 因此在无人驾驶、机器人视觉等领域[2-3]常常被应用于可行驶区域的感知系统, 这些应用领域对快速交互或响应速度有很高的要求.文献[4]提出使用全卷积神经网络(fully con-volutional networks, FCN)实现端到端的语义分割, 通过卷积和池化层对输入图像逐步下采样获得具有强鲁棒性的特征, 但也导致特征分辨率降低, 对目标边界的分割不够精细. 此后, 为更加精确地恢复高分辨率的特征, 文献[5-6]使用编码器来获得深层特征的语义信息, 使用解码器融合浅层和深层特征, 逐步恢复空间和细节信息. 此外, 文献[7]提出放弃编码器最后2次下采样操作, 使用空洞卷积保持算法的整体感受野不变, 并在网络末端增加全连接条件随机场进一步精细化网络的分割结果. 为避免特征图分辨率变小、定位精度过低等问题, DeepLabv2[8]将空洞卷积与空间金字塔池化方法结合, 提出空洞空间金字塔池化(atrous spatial pyramid pooling, ASPP)模块整合多尺度特征、增大感受野, 进而提高分割精度. 基于以上2种方法, DeepLabv3[9]进一步讨论了空洞卷积的并联和串联方式对算法分割效果的影响, 改进ASPP模块, 进而获取不同感受野信息, 提高了分割不同尺度目标的能力, 取得更好的语义分割效果. 文献[10]认为, 丰富的上下文信息可以增强网络的信息丰富度与类别区分度, 使得网络模型具有更好的语义分割能力.以上工作主要解决因网络下采样造成特征的空间信息丢失问题, 虽然提高了分割精度, 但是分割速度较慢, 无法满足实时分割任务需求. 目前实时语义分割算法大都以牺牲分割精度为前提达到实时分割速度. 实时语义分割算法SegNet[11]采用编码器-解码器结构, 在编码过程中通过多次卷积与池化运算提取特征, 在解码过程中使用池化索引执行非线性上采样, 减少内存占用并提升了速度. 为追求模型轻量化, 实时语义分割算法ENet[12]放弃最后下采样阶段, 感受野不足以覆盖比较大的对象, 导致算法分割精度较低, 但是该算法具有分割速度快的优点. 实时语义分割算法BiSeNet[13]使用浅层网络处理高分辨率图像, 并提出一种快速下采样的深层网络以平衡分类能力和感受野大小, 此算法可取得较高的分割速度和分割精度.文献[14]指出浅层特征中存在不准确的语义信息, 将深层和浅层特征直接叠加会产生大量噪声, 导致模型分割精度降低. 为解决此问题, 本文使用注意力机制为浅层特征分配像素级权重可抑制不准确语义信息的输出. 除此之外, 本文认为输入图像在神经网络中经下采样后得到的特征图中, 每个特征点在空间位置上与图像中的若干个像素点组成的区域相对应, 而这些像素点所属的类别可能不同, 因此采用多标签分类损失函数显式地监督训练网络, 使每个特征点可以具有多种类别信息, 提升特征语义信息的准确性, 进而提升算法分割精度.1本文算法为使本文算法具有实时分割速度且具有较高分割精度, 本文在当前最先进的语义分割算法DeepLabv3基础上轻量化特征提取网络, 即使用ResNet34[15]作为基础网络结构; 使用特征金字塔结构为其增加解码结构, 使其具有实时分割速度; 将优化后的网络结构作为本文算法的基础网络, 同时增加注意力机制模块和多标签损失函数监督训练, 进一步提升算法分割精度. 本节将详细介绍这3点改进.本文算法整体结构如图1所示, 其中backbone 表示ResNet34, /2和×2分别表示特征图下采样2倍和特征图上采样2倍, 虚线箭头表示在该阶段使用多标签分类损失监督网络训练, “+”表示特征图以相加的方式融合.1.1神经网络结构在自然图像中的对象往往具有不同尺度和纵横比, 如街景图像中天空、建筑、马路与路灯、广告牌尺度差别较大, 具有不恰当感受野的神经网络将无法给不同尺度目标均衡的关注. 比如, 具有小感受野的神经网络将会更加关注小目标或者将大目标分割成多个部分, 相反, 具有大感受野的神经网络将忽视小目标. 因此, 获得多感受野网络对于精细分割具有重要的意义, 而DeepLabv3利用图像的空间局部关联性和空洞卷积的采样特点对图像进行卷积运算, 既可以获取多种感受视野信息, 又可以保留特征的空间信息; 但这导致大量参数与高分辨率特征作点积运算, 降低分割速度. 而本文在保留DeepLabv3多感受野性能的前提下对分割速度进行优化. 本文算法整体结构如图1所示, 首先使用相对轻量级的ResNet34代替DeepLabv3第1期高翔, 等: 基于注意力和多标签分类的图像实时语义分割 61图1 本文算法整体结构中的ResNet101[15]作为特征编码器, 其次增加特征金字塔网络(feature pyramid networks, FPN)[16]作为解码器逐层上采样恢复特征空间信息和语义信息, 最后压缩空洞金字塔池化网络中的参数数量. 其中, ResNet34残差连接单元可避免梯度在反向传播阶段消失; FPN 是一种融合不同层级特征图的方式, 解码阶段使用FPN 重用浅层特征修复深层特征图的空间细节信息, 可以进一步增强特征图鲁棒性. 如图2所示, ASPP 由不同空洞率r 的卷积并联组成, 特征经过FPN 处理之后具有较强的鲁棒性, 因此只需要较少参数的ASPP 模块即可实现多尺度目标分割.图2 ASPP从网络组成方面来讲, 本文算法的网络主要由卷积层、激活层、空洞卷积层和批标准化层(batch normalization, BN)[17]共4种基础单元堆叠而成, 其中卷积层负责提取图像特征, 激活层负责提高网络的非线性程度, 空洞卷积层负责在保留特征空间信息的前提下增大算法感受野、提升特征的鲁棒性[7], 而BN 通过对网络不同层之间传递的数据进行标准化以消除内部协变量移位现象[17], 进而提高算法的收敛速度和精度.1.2 跨级注意力模块文献[14]指出, 浅层特征中存在不准确的语义信息, 而FPN 将深浅特征图直接相加, 这种特征融合方式将浅层特征中错误信息或冗余信息加到深层特征, 影响了算法分割精度. 鉴于此, 在深浅层特征融合的过程中, 本文引入跨级注意力模块(cross-level attention mechanism, CAM)抑制错误信息或冗余信息的输出. 注意力机制[18]的作用机制类似人类观察环境, 往往只关注某些特别重要的局部, 获取不同局部的重要信息, 抑制对当前识别作用不大的特征, 增强有效特征的作用. 注意力特征有助于增强模型的特征表达能力, 综合不同信息, 提高模型的理解能力[19].本文提出的CAM 如图3所示, 深层特征经过3×3卷积、BN 、激活处理后, 得到与编码器中浅层特征图尺度相对应的可解释权重矩阵, 然后与浅层特征图相乘, 最后将加权后的浅层特征图与深层特征图相加. 该模块以一种简单的方式使用深层特征指导浅层特征加权, 为浅层特征图提供像素级注意力, 使其关注更加具有信息量的特征点, 即在有限参数量下尽可能表达重要的信息. 该模块能够更好地平衡修剪模型架构与增强模型表达力.图3 CAM1.3 联合多标签分类监督训练不同于传统的单标签学习任务中每个样本只与一种类别信息有关, 多标签学习[20]需要输出多个标签信息, 其中每个实例可以与一组标签相关联. 假设n X = 表示n 维实例空间, {}12,,,= q Y y y y 表示标签空间, 该标签空间有q 种可能的标签类62计算机辅助设计与图形学学报 第33卷别. 多标签学习的任务是从多标签训练数据集D =(){},|1i i x Y i m ≤≤中学习一个函数:2Y f X →, 对于任意一个测试实例x X ∈, 多标签分类器()f ⋅预测x 的标签集合()f x Y ∈. 如图4所示, 在单标签分类任务中, 可能将图4中箭头所指位置的标签分配为马或者人, 但是在多标签分类任务中, 将会同时分配这2种类别, 并以此作为神经网络的标签, 监督网络同时学习这2种类别的特征.图4 多标签分类图示在当前语义分割技术中, 使用线性插值对特征进行上采样恢复语义信息类似于多标签分类任务中提取特征的过程. 如图4所示, 在语义分割编码阶段, 输入图像经卷积和下采样运算后输出原图像1/K 大小的特征图, 其中, K 为下采样倍数. 图4中本文根据下采样倍数对图像按照空间位置划分多个网格, 根据卷积神经网络输出特征与原图的映射关系可知, 特征图中的每个特征点与每个网格一一对应, 与网格中的像素点为一对多的关系. 通过图4可以发现, 在一部分网格中只存在背景类别的像素, 另一部分网格中存在马、人、背景共3种类别像素, 因此, 在某些网格内部存在多种类别目标的边界交汇. 在解码器上采样阶段使用线性插值对特征点采样提高特征图分辨率, 由于线性插值是基于空间不变模型的方法, 无法捕捉边缘快速变化的信息, 会产生边缘模糊效果[21].因此, 为进一步提高解码阶段上采样特征的准确性, 本文在特征图分辨率为原图1/32和1/16大小的特征图上进行上采样时, 引入多标签分类损失函数显式地监督网络训练. 这样可以使特征点包含的语义信息与图像中对应网格区域中的像素类别信息一致, 进而可以在恢复特征图分辨率的同时保证类别信息准确性, 并且不会降低算法的分割速度. 本文使用多标签分类损失函数和交叉熵损失函数共同监督网络学习, 修正目标边界信息. 损失函数描述为()()()1CE 2BCE 16163BCE 3232ˆˆ,,ˆ ,.=++L L y y L s s L s sλλλ其中, CE L 表示交叉熵损失函数; BCE L 表示多标签损失函数, 本文中多标签损失函数使用二进制交叉熵损失函数; 1632, , ⨯∈ H W y s s 表示真实标签;16s 和32s 表示标签分辨率大小分别为原标签分辨率大小的1/16和1/32; 1632ˆˆˆ, , ⨯∈ H W ys s 表示对应预测值; 123, , λλλ表示控制3个损失函数权重的3个超参数.2 实验与分析2.1 评价指标在本文实验中使用平均交并比(mean intersec-tion-over-union, mIoU), 处理每幅图像所用时间t (ms)和图像处理速度v (帧/s)作为算法性能评价指标.mIoU 为语义分割的标准度量, 计算2个集合的交集与并集之比, 在每个像素类别内计算交并比(intersection-over-union, IoU), 然后计算平均值. 使用处理每幅图像所用的时间t (ms)和图像处理速度v (帧/s)来衡量算法的速度, mIoU 和v 计算公式分别为001mIoU ,1kiik ki ij ji iij j p k p p p ====++-∑∑∑.NiiNv t =∑其中, ii p 表示分割正确的数量; ij p 表示本属于i 类但预测为j 类的像素数量; ji p 表示本属于j 类被预测为i 类的像素数量; N 表示图像数量; t 表示处理每幅图像所用的时间.2.2 实验数据与实验环境Cityscapes 数据集[22]包含来自不同城市、不同季节拍摄的5 000幅精确标注和20 000幅粗略标注的街景图像, 每幅图像分辨率为1 024像素×2 048像素. 数据集共19个街景类别. 在本文实验中, 仅使用精确标注的5 000幅图像, 其中3 475幅用于训练模型, 1 525幅用于测试, 测试数据没有提供真实标签, 需要提交其官方服务器测评.CamVid 数据集[23]包含从视频序列中提取的701幅分辨率为760像素×960像素的图像, 其中367幅用于训练, 101幅用于验证, 233幅用于测试,第1期高翔, 等: 基于注意力和多标签分类的图像实时语义分割 63在本文实验中共测试11个语义类别.本文所有实验的仿真实验环境为Ubuntu18.04, Python3.7.4, Pytorch1.1.0, 显卡为NVIDIA Titan RTX 和GTX1060. 模型编码器基础网络是在ImageNet 数据集[24]上预训练的ResNet34. 初始学习速率设置为0.005, 学习速率调整策略使用多项式衰减策略, 权重衰减使用L 2正则化, 衰减系数设置为0.000 5, 动量设置为0.9.2.3 算法性能分析与比较首先, 本文基于Cityscapes 和CamVid 数据集的分割结果在速度和精度2个方面进行模块有效性评估, 之后与模型FCN-8s, DeepLabv2, ENet, SegNet, ICNet [25]和BiSeNet 对比, 最后通过可视化结果进一步分析算法的分割性能. 2.3.1 模块有效性评估为评估本文提出的CAM 的效果, 首先对轻量化后的DeepLabv3模型进行评估, 记为Baseline, 然后对应用了本文CAM 模型进行评估, 记为Baseline+CAM. 表1所示为在显卡为NVIDIA Titan RTX 的实验环境下的消融实验结果, 在Cityscapes 和CamVid 数据集上的实验结果显示, 引入该CAM 后mIoU 分别提高了1.2%和1.4%, 处理每幅图像的运算时间分别增加1.7 ms 和1.4 ms, 证明该注意力模块可以在消耗很少运算时间的前提下提高模型分割精度, 并且说明深层特征图可以有效地指导浅层特征图保留有效信息, 防止传入过多干扰信息.表1 各模块有效性评估对比表Cityscapes(768×1536) CamVid(720×960)算法mIoU/%/ms t mIoU/% /ms t Baseline 65.3 21.9 71.5 9.8 Baseline+CAM 66.5 23.6 72.9 11.2Baseline+CAM+L ML68.123.674.111.2为评估本文提出的多标签分类损失函数辅助监督算法(记为Baseline+CAM+L ML )的有效性, 将仅使用交叉熵损失函数的算法与在解码器阶段使用多标签分类损失函数的算法对比. 其中, 关于多标签损失函数的超参数设定, 本文通过经验方式人为确定几组不同超参数值对比网络性能, 从中选取一组较优的超参数作为本文多标签损失函数的超参数, 最终超参数设定为121, 0.3,==λλ30.7=λ. 由表1可以看出, 相对于仅使用交叉熵损失函数的算法, 使用多标签分类函数监督网络训练的精度在Cityscapes 和CamVid 数据集上分别提高了1.6%和1.2%, 并且对网络运行速度没有影响, 可以说明本文采用多标签分类损失函数监督网络解码训练的有效性. 此外, 通过改变网络监督方式提升网络性能, 对于图像实时语义分割是一种非常有效的方式, 既能提高分割精度又不会影响分割速度.2.3.2 算法整体分析与比较为验证本文算法的有效性, 实验将本文算法与DeepLabv2, FCN-8s, SegNet, ENet 在Cityscapes 和CamVid 数据集上进行分割精度和速度的对比, 统一采用mIoU 衡量语义分割精度, 训练参数设置见第2.2节.从算法分割精度和处理速度上分析, 在Cityscapes 数据集上使用分辨率为768像素×1 536像素, 512像素×1 024像素进行训练、测试, 在CamVid 数据集上分别使用720像素×960像素, 384像素×480像素图像进行训练和测试. 表2是在显卡为NVIDIA Titan RTX 的实验环境下的部分实验结果. 其中, 本文算法在2个数据集上的处理速度分别为42帧/s 和89帧/s, 虽然分割速度略慢于ENet, 但是在2个数据集上的mIoU 比ENet 分别提高了9.8%和5.8%. 与SegNet 相比, 在速度更快的前提下, 在2个数据集上的mIoU 提高了11.1%和8.9%. 可见, 本文算法在速度与精度上与现有的实时分割算法相比有较好的表现. 与DeepLabv2和FCN-8s 非实时语义分割模型相比, 本文算法的分割精度也具有较大优势. 综合分析, 本文算法可以在分割速度与分割精度之间取得较好的平衡, 可以实现精确高效的分割.表2 不同算法在2个数据集上的性能对比Cityscapes(768×1 536) CamVid(720×960) 算法mIoU/%/ms t v /(帧·s –1) mIoU/% /ms t v /(帧·s –1)DeepLabv2[8] 63.1 4 000.2<1 71.3 830.0 1 FCN-8s [4] 60.4 250.0 4 66.9 76.913 SegNet [11] 57.0 31.332 65.2 13.574 ENet [12]58.3 11.983 68.3 7.3136 本文-ResNet3468.123.64274.111.289图5所示为本文算法在2个数据集上训练的损失函数曲线, 其中非联合多标签分类的交叉熵损失仅使用交叉熵损失. 可以看出, 使用本文提出的多标签分类损失函数作为损失函数辅助训练的算法在训练过程中损失值平稳下降, 相较于仅使用交叉熵损失函数有更好的表现, 进而可有效地监督算法在低分辨率特征图上进行多标签分类.64计算机辅助设计与图形学学报 第33卷a. Cityscapes 数据集b. CamVid 数据集图5 不同算法在2个数据集上的损失对比曲线表3和表4为6 GB 显存GTX1060显卡实验环境下的实验性能对比, 训练过程中无任何数据扩充, 图6~图8所示为相应分割效果展示图. 表3和图6分别展示本文算法(本文-ResNet34)与ICNet, BiSeNet 在Cityscapes 数据集上的分割结果对比. 根据表3在Cityscapes 和CamVid 数据集上的分割实验数据可知, 相比于ICNet, 本文算法在分割精度和分割速度上均有较大优势. 相比于BiSeNet, 本文算法分割精度有所提高, 但是分割速度稍慢. 根据图6分割效果图可以看出, 从分割细节来讲, ICNet 丢失细节较多, BiSeNet 对细节分割好于ICNet, 而本文算法好于BiSeNet, 这一点可以通过图6中的杆状目标分割效果看出; 另一方面, 这3种实时分割算法对图像的整体均能取得较好的分割效果, 但依据图6第2行和第3行的展示图可以看出, 本文算法和BiSeNet 对马路的分割效果好于ICNet. 而Baseline-ResNet34虽然对细节分割效果好于ICNet, 但对图像整体分割效果较差.表3 不同算法在2个数据集上性能对比Cityscapes(512×1 024) CamVid(360×480)mIoU/%/ms t v /(帧·s −1) mIoU/% /ms t v /(帧·s −1)ICNet [25] 65.3 60.016 63.4 40.025 BiSeNet [13] 66.2 20.651 65.1 14.967 本文-ResNet3466.823.04365.317.059表4 本文算法和DeepLabv3使用MobileNetv2性能对比mIoU/%t/msv/(帧·s −1) mIoU/%t/msv/(帧·s −1)Deeplabv3- MobileNetv2 64.330.2 31 62.3 22.245 本文-MobileNetv265.420.05063.9 15.963a. 输入图像b. Baseline-ResNet34c. ICNet [25]d. BiSeNet [13]e. 本文-ResNet34图6 实时语义分割算法结果对比算法算法Cityscapes(512×1 024) CamVid(360×480)第1期高翔, 等: 基于注意力和多标签分类的图像实时语义分割65本文算法与DeepLabv3使用MobileNetv2[26]作为特征提取网络的分割效果对比如图7所示, 其对应的实验数据如表4所示. 从图7可以看出, 本文算法在分割细节方面稍好于DeepLabv3, 但DeepLabv3的分割效果表现出该模型丢失了图像中更多的小目标和细节现象; 从图像整体分割效果来看, 这2种模型均能取得较好的分割效果.再结合图6和图7分析ResNet34和MobileNetv2作为特征提取网络对本文算法分割性能的影响. 可以看出, 使用ResNet34作为特征提取网络的模型在图像细节分割效果上要好于使用MobileNetv2的模型, 在图像整体分割效果上二者分割效果大体相似. 而结合表3和表4数据可以看出, 本文- ResNet34与本文-MobileNetv2模型分割速度相近. 本文分析这与测试Batch大小有关, 本文实验测试Batch为1, 当Batch增大, 本文-MobileNetv2模型占用显存少的优势更明显, 可被并行处理的图像数量增多, 其分割速度会有显著提升.a. 输入图像b. Deeplabv3-MobileNetv2c. 本文-MobileNetv2图7 Deeplabv3与本文算法在Cityscapes数据集上的对比接下来通过图8分析本文算法性能. 图8c所示为Baseline, 其提取特征网络为ResNet34, 该模型不含有CAM和多标签分类; 图8d所示为使用MobileNetv2作为提取特征网络的DeepLabv3模型; 图8e和图8f分别为使用MobileNetv2和ResNet34作为提取特征网络并与CAM和多标签分类相结合的模型, 即本文最终模型. 对于CamVid数据集, 其图像中细节部分较多, 实验中的图像分辨率只有360像素×480像素, 此分辨率较低, 因此在该数据集进行实验更加考察模型分割图像细节的能力. 从图8可以看出, Baseline对细节丢失最多, 而本文算法和DeepLabv3对细节的分割要好于Baseline, 可见本文算法的有效性.3 结语本文提出了一种基于注意力机制和多标签分类的实时图像语义分割网络, 首先优化DeepLabv3神经网络架构至满足实时分割的要求, 在此基础上设计了跨级特征注意力模块和多标签分类损失函数. CAM利用蕴含丰富语义信息的深层特征对浅层特征进行像素级加权, 实现更精确的空间信息选取, 同时, 使用多标签分类损失函数辅助监督网络学习, 在恢复特征图分辨率时有效提高类别信息准确性, 二者共同作用使得类别边界像素分割更精细. 最后, 在Cityscapes数据集和CamVid数据集上进行了一系列对比实验, 实验结果表明,本文算法能够更加准确地处理复杂场景图像中图66计算机辅助设计与图形学学报 第33卷a. 输入图像b. 真实标签c. Baselined. DeepLabv3e. 本文-MobileNetv2f. 本文-ResNet34图8 DeepLabv3与本文算法在CamVid 数据集上的对比像分割问题, 显著改善类别边缘区域分割效果, 同时证明本文算法是一种分割精度高、分割速度快的图像实时语义分割算法.参考文献(References):[1] Csurka G , Perronnin F. An efficient approach to semantic seg-mentation[J]. International Journal of Computer Vision, 2011, 95(2): 198-212[2] He Y H, Wang H, Zhang B. Color-based road detection in ur-ban traffic scenes[J]. IEEE Transactions on Intelligent Trans-portation Systems, 2004, 5(4): 309-318[3] An Zhe, Xu Xiping, Yang Jinhua, et al . Design of augmentedreality head-up display system based on image semantic seg-mentation[J]. Acta Optica Sinica, 2018, 38(7): 77-83(in Chi-nese)(安喆, 徐熙平, 杨进华, 等. 结合图像语义分割的增强现实型平视显示系统设计与研究[J]. 光学学报, 2018, 38(7):77-83)[4] Long J, Shelhamer E, Darrell T. Fully convolutional networksfor semantic segmentation[C] //Proceedings of the IEEE Con-ference on Computer Vision and Pattern Recognition. Los Alamitos: IEEE Computer Society Press, 2015: 3431-3440 [5] Lin G S, Milan A, Shen C H, et al . Refinenet: multi-path re-finement networks for high-resolution semantic segmenta-tion[C] //Proceedings of the IEEE Conference on Computer Vi-sion and Pattern Recognition. Los Alamitos: IEEE Computer Society Press, 2017: 5168-5177[6] Ronneberger O, Fischer P, Brox T. U-Net: convolutional net-works for biomedical image segmentation[C] //Proceedings ofMedical Image Computing and Computer Assisted Interven-tion. Heidelberg: Springer, 2015: 234-241[7] Chen L C, Papandreou G , Kokkinos I, et al . Semantic imagesegmentation with deep convolutional nets and fully connected crfs[OL]. [2020-02-16]. https:///abs/1412.7062, 2014 [8] Chen L C, Papandreou G , Kokkinos I, et al . DeepLab: semanticimage segmentation with deep convolutional nets, atrous con-volution, and fully connected CRFs[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2018, 40(4): 834-848 [9] Chen L C, Papandreou G , Schroff F, et al . Rethinking atrousconvolutionforsemanticimagesegmentation[OL].[2020-02-16]. https:///abs/1706.05587[10] Yue Shiyi. Image semantic segmentation based on hierarchicalcontext information[J]. Laser & Optoelectronics Progress, 2019, 56(24): 107-115 (in Chinese)(岳师怡. 基于多层级上下文信息的图像语义分割[J]. 激光与光电子学进展, 2019, 56(24): 107-115)[11] Badrinarayanan V , Kendall A, Cipolla R. SegNet: a deep con-volutional encoder-decoder architecture for image segmenta-tion[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2017, 39(12): 2481-2495[12] Paszke A, Chaurasia A, Kim S, et al . ENet: a deep neural net-work architecture for real-time semantic segmentation[OL]. [2020-02-16]. https:///abs/1606.02147[13] Yu C Q, Wang J B, Peng C, et al. BiSeNet: bilateral segmen-tation network for real-time semantic segmentation[C] //Proceedings of the European Conference on Computer Vision. Heidelberg: Springer, 2018: 334-349[14] Ghiasi G , Fowlkes C C. Laplacian pyramid reconstruction andrefinement for semantic segmentation[C] //Proceedings of the European Conference on Computer Vision. Heidelberg: Springer, 2016: 519-534。
背角无齿蚌钙调蛋白基因的克隆及Ca2+和Cd2+对其表达的影响
228
生态毒理学报
第 16 卷
ana and its expressions were analyzed. The complete cDNA sequence of AwCaM1 was comprised of 692 bp contained a 516 bp open reading frame which was encoded 172 amino acids. AwCaM1 contained four putative Ca2+ -
人的指纹的介绍作文英语
Fingerprints are a unique biological feature that every individual possesses. They are the patterns of ridges and grooves found on the skin of our fingertips. These patterns are formed during the early stages of fetal development and remain unchanged throughout a persons life. Heres an introduction to the fascinating world of fingerprints:1. Uniqueness: No two people have the same fingerprints, not even identical twins. This uniqueness makes fingerprints an ideal tool for personal identification.2. Formation: Fingerprints are formed during the 13th to 19th week of gestation. The pattern is influenced by genetic and environmental factors, such as the position of the fingers in the womb.3. Types: There are three main types of fingerprint patterns: arches, loops, and whorls. Each type has its own subcategories and characteristics.4. Functions: Beyond identification, fingerprints also serve a practical purpose. The ridges and grooves on our fingertips increase the friction between our fingers and objects, helping us grip and hold things more effectively.5. Use in Forensics: Fingerprints have been used in forensic science for over a century. They are often left at crime scenes and can be lifted using various techniques, such as dusting with powder or using chemical reagents to make them visible.6. Biometrics: In modern technology, fingerprints are used as a form of biometric identification. They can be scanned and used to unlock devices, authenticate transactions, and secure access to facilities.7. Health Indicators: Some studies suggest that fingerprints can also indicate certain health conditions. For example, the patterns and ridge characteristics may be linked to the risk of developing certain diseases.8. Cultural Significance: In various cultures, fingerprints have been used symbolically. For instance, in some Asian cultures, a red thumbprint is used as a personal seal or signature on important documents.9. Evolutionary Perspective: From an evolutionary standpoint, fingerprints may have developed to provide our ancestors with better grip and dexterity, which would have been crucial for survival.10. Future Research: Ongoing research continues to explore the full potential offingerprints, including their role in genetics, disease markers, and even as a means of enhancing security and privacy in digital transactions.Fingerprints are not just a means of identification they are a testament to the complexity and individuality of human biology. As technology advances, the applications and understanding of fingerprints are likely to expand even further.。
探索不确定性与遥感数据论文 英译汉
Exploring uncertainty in remotely sensed data withparallel coordinate plotsYong Ge , Sanping Li , V. Chris Lakhan , Arko LucieerAbstract The existence of uncertainty in classified remotely sensed data necessitates the application of enhanced techniques for identifying and visualizing the various degrees of uncertainty. This paper, therefore, applies the multidimensional graphical data analysis technique of parallel coordinate plots (PCP) to visualize the uncertainty in Landsat Thematic Mapper (TM) data classified by the Maximum Likelihood Classifier (MLC) and Fuzzy C-Means (FCM). The Landsat TM data are from the Yellow River Delta, Shandong Province, China. Image classification with MLC and FCM provides the probability vector and fuzzy membership vector of each pixel. Based on these vectors, the Shannon’s entropy (S.E.) of each pixel is calculated. PCPs are then produced for each classification output. The PCP axes denote the posterior probability vector and fuzzy membership vector and two additional axes represent S.E. and the associated degree of uncertainty. The PCPs highlight the distribution of probability values of different land cover types for each pixel, and also reflect the status of pixels with different degrees of uncertainty. Brushing functionality is then added to PCP visualization in order to highlight selected pixels of interest. This not only reduces the visualization uncertainty, but also provides invaluable information on the positional and spectral characteristics of targeted pixels.1. IntroductionA major problem that needs to be addressed in remote sensing is the analysis,identification and visualization of the uncertainties arising from the classification ofremotely sensed data with classifiers such as the Maximum Likelihood Classifier (MLC)and Fuzzy C-Means (FCM). While the estimation and mapping of uncertainty has beendiscussed by several authors (for example, Shi and Ehlers, 1996; van der Wel et al., 1998;Dungan et al., 2002; Foody and Atkinson, 2002; Lucieer and Kraak, 2004; Ibrahim et al.,2005; Ge and Li, 2008a), very little research has been done on identifying, targeting andvisualizing pixels with different degrees of uncertainty. This paper, therefore, appliesparallel coordinate plots (PCP) (Inselberg, 1985, 2009; Inselberg and Dimsdale, 1990) tovisualize the uncertainty in sample data and classified data with MLC and FuzzyC-Means. A PCP is a multivariate visualization tool that plots multiple attributes on theX-axis against their values on the Y-axis and has been widely applied to data mining andvisualization (Inselberg and Dimsdale, 1990; Guo, 2003; Edsall, 2003; Gahegan et al., 2002; Andrienko and Andrienko, 2004; Guo et al., 2005; Inselberg, 2009). The PCP is useful for providing a representation of high dimensional objects for visualizing uncertainty in remotely sensed data compared with two-dimensional, three-dimensional, animation and other visualization techniques (Ge and Li, 2008b). Several advantages of the PCP technique for visualizing multidimensional data have been outlined by Siirtoia and Ra¨iha¨ (2006).Data for the PCPs are from a 1999 Landsat Thematic Mapper (TM) image acquired over the Yellow River Delta, Shandong Province, China. After classifying the data the paper emphasizes the uncertainties arising from the classification process. The probability vector and fuzzy membership vector of each pixel, obtained with classifiers MLC and FCM, are then used in the calculation of Shannon’s entropy (S.E.), a measure of the degree of uncertainty. Axes on the PCP illustrate S.E. and the degrees of uncertainty. Brushing is then added to PCP visualization in order to highlight selected pixels. As demonstrated by Siirtoia and Ra¨iha¨(2006) the brushing operation enhances PCP visualization by allowing interaction with PCPs whereby polylines can be selected and highlighted.2. Remarks on parallel coordinate plots and brushingParallel coordinate plots, a data analysis tool, can be applied to a diverse set of multidimensional problems (Inselberg and Dimsdale, 1990). The basis of PCP is to place coordinate axes in parallel, and to present a data point as a connection line between coordinate values. A given row of data could be represented by drawing a line that connects the value of that row to each corresponding axis (Kreuseler, 2000; Shneiderman, 2004). A single set of connected line segments representing one multidimensional data item is called a polyline (Siirtoia and Ra¨iha¨, 2006). An entire dimensional dataset can be viewed whereby all observations are plotted on the same graph. All attributes in two-dimensional feature space could, therefore, be distinguished thereby allowing for the recognition of uncertainties and outliers in the dataset.The PCP can be used to visualize not only multidimensional data but alsonon-numerical multidimensional data. Jang and Yang (1996) discussed applications of PCP especially its usefulness as a dynamic graphics tool for multivariate data analysis. In this paper the PCP is applied to Landsat TM multidimensional data. This paper follows the procedures of Lucieer (2004) and Lucieer and Kraak (2004) who adopted the PCP from Hauser et al. (2002) and Ledermann (2003) to represent the uncertainties in the spectral characteristics of pixels and their corresponding fuzzy membership. To enhance the visualization of the PCP tinteractive brushing functionality is employed. In the brushing operation, pixels of interest are selected and then highlighted. Brushing permits not only highlighting but also masking and deletion of specific pixels and spatial areas. 3. Use of PCP to explore uncertainty in sample dataIn the supervised classification process of remotely sensed imagery, the quantity of samples is a major factor affecting the accuracy of the image classification. In this section, the uncertainty in the sample data is, therefore, first explored with PCP.3.1. Data acquiredIn this paper, the Landsat TM image representing the study area was acquired on August 28, 1999, and covers an area of the Yellow River delta. The image area is at the intersection of the terrain between Dongying and Binzhou, Shandong Province. The upper left latitude and longitude coordinates of the image are 11880034.0700E and 37822024.0000N, respectively. The lower right latitude and longitude coordinates are 118810052.8300E and 37813058.1300N, respectively. The image size is 515 by 515 pixels, with each pixel having a resolution of 30 m. Fig. 1 is a pseudo-color composite image of bands 5, 4 and 3.3.2. Exploring the uncertainty in sample dataThe image includes six land cover types, namely Water, Agriculture_1, Agriculture_2 (Agriculture_1 and Agriculture_2 are different crops), Urban, Bottomland (the channel of the Yellow River), and Bareground. Sample data are selected from the region of interest (see Fig. 2), and represents a total of 26,639 pixels.The Parbat software developed by Lucieer (2004) and Lucieer and Kraak (2004) is used to produce the PCP. The PCP depicts the multidimensional characteristics of pixels in the remote sensing image througha set of parallel axes and polylines in a twodimensional plane (Fig. 3). It is noticeable that there is a clear representation of sample data from different land cover types, as shown by clustering of spectral signatures, and the dispersion and overlapping of spectral signatures from different land cover types. The digital numbers (DNs) of all pixels in the land cover type,Fig. 1. Pseudo-color composition image of the study area.Fig. 2. Sample data in the region of interestBottomland, are very concentrated within a narrow band. The range of DNs of pixels in the Water class is also narrow, except for band 3. The land cover types of Water and Bottomland in Fig. 3 can be easily distinguished from the other land covers in bands 5 and 7. Further differentiation is provided in band 3. The radiation responses for Agriculture_1 and Agriculture_2 have close similarity in bands 1, 2, 3, 6 and 7, with a degree of overlap in bands 4 and 5. There is an almost perfect positive correlation between bands 1 and 2 for all categories. This occurrence presents difficulties in clearly differentiating pixels for Agriculture_1 and Agriculture_2. Hence, it is evident that there is uncertainty in differentiating pixels for the land cover types Agriculture_1 and Agriculture_2.4. Classification and measurement of uncertainty in classified remote sensing images 4.1. Uncertainties arising from the classification processThe supervised classifiers, MLC and FCM, are applied to the image, with the condition that no pixel is assigned to a null class. The classified images are shown in Fig. 4a and b.Comparison of Fig. 4a and b reveals that the classified results from MLC and FCM are not identical for the data pertaining to the same region of interest (ROI).Fig. 3. PCP of sample data.The difference images between these two classified results are presented in Fig. 5a and b, with Fig. 5a showing the classified results for MLC. The classified results of FCMfor the difference pixels are illustrated in Fig. 5b. The number of difference pixels total 16,416. These difference pixels are distributed mainly on the banks of the river, and in mixed areas of Bareground, Agriculture_1, and Agriculture_2. For the MLC classified result, 57.1% of the difference pixels are classified as Agriculture_1, and 36.7% are classified as Agriculture_2. The FCM classified results, however, demonstrate that 90.5% of difference pixels are Bareground while 9.3% are Agriculture_2.The number of pixels in the ROI and in each of the classification categories from MLC and FCM are illustrated in Fig. 6. Evidently, there is a significant difference in the number of pixels for Agriculture_1, Agriculture_2, and Bareground, while the number of pixels for Water and Bottomland are very similar. This is also demonstrated in Fig.Fig. 4. (a) Classification result from MLC; (b) classification result from fuzzy CmeansFig. 5. Spatial distribution of difference pixels between MLC and FCM: (a) difference pixels in the classified result from MLC; (b) difference pixels in the classified result from FCM.Fig. 6. Comparison of numbers of pixels in the ROI; each category from MLC and FCM.5a and b. Based on the classified results from MLC and FCM it is possible to claim that there are relatively high uncertainties in identifying Agriculture_1, Agriculture_2 and Bareground. There are, however, lower uncertainties in the identification of Water and Bareground.4.2. Measurement4.2.1. Probability/fuzzy membershipFCM produces a fuzzy membership vector for each pixel. This fuzzy membership can be taken as the area proportion within a pixel (Bastin et al., 2002). It is possible for pixels to have the same class type, but their posterior probabilities or fuzzy memberships could be different. Hence, the probability vector or fuzzy membership is normally used as a measure for uncertainty on a pixel scale. The posterior probability and fuzzy membership of Water and Agriculture_2 are illustrated in Figs. 7 and 8, respectively. The uncertainty in spatial distributions can be clearly observed in Figs. 7 and 8. In Fig. 7, pixels belonging to the Water class have larger posterior probability or fuzzy membership, thereby indicating that these pixels have smaller uncertainties. While there are insignificant variations between Fig. 7a and b there are, however, noticeable differences between Fig.8a and b, especially for the class,Agriculture_2.Fig. 7. Water class: (a) posterior probability from MLC; (b) fuzzy membership fromFCM. 4.2.2. Shannon entropy [S.E.]Entropy is a measure of uncertainty and information formulated in terms of probability theory, which expresses the relative support associated with mutually exclusive alternative classes (Foody, 1996; Brown et al., 2009). Shannon ’s entropy (Shannon, 1948), applied to the measurement of the quality of a remote sensing image, is defined as the required amount of information that determines a pixel completely belonging to a category and expresses the overall uncertainty information of the probability or membership vector, and all elements in the probability vector are used in the calculation (Maselli et al., 1994). Therefore, it is an appropriate method to measure the uncertainty for each pixel in classified remotely sensed images (van der Wel et al., 1998). The use of S.E. in this paper is represented by considering the following. Given U as the universe of discourse in the remotely sensed imagery, U contains all pixels in this image and is partitioned by {X1, X2, . . ., Xn} where n is the number of classes. The probability of each partition is denoted as pi = P(Xi) giving the S.E. as()1log 2n i ii H X p p ==∑ (1) where H(X) is the information entropy of the information source. When pi = 0, theequation becomes 0 log 0 = 0 (Zhang et al., 2001; Liang and Li, 2005). It is accepted that: 0 _ H(X) _ log n.On the basis of Bayesian decision rules, MLC determines the class type of every pixel according to its maximum posterior probability in a probability vector. The classification process could, therefore, be associated with uncertainty. S.E. derived from a probability vector or fuzzy membership vector can represent the variation in the posterior probability vector and can be taken as one of the measures on a pixel scale (van der Wel et al., 1998). Similarly, applying FCM to remotely sensed data can produce fuzzy membership values in all land cover classes. Hence, fuzzy membership can also be considered as a measure of uncertainty on a pixel scale. On the basis of the fuzzy membership of each pixel the corresponding S.E. can be calculated. To compare the MLC and FCM methods it is necessary to normalize the computed S.E. values.Fig. 8. Agriculture_2 class: (a) posterior Fig. 9.probability from MLC; (b) fuzzymembership from FCM.From Eq. (1) S.E. can be calculated through posterior probability and fuzzy memberships. Fig. 9a and b displays normalized S.E. values of classified pixels from MLC and FCM, respectively. When the grey value of a pixel is zero then the uncertainty is zero, and when the grey value is 255 then the uncertainty will be at the maximum value of 1. From Fig. 9a and b it can be observed that the classes of Water and Bottomland have lower uncertainties while the classes of Bareground, Agriculture_1 and Agriculture_2 have higher uncertainties. These results emphasize that S.E. values calculated from MLC are comparatively higher than those obtained from FCM. Obviously, FCM produces more information about the end members within a pixel than MLC.4.2.3. Degree of uncertaintyWhile S.E. provides information on the uncertainties of pixels it is, however, known that when there is a large range of greyscale values representing brightness values [0,255] the subtle differences in greyscales are not easily discernible to humans. Hence, there are difficulties in differentiating the degree of uncertainty. To overcome this problem the S.E. is, therefore, discretized equidistantly. For instance, the S.E. is discretized into the following intervals: 0.00, (0.00, 0.20], (0.20, 0.40], (0.40, 0.60], (0.60, 0.80], (0.80, 1.00]. Measurements falling into the same interval have the same degree of uncertainty. By assigning a color to each degree of uncertainty, a pixel-based uncertainty visualization is produced (see Fig. 10a and b). This discretization clearly highlights the degrees of uncertainty in the classified remotely sensing image. In Fig. 10a, representing MLC, the degrees of uncertainty for most of the pixels are 0, 1 and 2 while in Fig. 10b, associated with FCM, the degrees of uncertainty for most of the pixels are 1, 2 and 3 and occasionally 4. It is worthwhile to note that the interval map of S.E. permits a comparison of the degrees of uncertainty in classified results from different classifiers.5. PCP and brushingThe PCP is useful for visually exploring the degree of dispersion or aggregation of the DN values of pixels in each band, and can be conveniently used to investigate the reasons contributing to uncertainty. With the brushing operation, the pixels of interest could be selected and highlighted.5.1. PCPFrom the MLC results two sets of pixels have been, respectively, randomly selected from the classes of Water and Agriculture_2 tobe represented on a PCP. The Parbat software (Lucieer, 2004;Fig. 10. Degree of uncertainty: (a) derived from MLC; (b) derived from FCM.Lucieer and Kraak, 2004) is used to produce a PCP of Water (see Fig. 11) and Agriculture_2 (see Fig. 12) classified by MLC. The first six axes are posterior probabilities of the six land cover types and the last two axes are the S.E. and degree of uncertainty. For instance, Fig. 12 illustrates the uncertainties of the pixels selected from the class type of Agriculture_2 and their distributions. The posterior probability of Agriculture_2 to each pixel is relatively higher than other categories. Of significance, there is negative correlation between Agriculture_1 and Agriculture_2. In this example, the S.E. is equally divided into five intervals to obtain the degree of uncertainty. The uncertainties of the pixels selected from Water and Agriculture_2, respectively, and their distributions are illustrated by Figs. 11 and 12.Figs. 13 and 14 are obtained by placing all the DNs of these pixels, fuzzy memberships, S.E., and associated degree of uncertainty as attribute dimensions on the horizontal axis of the PCP. Similarly, pixels are randomly selected to be represented on the PCP, and the S.E. is divided into five equal intervals to obtain the degree of uncertainty. Fig. 13 highlights the uncertainty characteristics associated with the spectralsignatures of pixels from the Water class while Fig. 14 shows the spectral signatures of pixels from the Agriculture_2 class and their related uncertainty characteristics. B.1–B.7 and C.1–C.6 denote the seven bands of the TM image and their fuzzy memberships in the six land cover types, which are Water, Agriculture_1, Agriculture_2, Urban, Bottomland, and Bareground, respectively. The S.E. from the FCM classifier and the degree of uncertainty within the range 0–5 are denoted by ShE and UnL. The red line within bands 1–7 emphasize pixels with a degree of uncertainty of zero. Different colors denote different degrees of uncertainty in the PCP. From the PCP it is possible to discern the distribution of fuzzy memberships and Shannon entropies of pixels with different degrees of uncertainty.A comparison of Figs. 11 and 13 reveals that the MLC posterior probabilities results on classified pixels in the Water class have a non-zero value which are concentrated in the Bareground class. For the FCM (see Fig. 13) most of the classified pixels in the Water class have a fuzzy membership closer to 1. The fuzzy memberships of some pixels in the Bareground and Agriculture_1 and 2 classes are away from zero. It is, therefore, apparent that pixels with a highdegree of uncertainty are mixed pixels in the Water and Bareground classes, namely the boundary pixels between Water and Bareground. As expected, the spectral response for these pixels contains characteristics of both land cover types.Fig. 13 provides additional information on the spectral characteristics of Water and its uncertainty distribution. For instance, the degrees of dispersion in bands 2, 3, 4 and 5 are high, and the distance to the red line in these four bands is also relatively high. As such, their uncertainties are high. Pixels with a degree of uncertainty of 3, represented by the blue line, are relatively far The fuzzy memberships of pixels for Bareground are in the range 0.2–0.4 thereby giving rise to a high degree of uncertainty.from MLC.Fig. 13. Spectral features of the Water class and their uncertainty characteristics.Pixels with a degree of uncertainty of 4, represented by the pink line, are relatively far from those pixels with a degree of uncertainty of zero, as shown by the red line in band 4. There are two independent distribution ranges in band 3. In the case of the Water class pixels with a high degree of uncertainty, assuming their fuzzy membership on Bottomland is high, then the DNs in band 3 will be greater than the DNs of pixels with a degree ofuncertainty of zero.Fig. 14. Spectral features of pixels from the Agriculture_2 class and their uncertainty characteristicsFig. 15. (a) Class of Water: pixels with a low degree of uncertainty in the PCP; (b) class of Water: pixels with a high degree of uncertainty degree in the PCP.A comparison of Figs. 12 and 14 reveals that the classes of Agriculture_1 and Bareground have the highest uncertainty in Agriculture_2. The difference between the PCPs from MLC and FCMis because the influence of Agriculture_1 on Agriculture_2 in MLC is greater than the influence of Bareground on Agriculture_ 2. For FCM the two influences on Agriculture_2 are almost similar. When the fuzzy memberships of pixels for Agriculture_1 are greater, their DNs in bands 4 and 5 are less than that of pixels with a degree of uncertainty of zero. The DNs in all bands are dispersed for pixels with large fuzzy membership for Bareground.5.2. BrushingFrom Figs. 11 and 14 it is demonstrated that when different colors are used to represent different degrees of uncertainty, a certain amount of overlap develops between color lines, especially on the axes where the distribution of polylines is concentrated. The superposition of polylines definitely increases the difficulty in visual uncertainty analysis.A new ‘‘visual’’uncertainty is, thereby, introduced. To improve on this visualization the brushing operation is introduced to the PCP (Hauser et al., 2002; Ledermann, 2003). The difference from that of a conventional approach is that the user selects pixels of interest, and then highlights them with a brush, instead of using colored polylines. This is a suitable and convenient method for conducting targeted analysis on the spectral characteristics of pixels and their associated uncertainty.Brushing is applied to a set of pixels with a low degree of uncertainty (see Fig. 15a) and a set of pixels with a high degree of uncertainty (see Fig. 15b). The pixels targeted for investigation belong to the class type,Water.Agreen polyline denotes the pixels being brushed, while a grey polyline represents the distribution characteristics of all pixels belonging to the class type, Water. The red line for bands 1–7 represents pixels with a zero degree of uncertainty. In Fig. 15a, there is a strong negative correlation between C.5 (Bottomland) and C.6 (Bareground). This means thatthe larger the memberships of pixels in the class of Bottomland, the smaller the memberships of pixels in the class of Bareground. To further investigate the class type, Agriculture_2, PCP and brushing are used to visualize pixels with low uncertainty (seeFig. 16a) and pixels with high uncertainty (see Fig. 16b). From the Figures it becomes noticeable that when PCP is combined with brushing the user could focus on the spectral characteristics and the uncertainty distribution of pixels of interest. This effectively reduces the uncertainty introduced by the visualization of the remotely sensed data.Fig. 16. (a) Class of Agriculture_2: pixels with a low degree of uncertainty in the PCP; (b) class of Agriculture_2: pixels with a high degree of uncertainty in the PCP.6. Discussion and conclusionA major unresolved problem in image processing is how to identify and visualize the uncertainty arising out of the classification of remotely sensed data. Without doubt, targeting uncertainties will not only permit better visualization of features in geographical space but also enhance the capabilities of policymakers who have to make reliabledecisions on a broad range of geospatial issues. This paper has demonstrated the effectiveness of the combined PCP and brushing operation to explore and visualize the uncertainties in remotely sensed data classified with MLC and FCM.The MLC and FCM results demonstrate that water class pixels, with a high degree of uncertainty, have high posterior probability or fuzzy membership for the type Bottomland. Furthermore, some Agriculture_2 class pixels, with a high degree of uncertainty, have high fuzzy membership for the Bareground class type. Therefore, it is possible to compare the pixel distribution for Water and Bottomland or Agriculture_2 and Bareground, and further analyze the possible reasons causing the uncertainty by investigating the spectral characteristics for all bands. Essentially, it is necessary to use the probability vector and fuzzy membership vector of each pixel to compute the S.E. The degree of uncertainty of each pixel can then be represented on a PCP. As illustrated in this paper two axes on the PCP represent Shannon’s entropy and the degree of uncertainty.The PCP technique is also advantageous for highlighting the distribution of probability values of different land covers of each pixel, and also reflects the status of pixels with different degrees of uncertainty. Moreover, a PCP can be produced for the spectral characteristics of sample data and uncertainty attributes of classified data. The class type of the sample data can be included in the PCP to evaluate the quality of the data. Moreover, the sample data can then be compared to the classified data to evaluate whether the sample data are a reasonable reflection of the spectral characteristics of all bands. The identification of any dissimilarities or uncertainties is a definite indication of improvement in the visualization process. This paper demonstrates that there could be enhancements in PCP visualization with the addition of the brushing operation. Instead of using color polylines, as done with previous approaches, brushing permits the user to select pixels of interest. These pixels could then be highlighted with brushing instead of with color polylines. Evidently, brushing facilitates targeted analysis of the spectral characteristics of pixels and any associated uncertainty. It could, therefore, be concluded that the integration of PCP with the brushing operation is beneficial for not only visualizing uncertainty but also gaining insights on the spectral characteristics andattribute information of pixels of interest. By interacting with the PCP through the brushing operation it is possible to conduct an exploration of uncertainty, even at the sub-pixel level.AcknowledgementsThis research received partial support from the National Natural Science Foundation of China (Grant No. 40671136) and the National High Technology Research and Development Program of China (Grant No. 2006AA120106).ReferencesAndrienko, G., Andrienko, N., 2004. Parallel coordinates for exploring properties of subsets CMV. In: Roberts, J. (Ed.), Proceedings International Conference on Coordinated & Multiple Views in Exploratory Visualization, London, England, July 13, 2004, pp. 93–104.Bastin, L., Fisher, P.F., Wood, J., 2002. Visualizing uncertainty in multi-spectral remotely sensed imagery. Computers & Geosciences 28, 337–350.Brown, K.M., Foody, G.M., Atkinson, P.M., 2009. Estimating per-pixel thematicuncertainty in remote sensing classifications. International Journal of RemoteSensing 30, 209–229.Dungan, J.L., Kao, D., Pang, A., 2002. The uncertainty visualization problem in remote sensing analysis. In: Proceedings of IEEE International Geoscience and Remote Sensing Symposium, Toronto, Canada, June 2, 2002, pp. 729–731.Edsall, R.M., 2003. The parallel coordinate plot in action: design and use for geographic visualization. Computational Statistics and Data Analysis 43, 605–619.Foody, G.M., Atkinson, P.M., 2002. Uncertainty in Remote Sensing and GIS. Wiley Blackwell, London.Foody, G.M., 1996. Approaches for the production and evaluation of fuzzy land cover classifications from remotely-sensed data. International Journal of Remote Sensing 17, 1317–1340.Gahegan, M., Takatsuka, M., Wheeler, M., amd Hardisty, F., 2002. Introducing geo VISTA studio: an integrated suite of visualization and computational methods for。
外文翻译----数字图像处理和模式识别技术关于检测癌症的应用
引言英文文献原文Digital image processing and pattern recognition techniques for the detection of cancerCancer is the second leading cause of death for both men and women in the world , and is expected to become the leading cause of death in the next few decades . In recent years , cancer detection has become a significant area of research activities in the image processing and pattern recognition community .Medical imaging technologies have already made a great impact on our capabilities of detecting cancer early and diagnosing the disease more accurately . In order to further improve the efficiency and veracity of diagnoses and treatment , image processing and pattern recognition techniques have been widely applied to analysis and recognition of cancer , evaluation of the effectiveness of treatment , and prediction of the development of cancer . The aim of this special issue is to bring together researchers working on image processing and pattern recognition techniques for the detection and assessment of cancer , and to promote research in image processing and pattern recognition for oncology . A number of papers were submitted to this special issue and each was peer-reviewed by at least three experts in the field . From these submitted papers , 17were finally selected for inclusion in this special issue . These selected papers cover a broad range of topics that are representative of the state-of-the-art in computer-aided detection or diagnosis(CAD)of cancer . They cover several imaging modalities(such as CT , MRI , and mammography) and different types of cancer (including breast cancer , skin cancer , etc.) , which we summarize below .Skin cancer is the most prevalent among all types of cancers . Three papers in this special issue deal with skin cancer . Y uan et al. propose a skin lesion segmentation method. The method is based on region fusion and narrow-band energy graph partitioning . The method can deal with challenging situations with skin lesions , such as topological changes , weak or false edges , and asymmetry . T ang proposes a snake-based approach using multi-direction gradient vector flow (GVF) for the segmentation of skin cancer images . A new anisotropic diffusion filter is developed as a preprocessing step . After the noise is removed , the image is segmented using a GVF1snake . The proposed method is robust to noise and can correctly trace the boundary of the skin cancer even if there are other objects near the skin cancer region . Serrano et al. present a method based on Markov random fields (MRF) to detect different patterns in dermoscopic images . Different from previous approaches on automatic dermatological image classification with the ABCD rule (Asymmetry , Border irregularity , Color variegation , and Diameter greater than 6mm or growing) , this paper follows a new trend to look for specific patterns in lesions which could lead physicians to a clinical assessment.Breast cancer is the most frequently diagnosed cancer other than skin cancer and a leading cause of cancer deaths in women in developed countries . In recent years , CAD schemes have been developed as a potentially efficacious solution to improving radiologists’diagnostic accuracy in breast cancer screening and diagnosis . The predominant approach of CAD in breast cancer and medical imaging in general is to use automated image analysis to serve as a “second reader”, with the aim of improving radiologists’diagnostic performance . Thanks to intense research and development efforts , CAD schemes have now been introduces in screening mammography , and clinical studies have shown that such schemes can result in higher sensitivity at the cost of a small increase in recall rate . In this issue , we have three papers in the area of CAD for breast cancer . Wei et al. propose an image-retrieval based approach to CAD , in which retrieved images similar to that being evaluated (called the query image) are used to support a CAD classifier , yielding an improved measure of malignancy . This involves searching a large database for the images that are most similar to the query image , based on features that are automatically extracted from the images . Dominguez et al. investigate the use of image features characterizing the boundary contours of mass lesions in mammograms for classification of benign vs. Malignant masses . They study and evaluate the impact of these features on diagnostic accuracy with several different classifier designs when the lesion contours are extracted using two different automatic segmentation techniques . Schaefer et al. study the use of thermal imaging for breast cancer detection . In their scheme , statistical features are extracted from thermograms to quantify bilateral differences between left and right breast regions , which are used subsequently as input to a fuzzy-rule-based classification system for diagnosis.Colon cancer is the third most common cancer in men and women , and also the third mostcommon cause of cancer-related death in the USA . Y ao et al. propose a novel technique to detect colonic polyps using CT Colonography . They use ideas from geographic information systems to employ topographical height maps , which mimic the procedure used by radiologists for the detection of polyps . The technique can also be used to measure consistently the size of polyps . Hafner et al. present a technique to classify and assess colonic polyps , which are precursors of colorectal cancer . The classification is performed based on the pit-pattern in zoom-endoscopy images . They propose a novel color waveler cross co-occurence matrix which employs the wavelet transform to extract texture features from color channels.Lung cancer occurs most commonly between the ages of 45 and 70 years , and has one of the worse survival rates of all the types of cancer . Two papers are included in this special issue on lung cancer research . Pattichis et al. evaluate new mathematical models that are based on statistics , logic functions , and several statistical classifiers to analyze reader performance in grading chest radiographs for pneumoconiosis . The technique can be potentially applied to the detection of nodules related to early stages of lung cancer . El-Baz et al. focus on the early diagnosis of pulmonary nodules that may lead to lung cancer . Their methods monitor the development of lung nodules in successive low-dose chest CT scans . They propose a new two-step registration method to align globally and locally two detected nodules . Experments on a relatively large data set demonstrate that the proposed registration method contributes to precise identification and diagnosis of nodule development .It is estimated that almost a quarter of a million people in the USA are living with kidney cancer and that the number increases by 51000 every year . Linguraru et al. propose a computer-assisted radiology tool to assess renal tumors in contrast-enhanced CT for the management of tumor diagnosis and response to treatment . The tool accurately segments , measures , and characterizes renal tumors, and has been adopted in clinical practice . V alidation against manual tools shows high correlation .Neuroblastoma is a cancer of the sympathetic nervous system and one of the most malignant diseases affecting children . Two papers in this field are included in this special issue . Sertel et al. present techniques for classification of the degree of Schwannian stromal development as either stroma-rich or stroma-poor , which is a critical decision factor affecting theprognosis . The classification is based on texture features extracted using co-occurrence statistics and local binary patterns . Their work is useful in helping pathologists in the decision-making process . Kong et al. propose image processing and pattern recognition techniques to classify the grade of neuroblastic differentiation on whole-slide histology images . The presented technique is promising to facilitate grading of whole-slide images of neuroblastoma biopsies with high throughput .This special issue also includes papers which are not derectly focused on the detection or diagnosis of a specific type of cancer but deal with the development of techniques applicable to cancer detection . T a et al. propose a framework of graph-based tools for the segmentation of microscopic cellular images . Based on the framework , automatic or interactive segmentation schemes are developed for color cytological and histological images . T osun et al. propose an object-oriented segmentation algorithm for biopsy images for the detection of cancer . The proposed algorithm uses a homogeneity measure based on the distribution of the objects to characterize tissue components . Colon biopsy images were used to verify the effectiveness of the method ; the segmentation accuracy was improved as compared to its pixel-based counterpart . Narasimha et al. present a machine-learning tool for automatic texton-based joint classification and segmentation of mitochondria in MNT-1 cells imaged using an ion-abrasion scanning electron microscope . The proposed approach has minimal user intervention and can achieve high classification accuracy . El Naqa et al. investigate intensity-volume histogram metrics as well as shape and texture features extracted from PET images to predict a patient’s response to treatment . Preliminary results suggest that the proposed approach could potentially provide better tools and discriminant power for functional imaging in clinical prognosis.We hope that the collection of the selected papers in this special issue will serve as a basis for inspiring further rigorous research in CAD of various types of cancer . We invite you to explore this special issue and benefit from these papers .On behalf of the Editorial Committee , we take this opportunity to gratefully acknowledge the autors and the reviewers for their diligence in abilding by the editorial timeline . Our thanks also go to the Editors-in-Chief of Pattern Recognition , Dr. Robert S. Ledley and Dr.C.Y. Suen , for their encouragement and support for this special issue .英文文献译文数字图像处理和模式识别技术关于检测癌症的应用世界上癌症是对于人类（不论男人还是女人）生命的第二杀手。
生成对抗网络人脸生成及其检测技术研究
1 研究背景近年来，AIGC（AI Generated Content）技术已经成为人工智能技术新的增长点。
2023年，AIGC开启了人机共生的时代，特别是ChatGPT的成功，使得社会对AIGC的应用前景充满了期待。
但AIGC在使用中也存在着诸多的风险隐患，主要表现在：利用AI生成不存在的虚假内容或篡改现有的真实内容已经达到了以假乱真的效果，这降低了人们对于虚假信息的判断力。
例如，2020年，MIT利用深度伪造技术制作并发布了一段美国总统宣布登月计划失败的视频，视频中语音和面部表情都高度还原了尼克松的真实特征，成功地实现了对非专业人士的欺骗。
人有判断力，但AI没有，AI 生成的内容完全取决于使用者对它的引导。
如果使用者在使用这项技术的过程中做了恶意诱导，那么AI所生成的如暴力、极端仇恨等这样有风险的内容会给我们带来很大的隐患。
因此，对相关生成技术及其检测技术的研究成为信息安全领域新的研究内容。
本文以A I G C在图片生成方面的生成技术为目标，分析现有的以生成对抗网络（G e n e r a t i v e Adversarial Network，GAN）为技术基础的人脸生成技术。
在理解GAN的基本原理的同时，致力于对现有的人像生成技术体系和主要技术方法进行阐述。
对于当前人脸伪造检测的主流技术进行综述，并根据实验的结果分析检测技术存在的问题和研究的方向。
2 GAN的基本原理GAN由Goodfellow等人[1]于2014年首次提出。
生成对抗网络人脸生成及其检测技术研究吴春生，佟 晖，范晓明（北京警察学院，北京 102202）摘要：随着AIGC的突破性进展，内容生成技术成为社会关注的热点。
文章重点分析基于GAN的人脸生成技术及其检测方法。
首先介绍GAN的原理和基本架构，然后阐述GAN在人脸生成方面的技术模式。
重点对基于GAN在人脸语义生成方面的技术框架进行了综述，包括人脸语义生成发展、人脸语义生成的GAN实现。
基于高光谱成像技术的茄子叶片灰霉病早期检测
ｒ ｇｏ ｓ ｆｏ ３ ０ ｏ ３ ａ ．Ｔｈ ｉ ｕ ｅ ｎ ｔ ｒ ｅ ｆ ａｕ ｅ ｗａ ｅｅ ｇ ｈ ｒ ｅｅ ｔｄ ｂ ｒ ｃ ａ ｅ ｉｎ ｗａ ｒ ｍ ８ ｔ １０ １ ｍ ｅ ｐ ｃ ｒｓ ｏ ｈ ｅ ｅ ｔ ｒ ｖ ｌｎ ｔ ｓ ｗｅ ｅ ｓ ｌｃ ｅ ｙ ｐ ｉ ｉ ｌ ｔ ｎ ｐ
Ｅａ ｌ ｅｅ ｔ ｎ ｏ ｒ ｙｍｏｄ ｏ ｇ ｐ ａ ｔｌａｅ ｓｎ ｙ ｅｓｅ ｔａ ｍａ ｉｇｔｃ ｎｑ ｅ ｏ ｒ ａ ｆＺ ｅｉｎ ｒｙｄ ｔｃｉ ｆｇ ａ ｌ ｎ ｅｇ ｌｎ ｅ ｖｓｕｉｇｈ ｐ ｒｐ ｃｒ ｌ ｏ ｉ ｇｎ ｅｈ ｉ ｕ ．Ｊ ｕ ｎ ｌ ｈ ｊ ｇ ｏ ａ
关 键 词 高光 谱 成 像 技 术 ；灰 霉 病 ；最 小 二 乘 支持 向 量 机 ； 续投 影 算 法 ； 成 分 分 析 ；茄子 连 主
中图分类号 ＴＰ ３ １ ３ ． １ ９ ；Ｓ４ ６ ４ １ 文 献标 志 码 Ａ
ＦＥＮＧ Ｌｅ ，ＺＨ Ａ ＮＧ Ｄｅ ｒ ｇ ｉ — ｏｎ ， ＣＨ ＥＮ Ｓｈｕ ｎｇ ｓ ａ — ｈｕａ ｇ ｎ ，ＦＥＮ Ｇ Ｂｉ ， ＸＩ ｎ。 Ｅ Ｃｈｕ ｎ ｑ ，Ｃ Ｈ ＥＮ Ｙ ｏ ａ－ｉ ｕ—
ＤＯＩ ０． ７ ５ ｊ ｉｓ ． ０ ８ ９ ０ ． ０ ２．３． １ ：１ ３ ８ ／ ．ｓ ｎ １ ０ — ２ ９ ２ １ ０ ０ ２
基 于 高 光谱 成 像技 术 的茄 子 叶片 灰霉 病 早期 检 测
生物的英语试题及答案
生物的英语试题及答案一、选择题（每题1分，共10分）1. What is the basic unit of life?A. CellB. OrganC. TissueD. System2. Which of the following is not a characteristic of living organisms?A. GrowthB. ReproductionC. RespirationD. Inertia3. What is the process by which plants convert sunlight into energy?A. RespirationB. PhotosynthesisC. FermentationD. Digestion4. Which of the following is a type of genetic mutation?A. Gene duplicationB. Chromosomal deletionC. Both A and BD. None of the above5. What is the term for the study of the relationships among species?A. TaxonomyB. PhylogeneticsC. EcologyD. Ethology6. What is the primary function of the mitochondria in a cell?A. DNA replicationB. Protein synthesisC. Energy productionD. Waste disposal7. Which of the following is a hormone?A. InsulinB. GlucoseC. OxygenD. Carbon dioxide8. What is the correct sequence of the biologicalclassification hierarchy?A. Kingdom, Phylum, Class, Order, Family, Genus, SpeciesB. Species, Genus, Family, Order, Class, Phylum, KingdomC. Kingdom, Species, Genus, Family, Order, Class, PhylumD. Phylum, Class, Order, Family, Species, Genus, Kingdom9. What is the process by which new species arise?A. EvolutionB. Natural selectionC. SpeciationD. All of the above10. What is the role of chlorophyll in photosynthesis?A. To absorb light energyB. To produce waterC. To release oxygenD. To store energy二、填空题（每题1分，共5分）11. The process by which an organism develops from a single cell to a mature individual is called ________.12. The study of the structure of organisms is known as________.13. In genetics, the basic unit of heredity is the ________.14. The largest organ in the human body is the ________.15. The scientific method of classifying organisms based on evolutionary relationships is called ________.三、简答题（每题5分，共10分）16. Explain the role of DNA in the cell.17. Describe the process of cellular respiration.四、论述题（每题15分，共15分）18. Discuss the importance of biodiversity and the threats it faces.答案：一、选择题1-5: A, D, B, C, B6-10: C, A, A, A, A二、填空题11. Development12. Anatomy13. Gene14. Skin15. Phylogenetics三、简答题16. DNA is the molecule that carries the genetic instructions used in the growth, development, functioning, andreproduction of all known living organisms and many viruses.It is the blueprint for the organism's traits and functions. 17. Cellular respiration is the process by which cellsconvert nutrients into energy in the form of ATP (adenosine triphosphate). It involves the breakdown of glucose in the presence of oxygen to produce carbon dioxide, water, and energy.四、论述题18. Biodiversity is crucial for the health of ecosystems, asit ensures the stability and resilience of these systems. It provides a variety of ecosystem services such as pollination, pest control, and nutrient cycling. Threats to biodiversity include habitat destruction, climate change, overexploitation, pollution, and the introduction of invasive species.结束语：通过这份生物英语试题及答案，我们不仅复习了生物学的基本概念和过程，还加深了对生物多样性重要性及其面临的挑战的理解。
DNA分子的结构青年教师大赛获奖示范课公开课一等奖课件省赛课获奖课件
3.沃森和克里克默契配合，发现DNA双螺旋构造的过程， 作为科学家合作研究的典范，在科学界传为佳话。 他们这种工作方式予以你哪些启示？
构成DNA的基本单位：脱氧核苷酸 构成DNA的碱基：
腺嘌呤（A）鸟嘌呤（G） 胞嘧啶（C）胸腺嘧啶（T） 因此，脱氧核苷酸也有4种
1953年，美国生物学家沃森 （J．D.Watson，1928—）和英国物理 学家克里克（F.Crick，1916—2004）， 共同提出了DNA分子的双螺旋构造模型。
这是20世纪继爱因斯坦发现相对论之后的 又一划时代发现，它标志着生物学的研究进入 分子的层次。因为这项“生物科学中最具有革 命性的发现”，两位科学家获得了1962年度诺 贝尔生理学或医学奖。
某些规律。
∵ A ＝ T ，G ＝ C
∴ A+G＝T+C
∴ A+G
T＋C
（A+T+C+G ） （A+T+C+G
）50％
也能够写成下列形式：
A + G （ A + C ） （ T + G ） …… 1 T＋C （ T+G ） （A+C ）
规律概括：在DNA双链中，任意两个不互补碱基之 和 相等 ，并为碱基总数的 50％ 。
资料2：DNA是由许多个脱氧核苷酸连接 而成的长链。
资料3：1951年，英国科学家威尔金斯和 富兰克林提供了DNA的X射线衍射图谱 。
资料4：奥地利出名生物化学家查哥夫研究得出：腺嘌呤 （A）的量总是等于胸腺嘧啶（T）的量(A=T)，鸟嘌呤 （G）的量总是等于胞嘧啶（C）的量(G=C)。
不同公司萤光素底物在细胞活体成像中的运用比较
山东化工■146 -SHANDONG CHEMICAL INDUSTRY2021 年第 50 卷不同公司萤光素底物在细胞活体成像中的运用比较赵小鸽，陈妍科(西安交通大学医学部生物医学研究实验中心，陕西西安710061)摘要：目的：比较不同公司四种萤光素底物在细胞活体成像中的效果。
方法:将稳定表达G F P绿色荧光蛋白及萤火虫萤光素酶报告基因(lucifemo)的乳腺癌细胞M D A-M B-231分别接种至6孔板、96孔板中，分别加人四种萤光素底物，用小动物活体成像仪进行生物发光成像测试。
结果:6孔板实验中，A公司和B公司的底物生物发光强度无显著差异（J>0.05);96孔板实验中，B公司的萤光素底物生物发光强度明显高于A公司（J<0.05)。
A、B公司的萤光素底物生物发光强度明显高于C、D公司（J<0.05)，C公司明显低于D公司的荧光强度（J<0. 05)。
结论:不同公司的萤光素底物在细胞活体成像中呈现不同效果。
关键词：萤光素;生物发光;活体成像中图分类号：Q811.5 文献标识码：A文章编号：1008-021X(2021)01-0146-02Comparison of the Application of Luciferin Substrates from DifferentCompanies in Cell Vivo ImagingZhao Xiaoge，Chen Yanke(Biomedical Research Experimental Center，Medical Department of X i}n Jiaotong University，X i}n710061，C h i n a) Abstract：Objective：T o c o m p a r e the effects of four diferent luciferin substrates in cell in vivo imaging.M e t h o d s:Breast cancer cells M D A-M B-231expressing G F P green fluorescent protein a n d firefly luciferase reporter gene(luciferase) w ere seeded into6 -well a n d96-well plates respectively，four luciferin substrates were a d d e d.T h e bioluminescence imaging test w a sa small animal in vivo i mager.Results:In the6-well plate experiment，there w a s no significant difference in the bioluminescenceintensity b etween C o m p a n y A a n d c o m p a n y B( J>0.05)，in the96-well plate experiment，the bioluminescence intensity ofc o m p a n y B w a s significantly higher tlian c o m p a n y A( J<0.05) .T h e bioluminescence intensity of luciferin substrate from c o m p a n yA a n dB w a s significantly higher than that in c o m p a n yC a n d D(J<0.05)，c o m p a n y C w a s significantly lower thanD( J<0.05).Conclusion:T h e luciferin substrates of different companies showdifferent effects in cell K e y words：luciferin%bioluminescence%in vivo imaging生物发光(bioluminescence)是一些特殊生物自然发光的现象，生物体内合成的化学物质或者提取物不依赖机体对光的吸收，在特殊酶的作用下将化学能几乎全部转化为光能的化学发光现象[1]。
改进U-Net的遥感图像中建筑物变化检测
2021573遥感图像变化检测[1]是对同一地理位置不同时期获取的两幅或多幅图像进行分析和检测，从而获得地物变化信息的技术，在土地利用监控、农业监测、防灾预警、城市规划等许多领域都有广泛的应用。
目前，遥感图像的变化检测方法和技术是遥感图像处理领域的研究热点和难点之一。
传统的遥感图像变化检测方法流程[2]一般分为三步：（1）应用图像预处理技术对图像进行配准、去噪等，消除成像因素带来的图像差异；（2）以图像差分、图像比值等方法生成差分图像；（3）对差分图像进行分类，从中提取变化类特征，分析变化类特征得到变化图。
Gong等人[3-4]在小波域利用平均比值图像和对数比值图像的互补信息生成差分图像，并对FLICM算法进行改进，提出改良局部邻域模糊C均值（Reformulated Fuzzy Local改进U-Net的遥感图像中建筑物变化检测张翠军1，2，安冉1，2，马丽1，21.河北地质大学信息工程学院，石家庄0500312.河北地质大学人工智能与机器学习研究室，石家庄050031摘要：提出了一种改进U-Net的遥感图像中建筑物变化检测方法，将变化检测问题转化为像素级二分类问题，利用U-Net模型对图像进行分类，把图像中的每个像素划分为变化类或非变化类，并根据变化类的像素得到建筑物的变化检测结果图。
针对U-Net模型进行遥感图像中建筑物变化检测时，在训练中容易出现过拟合的现象，提出用非对称卷积块代替U-Net网络特征提取部分的标准卷积操作，增强卷积核的鲁棒性和网络的中心骨架，防止过拟合；针对变化检测数据集中图像背景复杂、小目标的变化情况容易被漏检的问题，提出在U-Net中引入注意力机制，抑制模型对非变化类像素特征的学习，加强对变化类特征的学习，提取到更适合的特征。
实验结果表明，在引入非对称卷积块和注意力机制后，变化检测的F1分数有明显的提升。
关键词：建筑物变化检测；U-Net；非对称卷积块；注意力机制文献标志码：A中图分类号：TP391.41doi：10.3778/j.issn.1002-8331.2005-0331Building Change Detection in Remote Sensing Image Based on Improved U-NetZHANG Cuijun1，2,AN Ran1，2,MA Li1，21.School of Information Engineering,Hebei GEO University,Shijiazhuang050031,Chinaboratory of Artificial Intelligence and Machine Learning,Hebei GEO University,Shijiazhuang050031,ChinaAbstract：A method of building change detection in remote sensing image based on improved U-Net is proposed.The U-Net model is used to classify the image,and the problem of change detection is transformed into a two classification problem.Each pixel in the image is divided into a change class or a non change class,the change detection results are obtained according to the change class pixels.In order to improve the robustness of the convolution kernel and prevent over fitting,asymmetric convolution block is proposed to replace the standard convolution operation of the feature extraction part of U-Net network.Aiming at the problem that the background of the image in the change detection data set is complex and the change of the small target is easy to be missed,the attention mechanism is introduced into U-Net to restrain the learning of the model to the non-change type pixel features,strengthen the learning of the change type features,and extract the more suitable features.The experimental results show that after the introduction of asymmetric convolution block and attention mechanism,the F1score of change detection is significantly improved.Key words：building change detection;U-Net;asymmetric convolution block;attention mechanism基金项目：河北省高等学校科学技术研究重点项目（ZD2018043，ZD2019134）；河北省自然科学基金（F2020403030）。
- 1、下载文档前请自行甄别文档内容的完整性，平台不提供额外的编辑、内容补充、找答案等附加服务。
- 2、"仅部分预览"的文档,不可在线预览部分如存在完整性等问题,可反馈申请退款(可完整预览的文档不适用该条件!)。
- 3、如文档侵犯您的权益，请联系客服反馈,我们会尽快为您处理(人工客服工作时间：9:00-18:30)。
PIXEL CLASSIFICATION:A FUZZY-GENETIC APPROACHUlrich Bodenhofer,Erich Peter KlementFuzzy Logic Laboratorium Linz-HagenbergInstitut f¨u r MathematikJohannes Kepler Universit¨a t,A-4040Linz,Austriaip@f lll.uni-linz.ac.atAbstract.In this paper a fuzzy method for pixel classification is proposed.It is one of the most important results of the development of an inspection system for a silk-screen printing process.The classification algorithm is applied to a reference image in the initial step of the printing process in order to obtain regions which are to be checked by applying different criteria.Tight limitations in terms of computation speed have necessitated very specific, efficient methods which operate locally.These methods are motivated and presented in detail in the following. Furthermore,the optimization of the parameters of the classification system with genetic algorithms is discussed. Finally,the genetic approach is compared with other probilistic optimization methods.Keywords:discrepancy norm,edge detection,genetic algorithm,pixel classification1IntroductionThe main goal of this project was to design an au-tomatic inspection system which does not sort out every print with defects,but only those which have visible defects in a way which is really unaccept-able for the consumer.It is clear that the visibility of a defect depends on the structure of the print in its neighborhood.While little spots can hardly be recognized in very chaotic areas,they can be dis-turbing in rather homogeneous areas.So,thefirst step towards a sensitive inspection is to partition the print into areas of different sensitivity which, consequently,should be treated differently.The following types were specified by experts of our partner company.For certain reasons,which can be explained with the special principles of the silk-screen printing process,it is sufficient to con-sider only these four types:Homogeneous area:uniformly colored area; Edge area:pixels within or close to visually sig-nificant edges;Halftone:area which looks rather homogeneous from a certain distance,although it is actu-ally obtained by printing small raster dots oftwo or more colors;Picture:rastered area with high,chaotic devia-tions,in particular small high-contrasted de-tails.The magnifications in Figure1show how these ar-eas typically look like at the pixel level.Of course, transitions between two or more of these areas are possible,hence a fuzzy model is recommendable.First of all,we should define precisely what,in our case,an image is:Definition1An matrix of the form(1) with three-dimensional entries(additive RGB model)is a model of a24bit color image of size.A coordinate pair stands for a pixel;the val-ues are called the gray values of pixel.It is near at hand to use something like the vari-ance or an other measure for deviations to distin-guish between areas which show only low devia-tions,such as homogeneous areas and halftone ar-eas,and areas with rather high deviations,such as edges or pictures.On the contrary,it is intuitively clear that such a measure can never be used to separate edge areas from picture areas,because of the neglection of any geometrical information.Experiments have shown that well-known stan-dard edge detectors,such as the Laplacian or the Mexican Hat,but also many other locally operat-ingfilter masks(see e.g.[7]),cannot distinguish sufficiently if deviations are chaotic or anisotropic.Homogeneous Edge HalftonePicture Figure1:Magnifications of typical representatives of the four typesAnother possibility we also took into considera-tion was to use wavelet transforms or more sophis-ticated image segmentation methods(see for in-stance[7]or[3]).Since we had to cope with se-rious restrictions in terms of computation speed, such highly advanced methods,although they are efficient,would require too much time.Finally,we found a fairly good alternative which is based on the discrepancy norm.This approach uses only, like the simpliestfilter masks,the closest neigh-borhood of a pixel.For an arbitrary butfixed pixel we defined the enumeration mapping as follows:k(,)(,)(,)(,) If we plot one color extraction of the eight neighbor pixels with respect to this enumeration, i.e.,where,wetypically get curves like those shown in Figure2.From these sketches it can be seen easily that a measure for the deviations can be used to distin-guish between homogeneous areas,halftones,and the other two types.On the other hand,the most eyecatching difference between the curves around pixels in pictures and edge areas is that,in the case of an edge pixel,the peaks appear to be more con-nected while they are mainly chaotic and narrow for a pixel in a picture area.So,a method which judges the shape of the peaks should be used in or-der to separate edge areas from pictures.A sim-ple but effective method for this purpose is the so-called discrepancy norm,for which there are al-ready other applications in pattern recognition(cf.[6]).2The Discrepancy NormDefinition2The mapping(2)is called discrepancy norm on.In words,is the absolute value of the max-imal sum of consecutive coordinates of the vector .Obviously,unlike conventional norms,the signs and the order of the entries play an essential role. Nevertheless,the mapping is a norm on.It can be seen easily that the more entries with equal signs appear successively,the higher the dis-crepancy norm is.On the contrary,for sequences with alternating signs it is close to the supremum norm.Therefore,can be used for judging the connectedness of the peaks with equal signs.Obviously,the computation of by strictly using the definition requires operations. The following theorem allows us to computewith linear speed:Theorem1For all we have(3)where the values denote the partial sums.A more detailed analysis of the discrepancy norm can be found in[2].3The Fuzzy SystemFor each pixel we consider its nearest eight neighbors enumerated as described above,which yields three vectors of gray values with8entries —one for each color extraction.As already men-tioned,the sum of the variances of the three vectorsHomogeneous Edge Halftone Picture88808 000Figure2:Typical gray value curves of the formcan be taken as a measure for the size of the devi-ations in the neighborhood of the pixel.Let us de-note this value with.On the other hand,thesum of the discrepancy norms of the vectors,wherewe subtract each entry by the mean value of all en-tries,can be used as a criterion whether the pixel iswithin or close to a visually significant edge.Of course,itself can be used as an edge detec-tor.Figure3shows how good it works comparedwith the commonly used Mexican Hatfilter mask.The fuzzy decision is then carried out for eachpixel independently:First of all,the char-acteristic values and are computed.These values are taken as the input of a small fuzzysystem with two inputs and one output.Let us de-note the linguistic variables on the input side withand.Since the position of the pixel is of norelevance for the decision in this concrete applica-tion,indices can be omitted here.The input spaceof the variable is represented by three fuzzy setswhich are labeled“low”,“med”,and“high”.Anal-ogously,the input space of the variable is repre-sented by two fuzzy sets,which are labeled“low”and“high”.Experiments have shown thatand are appropriate universes of discoursefor and,respectively.For the decompositionof the input domains simple Ruspini partitions(see[8])consisting of trapezoidal fuzzy subsets werechosen:Original image Discrepancy method Mexican HatW i t h o u t n o i seW i t h a d d e d n o i se Figure 3:Comparison between and a standardfilter maskmine the shape of the two fuzzy partitions.In the first step these parameters were tuned manually.Of course,we have also taken into consideration the use of (semi)automatic methods for finding the op-timal parameters.The general problem is not to find an appro-priate algorithm for doing that task,the difficulty is how to judge such a classification.Since the specification of the four types of areas is given in a vague,verbal form,no mathematical criterion is available for that.Hence,a model-based optimiza-tion process is,because of the lack of a model,not applicable.The alternative is a knowledge-based approach,which poses the question how to gener-ate this knowledge —the examples from which the algorithm should learn.Our optimization procedure consists of a paint-ing program which offers tools,such as a pencil,a rubber,a filling algorithm,and many more,which can be used to make a classification of a given rep-resentative image by hand.Then an optimization algorithm can be used to find that configuration of parameters which yields the maximal degree of matching between the desired result and the output actually obtained by the classification system.Assume that we havesample pixels for which the pairs of input valuesare computed and that we already have a ref-erence classification of these pixelsHo Ed Ha Pi,where .Since,as soon as the values andare computed,the geometry of the image plays no role anymore,we can switch to one-dimensionalindices here.Then one possibility to define the per-formance (fitness)of the fuzzy system would be(5)where Ho Ed Ha Pi .If is a function which maps each cluster to a representative value (e.g.,its center of gravity),we can define the fitness (objective)function as(6)whereHo EdHa PiIf the number of parts is chosen moderately(e.g.arectangular net which yields)the evaluation of thefitness function takes consid-erably less time than a direct application of formula(4).Note that in(6)thefitness is already trans-formed such that it can be regarded as a degreeof matching between the desired and the actuallyobtained classification measured in percent.Thisvalue has then to be maximized.In fact,fitness functions of this type are,in al-most all cases,continuous but not differentiableand have a lot of local maxima.Therefore,it ismore reasonable rather to use probabilistic opti-mization algorithms than to apply continuous op-timization methods,which make excessive use ofderivatives.This,first of all,requires a(binary)coding of the parameters.We decided to use a cod-ing which maps the parameters,,,,, and to a string of six8-bit integerswhich range from to.The following tableshows how the encoding and decoding is done:A class of probabilistic optimization methodswhich has come into fashion in the last years aregenetic algorithms(GAs).They can be regardedas simplified simulations of an evolution process,based on the principles of genetic reproduction em-ploying mechanisms,such as selection,mutation,and sexual reproduction.Another important differ-ence between GAs and conventional optimizationalgorithms is that GAs do not operate on singlepoints but on whole populations of points(which,in this case,are binary strings).Wefirst tried a standard GA(see[4]or[5])with proportional(standard roulette wheel)selec-tion,one-point crossing over with uniform selec-tion of the crossing point,and bitwise mutation.The size of the population was constant,the length of the strings was48(compare with the cod-ing above,see[4]for an overview of more sophis-ticated variants of GAs).In order to compare the performance of the GAswith other well-known probabilistic optimizationmethods,we additionally considered the followingmethods:Hill climbing:always moves to the best-fittedneighbor of the current string until a localmaximum is reached;the initial string is gen-erated randomly.Simulated annealing:powerful,often used prob-abilistic method which is based on the imi-tation of the solidification of a crystal underslowly decreasing temperature(see[9]for adetailed description)Each one of these methods requires only a few bi-nary operations in each step.Most of the time isconsumed by the evaluation of thefitness function.So,it is near at hand to take the number of evalua-tions as a measure for the speed of the algorithms. ResultsAll these algorithms are probabilistic methods,therefore their results are not well-determined,theycan differ randomly within certain boundaries.Inorder to get more information about their averagebehavior,we tried out each one of them timesfor one certain problem.For the given problem wefound out that the maximal degree of matching be-tween the reference classification and the classifi-cation actually obtained by the fuzzy system was94.3776%.See the table in Figure4.In this tableis thefitness of the best and is thefitnessof the worst solution;denotes the averagefitnessof the20solutions,denotes the standard devia-tion of thefitness values of the20solutions,and#stands for the average number of evaluations of thefitness function which was necessary to reach thesolution.The hill climbing method with a random selec-tion of the initial string converged rather quickly.Unfortunately,it was always trapped in a localmaximum,but never reached the global solution(atleast in these trials).The simulated annealing algorithm showedsimilar behavior at the very beginning.After tun-ing the parameters involved,the performance im-proved remarkably.The raw genetic algorithm was implementedwith a population size of20;the crossing overmax#94.365993.553686294.364893.5639151094.377394.26972196894.376094.2485991094.376094.2775746094.377694.369318631Figure4:A comparison of the results of various probabilistic optimization methodsprobability was set to,the mutation probabil-ity was for each byte.It behaved pretty well from the beginning,but it seemed inferior to the improved simulated annealing.Next,we tried a hybrid GA,where we kept the genetic operations and parameters of the raw GA,but every50th generation the best-fitted indi-vidual was taken as initial string for a hill climb-ing method.Although the performance increased slightly,the hybrid method still seemed to be worse than the improved simulated annealing algorithm. The reason that the effects of this modification were not so dramatic might be that the probabil-ity is rather high that the best individual is already a local maximum.So we modified the procedure again.This time a randomly chosen individual of every25th generation was used as initial solution of the hill climbing method.The result exceeded the expectations by far.The algorithm was,in all cases,nearer to the global solution than the im-proved simulated annealing was(compare with ta-ble in Figure4)but,surprisingly,sufficed with less invocations of thefitness function.The graph in Figure4shows the results graphically.Each line in this graph corresponds to one algorithm.The curve shows,for a givenfitness value,how many of the 20different solutions had afitness higher or equal to.It can be seen easily from this graph that the hybrid GA with random selection led to the best results.Note that the-axis is not a linear scale in thisfigure.It was transformed in order to make small differences visible.6ConclusionIn thefirst part of this paper we demonstrated the synergy which lies in the combination of fuzzy systems with,more or less,conventional methods.This combination is in particular suitable for de-signing specific algorithms for time-critical prob-lems.However,this specifity often results in a loss of universality.In the second part we showed the suitability of genetic algorithms forfinding the optimal parame-ters of a fuzzy system,especially if the analytical properties of the objective function are bad.More-over,hybridization has been discovered as an enor-mous potential for improvements of genetic algo-rithms.References1.P.Bauer,U.Bodenhofer,E.P.Klement.“A fuzzymethod for pixel classification and its application to print inspection”.Proc.IPMU'96,Granada,July 1996,pp.1301–1305.2.P.Bauer,U.Bodenhofer,E.P.Klement.“A fuzzy al-gorithm for pixel classification based on the discrep-ancy norm”.Proc.FUZZ-IEEE'97,New Orleans, September1996,pp.2007–2012.3.J.C.Bezdek,S.K.Pal.Fuzzy Models for PatternRecognition.IEEE Press,New York,1992.4. D.E.Goldberg.Genetic Algorithms in Search,Opti-mization,and Machine Learning.Addison-Wesley, Reading,MA,1989.5.J.H.Holland.Adaptation in Natural and ArtificialSystems.The MIT Press,Cambridge,MA,1992. 6.H.Neunzert,B.Wetton.“Pattern recognition usingmeasure space metrics”.Technical Report28,Uni-versit¨a t Kaiserslautern,Fachbereich Mathematik, November1987.7. A.Rosenfeld,A.C.Kak.Digital Picture Process-ing,volume II.Academic Press,San Diego,CA, second edition,1982.8. E.H.Ruspini.“A new approach to clustering”.In-formation&Control.15,22–32,1969.9.P.J.M.van Laarhoven,E.H.L.Aarts.SimulatedAnnealing:Theory and Applications.Kluwer Aca-demic Publishers,Dordrecht,1987.。