建筑物图像识别
- 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)。
建筑物图像识别研究
摘要:随着城市化进程的不断发展,城市人口、经济等要素密切影响着城市建筑物的几何形态与分布。从航拍图像或者卫星图像上快速准确的获取城市建筑物的信息不仅有利于地理空间数据的更新,而且对于有效监测新增建筑等城市专题信息有重要意义。不仅如此,近年来基于卫星图像或者航拍图像的建筑物识别与轮廓提取算法的提出与研究改进,更有助于实现对城市建筑物的识别和分类。其对GIS 数据的获得、对影像的理解、大比例尺制图和及其它许多对遥感数据的应用都具有重要的意义。
基于此,本文从建筑物轮廓的几何形态出发,对卫星图像中的建筑物轮廓的提取方法进行了研究。主要进行了如下几方面工作:
1、阐述了建筑物提取的基础理论。对典型城市建筑物进行了特征分析,学习并理解在建筑物的轮廓提取中所要用到的算法,从中选择合适的边缘检测与轮廓提取算法。
2、在对已有方法进行总结归纳的基础上,系统的研究了基于灰度特征与二值数学形态学的轮廓描述算法和基于边缘特征的Hough变换直线提取算法,对这两种算法进行改进,初步实现了典型城市建筑物的半自动提取。
3、针对本文出现的方法进行编程实现,并针对试验结果对这两种方法进行综合的检测与评价,分析了这两种方法的特点及不足。
关键词:卫星图像;建筑物轮廓;二值数学形态学;边缘检测;直线提取
Building Image Recognition and Research
Abstract: With the continuing developing of urbanization, the urban population, economic and other factors are closely affecting the city building’s geometry and distribution. It is not only conductive to the data updating of geospatial, but also has important significance of the thematic information for effective monitoring of new city buildings and so on that accessing information form the city buildings aerial images or satellite images fast and accurately. Moreover, in recent years, building on satellite images to identify and contour extraction or aerial imagery and research to improve the algorithm proposed, but also help achieve the identification and classification of city buildings. Its GIS data acquisition, image understanding, large scale mapping and many other applications are of great significance.
This paper is start form the geometry of the building outline, doing research of the extraction methods for the building outline which from high resolution images. Mainly for the following work areas:
1.Describes the basic theory of building extraction. Urban buildings typical
characteristic analysis, systematic study of image processing algorithms to use in extraction of the building, from which to select the appropriate algorithm for edge detection and contour extraction.
2.On the basis of summarizing the existing methods, the proposed feature
extraction based on gray and binary mathematical morphology the contour edge detection algorithm based on linear Hough transform algorithm, the semi-automatic extraction of typical urban buildings.
3.We were against the proposed method programming, and comprehensive
testing and evaluation of these two methods for the test results, the characteristics and drawbacks of the two methods.
Key words:Satellite images, Building outline, Binary mathematical morphology,Edge detection,Line Extraction
目录
1 绪论 (1)
1.1 研究背景和意义 (1)
1.1.1 研究背景 (1)
1.1.2 研究意义 (1)
1.2 研究现状 (2)
1.3 本文的主要内容 (3)
2 建筑物图像预处理的基础理论 (6)
2.1 建筑物的特征分析 (6)
2.2 建筑物几何特征分类 (6)
2.3 图像增强 (8)
2.3.1 灰度变换 (8)
2.3.2 直方图均衡变换 (14)
2.4 本章小结 (19)
3 基于灰度特征的建筑物轮廓提取 (20)
3.1 图像二值化 (20)
3.2 二值数学形态学 (25)
3.2.1 膨胀与腐蚀 (25)
3.2.2 开启与闭合 (27)
3.3 本章小结 (31)
4 基于边缘特征的建筑物轮廓提取 (32)
4.1 边缘检测 (32)
4.1.1 边缘检测技术 (32)
4.1.2 边缘的分析 (33)
4.1.3 关于梯度的概念 (33)
4.2 边缘检测算子 (34)
4.2.1 罗伯特(Robert)边缘算子 (35)
4.2.2 索贝尔(Sobel)边缘算子 (35)
4.2.3 Prewitt边缘算子 (35)
4.2.4 拉普拉斯(Laplacian)边缘算子 (36)
4.2.5 坎尼(Canny)边缘算子 (37)
4.2.6 边缘检测算子比较 (40)
4.3 直线段提取 (42)