赵_Chapter_7_Human_Skin_Texture_Analysis

合集下载

人脸识别中的姿态问题研究

-I-
哈尔滨工业大学工学博士学位论文
果更加精确。 (3) 提出了一种基于局部线性回归(LLR)的姿态校正方法。通过将姿态校正
形式化为从非正面人脸估计其虚拟正面视图的预测问题，本文提出了一种基于线性回归的正面视图预测方法，该方法可以在仅标定输入非正面人脸双眼中心位置的条件下，通过线性回归算法预测出其虚拟正面视图。考虑到非正面人脸视图与正面人脸视图之间的映射对于不同人而言实际上并不相同，我们进一步改进上述算法，提出了一种基于分块模式的姿态校正方法，即所谓的局部线性回归(LLR)姿态校正方法。该方法基于 3D 人脸结构的先验知识，将人脸区域分成若干小的面片，在小的面片上进行上述线性回归操作以获得更为精确的预测结果。实验结果表明，无论从视图校正的视觉效果，还是从人脸识别的精度来看，此方法都具有很好的性能。
(4) 提出了一种基于 3D 稀疏变形模型(3D SDM)的 3D 人脸重建及姿态校正方法。本方法利用人脸类的 3D 形状先验知识，根据单幅任意姿态下的人脸图像，重构其特定的 3D 形状信息。在 3D 人脸重构的基础上，通过图形绘制的方法得到正面姿态下的人脸视图，用于解决人脸识别中的姿态问题。该方法假设人脸的 3D 形状分布为高斯分布，将所有人脸的 3D 形状分布空间用主成分分析(PCA) 模型表示。按照预先定义的 2D 人脸图像上的关键特征点，从稠密的 PCA 模型中得到稀疏的版本，即稀疏变形模型。基于此，在输入图像 2D 特征点的驱动下，恢复得到该特定人的 3D 形状，进而重构其 3D 人脸，实现姿态校正及任意姿态下的虚拟视图生成。实验表明：通过 3D 人脸重建得到的虚拟视图结果，从视觉上来看更加接近于真实人脸图像。同时，以 Gabor PCA+LDA 方法作为识别策略，在 CMU PIE 数据库中 45 度以内的图像上测试，将姿态校正后的视图作为输入，平均识别率达到了 97.5%左右，极大的改善了识别系统对于人脸姿态图像的适应能力。

人体面部结构英语

人体面部结构英语The Human Facial Structure: An Insight into the Complexity and BeautyIntroduction:The human face is a remarkable structure that portrays emotions, thoughts, and personality. It is the most distinguishing feature of an individual, making each person unique. Understanding the human facial structure is essential in various fields, such as medicine, art, anthropology, and even everyday interactions. In this article, we will explore the intricate details of the human facial structure, its components, and their functions, and delve into the importance of studying facial anatomy in the English language.Components of the Facial Structure:The human face comprises various components that work together harmoniously to create the mesmerizing beauty we witness every day. These components include the forehead, eyebrows, eyes, nose, cheeks, mouth, chin, and ears.Forehead:Located at the upper part of the face, the forehead serves as a protective covering for the brain. It plays a significant role in facial expressions, especially when it comes to showcasing surprise, concern, or concentration.Eyebrows:The eyebrows, positioned above the eyes, have a dual purpose. Firstly, they protect the eyes from sweat, dust, and other foreign particles. Secondly, eyebrows play a crucial role in facial expressions, conveying emotions such as anger, surprise, and skepticism.Eyes:The eyes, often referred to as the "windows to the soul," are the most captivating feature of the human face. They facilitate vision, allowing us to perceive the worldaround us. Eyes also convey a wide range of emotions, including happiness, sadness, fear, and excitement, making them a vital part of non-verbal communication.Nose:The nose, positioned at the center of the face, is responsible for the sense of smell. Additionally, the shape, size, and structure of the nose contribute significantly to the overall facial aesthetics, as it is a prominent feature.Cheeks:The cheeks, located on the sides of the face, add volume and contour to the facial structure. They are involved in various facial expressions, including smiling, laughing, and pouting.Mouth:The mouth, consisting of the lips, teeth, and tongue, plays a crucial role in communication and digestion. It allows us to articulate speech, express emotions, and consume food. The lips, in particular, are known for their sensuality and play a vital role in non-verbal communication.Chin:The chin, located at the bottom of the face, provides support to the lower lip and adds definition to the jawline. It also helps in facial expressions, conveying emotions such as determination, defiance, and contemplation.Ears:The ears, situated on both sides of the face, are responsible for the sense of hearing. Apart from their auditory function, the ears contribute to the overall aesthetics of the face. They vary in shape, size, and position, adding uniqueness to each individual.Importance of Studying Facial Anatomy in the English Language:Studying facial anatomy in the English language is essential for various reasons, including communication, medical professions, and artistic endeavors.Communication:Facial expressions and body language are integral parts of effective communication, and the human face plays a central role in conveying emotions, intentions, and attitudes. Understanding the various facial structures and their functions helps individuals accurately interpret and express emotions in English, enhancing effective communication in both personal and professional settings.Medical Professions:In the medical field, understanding the human facial structure is crucial. Healthcare professionals, such as doctors, nurses, and dentists, need to have a comprehensive knowledge of facial anatomy to diagnose and treat facial deformities, perform surgeries, administer injections, and conduct dental procedures. Accurate and clear communication with patients, colleagues, and other medical professionals is essential, and proper use of English terminology related to the facial structure is a fundamental aspect of this.Artistic Endeavors:Artists, whether they specialize in painting, sculpting, or digital art, benefit greatly from studying facial anatomy. A deep understanding of the human facial structure enables artists to create realistic and expressive portraits, capturing the essence of their subjects. Knowledge of English terminology related to facial features allows artists to communicate effectively with a global audience, enhancing their reach and understanding.Conclusion:The human facial structure is an intricate and captivating creation, with its various components working together to express emotions, communicate, and define individuality. Understanding the complexity and beauty of the facial structure is essential in various fields, including communication, medicine, and art. By studying facial anatomy, both in terms of structure and English terminology, individuals can enhancetheir communication skills, medical expertise, and artistic endeavors, ultimately gaining a deeper appreciation for the beauty that lies within the human face.。

Image Anal Stereol 200120207-211 Original Research Paper DAMAGE MORPHOLOGICAL PARAMETERS

Original Research PaperDAMAGE MORPHOLOGICAL PARAMETERSJ EAN-L OUIS C HERMANT, G UILLAUME B OITIER, S EVERINE D ARZENS, M ICHEL C OSTER AND L ILIANE C HERMANTLERMAT, URA CNRS 1317, ISMRA, 6 Bd Maréchal Juin, 14050 Caen Cedex, France(Accepted October 27, 2001)ABSTRACTThis paper shows how it is possible to characterize and quantify the damages in materials using classical tools of automatic image analysis. Examples presented concern ceramic matrix composites, i.e. high tech materials. It gives important information to support the deformation and rupture mechanism of materials under mechanical solicitations.Keywords: ceramic matrix composites, damage parameters, mathematical morphology.INTRODUCTIONIn many brittle materials, under a mechanical solicitation with or without temperature, many damages appear such as microcracks, grain or fiber/matrix decohesion, etc. If one can be able to quantify these damages, it could be a way to bring new material parameters to the mechanical engineers. For example for such materials one can use the damage mechanics, as proposed by Kachanov (1956), based on the decrease of the Young's modulus as the damages are in progress. But no morphological parameters of the damages are introduced in the formalism of the damage mechanics. The aim of this paper is to show how it is possible to use the classical methods of automatic image analysis to quantify the damages of some ceramic matrix composites during creep investigations.MATERIALS AND METHODSMaterials investigated are CMCs, i.e. ceramic matrix composites, fabricated by SNECMA (Division Moteurs et Fusées, Saint-Médard en Jalles, France), according to a specific chemical vapor infiltration process into an architecture made of carbon or silicon carbide fibers (Christin et al., 1979; Naslain, 1999). That leads to C f-SiC or SiC f-SiBC composites depending on the chemical vapor content. These materials are in fact ceramic matrices reinforced by continuous ceramic fibers, for aeronautics and space applications (Lamicq, 1990; Spriet and Habarou, 1997).Creep tests were performed in tension, under argon, at different stresses (up to 220 MPa) and temperatures (up to 1673 K), using dogbone shaped specimens. Observations of damages were made through a scanning electron microscope, SEM (Jeol JSM 6400) or an optical microscope (Olympus BH2 type).Automatic image analysis was performed either with a Matra Pericolor 3100 (Matra Cap System MS2I) or the Aphelion software from ADCIS. After the damage detection, classical tools of mathematical morphology (Serra, 1982; Coster and Chermant, 1985; 1989) were used in order to extract and quantify the correct damage features.After tensile creep or compression, one observes on the surface (and in the bulk) of all these materials many brittle damages like crystal or fiber debonding, matrix microcracking, fiber and/or yarn bridging or fracture, etc. A major difficulty for such damage investigation is to select a correct magnification. When one increases the magnifications, the microcrack length and the tortuosity also increase until its size is reached at the atomic scale. But it does not mean that the cracks are fractal. So one must use a magnification which is related to the scale of the microstructure of the investigated material and compatible with a true and valid quantification (i.e. on a number of frames of measurements including enough cracks, to obtain a statistically valid number). For these CMCs one has to choose a magnification in order to have a yarn totally included in the frame of measurements to count all the microcracks inside (to avoid any bias if the yarn intersects the frame of measurements), or for the thickness of the cracks a magnification of about ×1000.RESULTSTwo types of results on the damage quantification will be presented: the surface fraction of microcracks in crept C f-SiC composites and the crack opening evolution in crept SiC f-SiBC composites.MICROCRACK SURFACE FRACTIONTo extract the microcracks which have the same grey tone level than the pores, one successively uses several steps, but to be valid the frame of measurements must, of course, totally include the yarn to be analyzed to avoid any bias (Serra, 1982; Coster and Chermant, 1985; 1989): x a threshold to extract microcracks + pores (Fig. 1b); y the skeleton function to reduce microcracks and pores to the thickness of 1 pixel; z a clipping until the smallest skeletons which are related to the pores, will be lost; then { microcracks are reconstructed to obtain their initial sizes (Fig. 1c) and by difference between image y and { one obtains the image of pores.On these images we have measured the mean values of the surface area of the yarns, A(Y), the surface fraction of the intra-yarn pores, A A(P), and of the microcracks, A A(f), in a 2.5D C f-SiC in its as-received state and after creep tests at 1273 K and 1673 K under 200 MPa. Values are given in Table 1.One can note from such investigations that there is a swelling of the yarns due to the development of the matrix microcracking process.CRACK OPENING EVOLUTIONIn the case of SiC f-SiBC, the matrix microcracks were followed, during creep and using step-creeping tests (Darzens, 2000), by the techniques of automatic image analysis as previously described. Then the maximum thicknesses of the microcracks were calculated from the greatest inscriptible square in the considered microcrack, without considering the pores when they are crossed by a microcrack. Fig. 2 illustrates that measurement.During the loading the microcracks completely cross the inter-yarn zone perpendicular to the stress direction and reach the longitudinal yarns. Thereafter there is a deviation of the microcracks in mode II in the pyrolytic carbon layer parallel to the stress direction, and the opening of these microcracks is then very low. So the damage quantification was only undertaken in the transverse yarns. Fig. 3 presents the evolution of a transverse microcrack during different step-creeping tests at 1473 K, under 120 MPa.One notices an increase of the opening of the microcracks during the creep process. Each given experimental result value (Fig. 3) is the mean for 35 microcrack measurements corresponding to 75 acquirements for each microcracks. The accuracy of measurements is 0.25 µm.Table 1. Mean values of the surface area of yarns, A(Y), of the surface fraction of the intra-yarn pores, A A(P), and of the microcracks, A A(f), in 2.5D C f-SiC specimens, as-received and after tensile creep at 1273 K and 1673 K, at 220 MPa under argon.Specimen A(Y) (µm2)A A(P) (%)A A(f) (%)as-received136 000 4.5 1.41273K, 220 MPa, 212 h154 000 5.5 6.01673K, 220 MPa, 127 h147 0008.86.4a)b)c)Fig. 1. Image processing to investigate automatically the microcrack morphology of a 2.5D C f-SiC, after creep tests at 1673 K and 220 MPa: a) image of a yarn totally included in the frame of measurements, b) microcracks and pores after threshold, c) reconstructed image of the microcracks alone.a)b)Fig. 2. Measurement of the maximum width of a crack using a square structuring element in a SiC f-SiBC composite, creep tested at 1473 K under 120 MPa, after 3 h (a) and 10 h (b).t = 2 h,εin = 0.37%t = 9 h 40 (rupture),εin = 0.63%t = 1 h,εin = 0.25%20 µmFig. 3. Evolution of a transverse matrix microcrack in a N1 SiC f-SiBC composite, during the creep at 1473 K,under 120 MPa, for different creep times (t) and inelastic deformations (εin ).The number of analyzed microcracks is, of course,too low to have a result statistically representative of the microcrack array in the hot zone of the crept specimens.Nevertheless, it gives some correct information on the trends. After 50 h of creep, the mean crack opening is 5 times larger than after the loading. As the standard deviation values are low, these results appear homogeneous with regard to the crack opening.Fig. 4 shows the change of the microcrack opening,e m , as a function of the inelastic strain, εin , for two types of composites N1 and N2 SiC f-SiBC, after creep tests at 1473 K under 120 MPa (the difference in N1 and N2 lies in the layer sequence; the inelastic strain corresponds to the total creep strain less the elastic strain calculated from the Hooke’s law). There is a linear dependence between e m and εin . It appears that for N2 composites the microcrack opening evolves more largely than for N1 composites, while the microcrack inter-distance remains similar: 650 µm and 610 µm, respectively for N1 and N2 composites.(%)e m ( µ m )ε i nFig. 4. Change in the microcrack opening, e m , as a function of the inelastic strain, εin , for N1 and N2SiC f -SiBC composites, creep tested at 1473 K, under 120 MPa.This difference informs on the existence of two different damage mechanisms which confirms the macroscopical and damage mechanics results (Darzens,2000).DISCUSSIONAlthough these morphological measurements are simple and classical, they bring very important information regarding the creep mechanisms.It has been previously shown that damage creep is a possible mechanism for these CMCs when applied temperatures and stresses are not high enough to activate dislocation motions or/and diffusionphenomena (Chermant, 1995; 2000). Several authors (Boitier et al ., 1999; Darzens, 2000) have shown that there is x first a rapid formation of a crack network until its saturation, and y then another mechanism which appears very slow. Both the changes in the surface areas and microcrack openings of the matrix microcracks confirm that the second mechanism can be assimilated to a slow crack growth process, SCG,well known for glass and ceramic materials (Wierderhorn et al ., 1968; Evans, 1972).These damage morphological parameters are not the only ones, others can be also investigated such as the mean number of cracks per unit surface, the mean crack length per unit surface, the mean distance between cracks, their orientation, etc. Some of these parameters were investigated in the case of concrete materials by several authors (Ringot, 1988; Gérard et al ., 1996; Nemati et al ., 1998; Ammouche et al ., 1999;2000).CONCLUSIONThis short paper has shown the importance of the morphological quantification of brittle damages in materials mechanically tested. It helps the mechanical engineers as it brings them morphological parameters of the damages in solicitated materials which have now to be introduced in the mathematical formalisms of the damage mechanics.ACKNOWLEDGEMENTSThis work has been performed in the frame of the "Pôle Traitement et Analyse d'Images" (Image Processing and Analysis Pole) of Basse-Normandie (Pôle TAI). It has been supported by SNECMA,Division Moteurs et Fusées, S t Médard en Jalles,France, and the CNRS for the ceramic matrix composite investigations. The authors thank the CNRS and Région of Basse-Normandie for the fellowships (GB and SD).REFERENCESAmmouche A (1999). Caractérisation automatique de lamicrofissuration des bétons par traitement d’images.Application à l’étude de différents faciès de dégradation.Thèse de Doctorat of the University of Bordeaux I, et Philisophiae Doctor of the University Laval-Québec.Ammouche A, Breysse D, Hornain H, Didry O, Marchand J(2000). A new image analysis technique for the quantitative assessment of microcracks in cement-based materials. Cem Concr Res 30:25-35.Boitier G, Chermant JL, Vicens J (1999). Multiscale investigation of the creep behavior of a 2.5D C f-SiC composites. J Mat Sci 34:1-9.Chermant JL (1995). Creep behavior of ceramic matrix composites. Sil Ind 60:261-73.Chermant JL (2000). Damage and creep of CMCs. Ceram Trans 103:409-27.Christin F, Naslain R, Bernard C (1979). A thermodynamic and experimental approach of silicon carbide CVD.Application to the CVD-infiltration of porous carbon-carbon composites. In: Sedwick TO, Lydtin H, eds.Proceedings of the 7th International Conference on CVD. Princeton: The Electrochemical Society, 499-514. Coster M, Chermant JL (1985; 1989). Précis d'analyse d'images. Les Editions du CNRS; 2nd edition, Les Presses du CNRS.Darzens S (2000). Fluage en traction sous argon et microstructure de composites SiC f-SiBC. Thèse de Doctorat of the University of Caen.Evans AJ (1972). A method for evaluating the time-dependent failure characteristics of brittle materials, and its application to polycrystalline alumina. J Mat Sci 7:1137-46.Gérard B, Breysse D, Ammouche A, Houdusse O, Didry O (1996). Cracking and permeability of concrete under tension. Mat Struct 29:141-51.Kachanov L (1958). Rupture time under creep conditions.Izv Akad Nauk SSR 8:26-31.Lamicq P (1990). La percée des composites thermo-structuraux. Les nouveaux matériaux, la mécanique en environnement sévère. In: Dunod, ed. Science et Défense 90:80-92.Naslain R (1999). Materials design and processing of high temperature ceramic matrix composites: state of the art and future trends. Adv Comp Mater 8:3-16.Nemati KM, Monteiro PJ, Scrivener KL (1998). Analysis of compressive stress-induced cracks in concrete. ACI Mater J 95:617-30.Ringot E (1988). Development of the map cracking in concrete under compressive loading. Cem Concr Res 18:933-42.Serra J (1982). Image Analysis and Mathematical Morphology. New York: Academic Press.Spriet P, Habarou G (1997). Applications of CMCs to turbojet engines: overview of the SEP experience. In: Fuentes M, Martinez-Esnaola JM, Daniel AM, eds.CMMC 96, San Sebastian, Spain, Sept. 9-12, 1996, Key Eng Mat 127-131:1267-76.Wiederhorn SM (1968). Moisture assisted crack growth in ceramics. Int l J Fract Mech 4:171-7.。

英文翻译

A Facial Aging Simulation Method Using flaccidity deformation criteriaAlexandre Cruz Berg Lutheran University of Brazil.Dept Computer ScienceRua Miguel Tostes, 101. 92420-280 Canoas, RS, Brazil berg@ulbra.tche.br Francisco José Perales LopezUniversitat les Illes Balears.Dept Mathmatics InformaticsCtra Valldemossa, km 7,5E-07071 Palma MallorcaSpainpaco.perales@uib.esManuel GonzálezUniversitat les Illes Balears.Dept Mathmatics InformaticsCtra Valldemossa, km 7,5E-07071 Palma MallorcaSpainmanuel.gonzales@uib.esAbstractDue to the fact that the aging human face encompasses skull bones, facial muscles, and tissues, we render it using the effects of flaccidity through the observation of family groups categorized by sex, race and age. Considering that patterns of aging are consistent, facial ptosis becomes manifest toward the end of the fourth decade. In order to simulate facial aging according to these patterns, we used surfaces with control points so that it was possible to represent the effect of aging through flaccidity. The main use of these surfaces is to simulate flaccidity and aging consequently.1.IntroductionThe synthesis of realistic virtual views remains one of the central research topics in computer graphics. The range of applications encompasses many fields, including: visual interfaces for communications, integrated environments of virtual reality, as well as visual effects commonly used in film production.The ultimate goal of the research on realistic rendering is to display a scene on a screen so that it appears as if the object exists behind the screen. This description, however, is somewhat ambiguous and doesn't provide a quality measure for synthesized images. Certain areas, such as plastic surgery, need this quality evaluation on synthesized faces to make sure how the patient look like and more often how the patient will look like in the future. Instead, in computer graphics and computer vision communities, considerable effort has been put forthto synthesize the virtual view of real or imaginary scenes so that they look like the real scenes.Much work that plastic surgeons put in this fieldis to retard aging process but aging is an inevitable process. Age changes cause major variations in the appearance of human faces [1]. Some aspects of aging are uncontrollable and are based on hereditary factors; others are somewhat controllable, resulting from many social factors including lifestyle, among others [2].1.1.Related WorkMany works about aging human faces have been done. We can list some related work in the simulation of facial skin deformation [3].One approach is based on geometric models, physically based models and biomechanical models using either a particle system or a continuous system.Many geometrical models have been developed, such as parametric model [4] and geometric operators [5]. The finite element method is also employed for more accurate calculation of skin deformation, especially for potential medical applications such as plastic surgery [6]. Overall, those works simulate wrinkles but none of them have used flaccidity as causing creases and aging consequently.In this work is presented this effort in aging virtual human faces, by addressing the synthesis of new facial images of subjects for a given target age.We present a scheme that uses aging function to perform this synthesis thru flaccidity. This scheme enforces perceptually realistic images by preserving the identity of the subject. The main difference between our model and the previous ones is that we simulate increase of fat and muscular mass diminish causing flaccidity as one responsible element for the sprouting of lines and aging human face.In the next section will plan to present the methodology. Also in section 3, we introduce the measurements procedure, defining structural alterations of the face. In section 4, we present a visual facial model. We describe age simulation thrua deformation approach in section 5. In the last section we conclude the main results and future work.2.MethodologyA methodology to model the aging of human face allows us to recover the face aging process. This methodology consists of: 1) defining the variations of certain face regions, where the aging process is perceptible; 2) measuring the variations of those regions for a period of time in a group of people and finally 3) making up a model through the measurements based on personal features.That could be used as a standard to a whole group in order to design aging curves to the facial regions defined.¦njjjpVM2.1Mathematical Background and AnalysisHuman society values beauty and youth. It is well known that the aging process is influenced by several parameters such: feeding, weight, stress level, race, religious factors, genetics, etc. Finding a standard set of characteristics that could possibly emulate and represent the aging process is a difficult proposition.This standard set was obtained through a mathematical analysis of some face measurements in a specific group of people, whose photographs in different ages were available [7]. To each person in the group, there were, at least, four digitized photographs. The oldest of them was taken as a standard to the most recent one. Hence, some face alterations were attained through the passing of time for the same person.The diversity of the generated data has led to the designing of a mathematical model, which enabled the acquiring of a behavior pattern to all persons of the same group, as the form of a curve defined over the domain [0,1] in general, in order to define over any interval [0,Į] for an individual face. The unknown points Įi are found using the blossoming principle [8] to form the control polygon of that face.The first step consisted in the selection of the group to be studied. Proposing the assessment of the face aging characteristics it will be necessary to have a photographic follow-up along time for a group of people, in which their face alterations were measurable.The database used in this work consisted of files of patients who were submitted to plastic surgery at Medical Center Praia do Guaíba, located in Porto Alegre, Brazil.3.MeasurementsAccording to anatomic principles [9] the vectors of aging can be described aswhich alter the position and appearance of key anatomic structures of the face as can be shown in figure 1 which compares a Caucasian mother age 66 (left side) with her Caucasian daughters, ages 37 (right above) and 33 (right below) respectively.Figure 1 - Observation of family groupsTherefore, basic anatomic and surgical principles must be applied when planning rejuvenative facial surgery and treating specific problems concomitantwith the aging process.4.Visual Facial ModelThe fact that human face has an especially irregular format and interior components (bones, muscles and fabrics) to possess a complex structure and deformations of different face characteristics of person to person, becomes the modeling of the face a difficult task. The modeling carried through in the present work was based on the model, where the mesh of polygons corresponds to an elastic mesh, simulating the dermis of the face. The deformations in this mesh, necessary to simulate the aging curves, are obtained through the displacement of the vertexes, considering x(t) as a planar curve, which is located within the (u,v ) unit square. So, we can cover the square with a regular grid of points b i,j =[i/m,j/n]T ; i=0,...,m; j=0,...,n. leading to every point (u,v ) asfrom the linear precision property of Bernstein polynomials. Using comparisons with parents we can distort the grid of b i,j into a grid b'i,j , the point (u,v )will be mapped to a point (u',v') asIn order to construct our 3D mesh we introduce the patch byAs the displacements of the vertexes conform to the certain measures gotten through curves of aging and no type of movement in the face is carried through, the parameters of this modeling had been based on the conformation parameter.4.1Textures mappingIn most cases the result gotten in the modeling of the face becomes a little artificial. Using textures mapping can solve this problem. This technique allows an extraordinary increase in the realism of the shaped images and consists of applying on the shaped object, existing textures of the real images of the object.In this case, to do the mapping of an extracted texture of a real image, it is necessary that the textureaccurately correspond to the model 3D of that is made use [9].The detected feature points are used for automatic texture mapping. The main idea of texture mapping is that we get an image by combining two orthogonal pictures in a proper way and then give correct texture coordinates of every point on a head.To give a proper coordinate on a combined image for every point on a head, we first project an individualized 3D head onto three planes, the front (x, y), the left (y, z) and the right (y, z) planes. With the information of feature lines, which are used for image merging, we decide on which plane a 3D-head point on is projected.The projected points on one of three planes arethen transferred to one of feature points spaces suchas the front and the side in 2D. Then they are transferred to the image space and finally to the combined image space.The result of the texture mapping (figure 2) is excellent when it is desired to simulate some alteration of the face that does not involve a type of expression, as neutral. The picture pose must be the same that the 3D scanned data.¦¦¦ mi nj lk n j m i lk k j i w B v B u B b w v u 000,,)()()(')',','(¦¦ m i nj n jmij i v B u B b v u 00,)()(),(¦¦ m i nj n j m i j i v B u B b v u 00,)()(')','(¦¦¦ mi nj lk n j m i lk k j i w B v B u B b w v u 000,,)()()(')',','(Figure 2 - Image shaped with texturemapping5.Age SimulationThis method involves the deformation of a face starting with control segments that define the edges of the faces, as¦¦¦ mi nj lk n j m i lk k j i w B v B u B b w v u 000,,)()()(')',','(Those segments are defined in the original face and their positions are changed to a target face. From those new positions the new position of each vertex in the face is determined.The definition of edges in the face is a fundamental step, since in that phase the applied aging curves are selected. Hence, the face is divided in influencing regions according to their principal edges and characteristics.Considering the face morphology and the modeling of the face aging developed [10], the face was divided in six basic regions (figure 3).The frontal region (1) is limited by the eyelids and the forehead control lines. The distance between these limits enlarges with forward aging.The orbitary region (2) is one of the most important aging parameters because a great number of wrinkles appears and the palpebral pouch increases [11]. In nasal region (3) is observed an enlargement of its contour.The orolabial region (4) is defined by 2 horizontal control segments bounding the upper and lower lips and other 2 segments that define the nasogenian fold. Figure 3 - Regions considering the agingparametersThe lips become thinner and the nasogenian fold deeper and larger. The mental region (5) have 8 control segments that define the low limit of the face and descend with aging. In ear curve (6) is observed an enlargement of its size. The choice of feature lines was based in the characteristic age points in figure 6.The target face is obtained from the aging curves applied to the source face, i.e., with the new control segment position, each vertex of the new image has its position defined by the corresponding vertex in the target face. This final face corresponds to the face in the new age, which was obtained through the application of the numerical modeling of the frontal face aging.The definition of the straight-line segment will control the aging process, leading to a series of tests until the visual result was adequate to the results obtained from the aging curves. The extremes of the segments are interpolated according to the previously defined curves, obtained by piecewise bilinear interpolation [12].Horizontal and vertical orienting auxiliary lines were defined to characterize the extreme points of the control segments (figure 4). Some points, that delimit the control segments, are marked from the intersection of the auxiliary lines with the contour of the face, eyebrow, superior part of the head and the eyes. Others are directly defined without the use of auxiliary lines, such as: eyelid hollow, eyebrow edges, subnasion, mouth, nasolabial wrinkle andnose sides.Figure 4 - Points of the control segmentsOnce the control segments characterize the target image, the following step of the aging process can be undertaken, corresponding to the transformations of the original points to the new positions in the target image. The transformations applied to the segments are given by the aging curves, presented in section 4.In the present work the target segments are calculated by polynomial interpolations, based on parametric curves [12].5.1Deformation approachThe common goal of deformation models is to regulate deformations of a geometric model by providing smoothness constraints. In our age simulation approach, a mesh-independent deformation model is proposed. First, connected piece-wise 3D parametric volumes are generated automatically from a given face mesh according to facial feature points.These volumes cover most regions of a face that can be deformed. Then, by moving the control pointsof each volume, face mesh is deformed. By using non-parallel volumes [13], irregular 3D manifolds are formed. As a result, smaller number of deformvolumes are necessary and the number of freedom incontrol points are reduced. Moreover, based on facialfeature points, this model is mesh independent,which means that it can be easily adopted to deformany face model.After this mesh is constructed, for each vertex on the mesh, it needs to be determined which particularparametric volume it belongs to and what valueparameters are. Then, moving control points ofparametric volumes in 3D will cause smooth facialdeformations, generating facial aging throughflaccidity, automatically through the use of the agingparameters. This deformation is written in matricesas , where V is the nodal displacements offace mesh, B is the mapping matrix composed ofBernstein polynomials, and E is the displacementvector of parametric volume control nodes.BE V Given a quadrilateral mesh of points m i,j ,, we define acontinuous aged surface via a parametricinterpolation of the discretely sampled similaritiespoints. The aged position is defined via abicubic polynomial interpolation of the form with d m,n chosen to satisfy the known normal and continuity conditions at the sample points x i,j .>@>M N j i ,...,1,...,1),(u @@>@>1,,1,),,( j j v i i u v u x ¦3,,),(n m n m n m v u d v u x An interactive tool is programmed to manipulate control points E to achieve aged expressions making possible to simulate aging through age ranges. Basic aged expression units are orbicularis oculi, cheek, eyebrow, eyelid, region of chin, and neck [14]. In general, for each segment, there is an associated transformation, whose behavior can be observed by curves. The only segments that do not suffer any transformation are the contour of the eyes and the superior side of the head.5.2Deformation approachThe developed program also performs shape transformations according to the created aging curves, not including any quantification over the alterations made in texture and skin and hair color. Firstly, in the input model the subjects are required to perform different ages, as previouslymentioned, the first frame needs to be approximately frontal view and with no expression.Secondly, in the facial model initialization, from the first frame, facial features points are extracted manually. The 3D fitting algorithm [15] is then applied to warp the generic model for the person whose face is used. The warping process and from facial feature points and their norms, parametric volumes are automatically generated.Finally, aging field works to relieve the drifting problem in template matching algorithm, templates from the previous frame and templates from the initial frame are applied in order to combine the aging sequence. Our experiments show that this approach is very effective. Despite interest has been put in presenting a friendly user interface, we have to keep in mind that the software system is research oriented. In this kind of applications an important point is the flexibility to add and remove test facilities. 6.Results The presented results in the following figuresrefer to the emulations made on the frontalphotographs, principal focus of this paper, with theobjective to apply the developed program to otherpersons outside the analyzed group. The comparisonswith other photographs of the tested persons dependon their quality and on the position in which theywere taken. An assessment was made of the new positions, of the control segments. It consisted in: after aging a face, from the first age to the second one, through the use of polynomial interpolation of the control segments in the models in the young age, the new positions are then compared with the ones in the model of a relative of older age (figure 5). The processed faces were qualitatively compared with theperson’s photograph at the same age. Figure 5 - Synthetic young age model,region-marked model and aged modelAlso the eyelid hollow, very subtle falling of the eyebrow, thinning of the lips with the enlarging of the nasion and the superior part of the lip, enlargingof the front and changing in the nasolabial wrinkle.7.ConclusionsModelling biological phenomena is a great deal of work, especially when the biggest part of the information about the subject involves only qualitative data. Thus, this research developed had has a challenge in the designing of a model to represent the face aging from qualitative data.Due to its multi-disciplinary character, the developed methodology to model and emulate the face aging involved the study of several other related fields, such as medicine, computing, statistics and mathematics.The possibilities opened by the presented method and some further research on this field can lead to new proposals of enhancing the current techniques of plastic face surgery. It is possible to suggest the ideal age to perform face lifting. Once the most affected aging regions are known and how this process occurs over time. Also missing persons can be recognized based on old photographs using this technique. AcknowledgementsThe project TIN2004-07926 of Spanish Government have subsidized this work.8. References[1] Burt, D. M. et al., Perc. age in adult Caucasianmale faces, in Proc. R. Soc., 259, pp 137-143,1995.[2] Berg, A C. Aging of Orbicularis Muscle inVirtual Human Faces. IEEE 7th InternationalConference on Information Visualization, London, UK, 2003a.[3] Beier , T., S. Neely, Feature-based imagemetamorphosis, In Computer Graphics (Proc.SIGGRAPH), pp. 35-42, 1992.[4] Parke, F. I. P arametrized Models for FacialAnimation, IEEE Computer & Graphics Applications, Nov. 1982.[5] Waters, K.; A Muscle Model for Animating ThreeDimensional Facial Expression. Proc SIGGRAPH'87,Computer Graphics, Vol. 21, Nº4, United States, 1987. [6] Koch, R.M. et alia.. Simulation Facial SurgeryUsing Finite Element Models, Proceedings of SIGGRAPH'96, Computer Graphics, 1996.[7] Kurihara, Tsuneya; Kiyoshi Arai, ATransformation Method for Modeling and Animation of the Human Face from Photographs, Computer Animatio n, Springer-Verlag Tokyo, pp.45-58, 1991.[8] Kent, J., W. Carlson , R. Parent, ShapeTransformation for Polygon Objects, In Computer Graphics (Proc. SIGGRAPH), pp. 47-54, 1992. [9] Sorensen, P., Morphing Magic, in ComputerGraphics World, January 1992.[10]Pitanguy, I., Quintaes, G. de A., Cavalcanti, M.A., Leite, L. A. de S., Anatomia doEnvelhecimento da Face, in Revista Brasileira deCirurgia, Vol 67, 1977.[11]Pitanguy, I., F. R. Leta, D. Pamplona, H. I.Weber, Defining and measuring ageing parameters, in Applied Mathematics and Computation , 1996.[12]Fisher, J.; Lowther, J.; Ching-Kuang S. Curveand Surface Interpolation and Approximation: Knowledge Unit and Software Tool. ITiCSE’04,Leeds, UK June 28–30, 2004.[13]Lerios, A. et al., Feature-Based VolumeMetamorphosis, in SIGGRAPH 95 - Proceedings,pp 449-456, ACM Press, N.Y, 1995.[14]Berg, A C. Facial Aging in a VirtualEnvironment. Memória de Investigación, UIB, Spain, 2003b.[15]Hall, V., Morphing in 2-D and 3-D, in Dr.Dobb's Journal, July 1993.。

AnatomySketchSeries（解剖素描系列）

M U S C L E S O FT H E U P P E R E X T R E M I T I E SP R E F A C EThis project was inspired by the beauty of the human body and the rigorous discipline devoted to unraveling its complexity. Each structure within the body does not exist alone but establishes intricate links with surrounding structures. It is the grouping of these structures that create the human body as we know it. Fueled by the oxygen and nutrients delivered in vessels and synced by neuronal networks, the human body is a marvelous piece of natural machinery, set to consolidate and accomplish the aspirations of our minds.Through this project, I have gained a deep appreciation for the meticulous and elegant composition of each muscle group, nerve bundle, and branch of vessels. Though good medicine aims to target selective receptor in specific organs, this project was a reminder that the body is intricately linked as a whole, and effects in one corner may ripple into another. To administer medicine responsibly, one must be cognizant of the potential of a drug, an intervention, or a procedure to effect the entire body.I sincerely thank the members of my advisory board, Dr. Paul Kingston, Dr. Mark Whitehead, and Dr. David Rapaport for their aide in the successful completion of this project.。

血字的研究(英文版)

19211王冠宝石案theadventureofthemazarinstone19222雷神桥之谜theadventureofthorbridge19233爬行人theadventureofthecreepingman19241吸血鬼theadventureofthesussexvampire19251三个同姓人theadventureofthethreegarridebs19252显贵的主顾theadventureoftheillustriousclient19261三角墙山庄theadventureofthethreegables192611皮肤变白的军人theadventureoftheblanchedsoldier192612狮鬃毛theadventureofthelionsmane19271退休的颜料商theadventureoftheretiredcolourman19272带面纱的房客theadventureoftheveiledlodger19274肖斯科姆别墅theadventureofschoscombeoldhisotherworksconfessionofjohnsmith约翰史密斯的告白1883greatboerwar伟大的布尔战争1900thelostworld失落的世界1912thenewrevelations新启示1918thematerialfact重要信息1919thehistoryofspiritualism唯灵论史地球病叫一声科幻marlakurtabyss马拉库特深渊科幻contenttheintroductionoftheauthertheintroductionofcharactersthepoltofthestorydetectivegamesdoylesaidthatthecharacterofsherlockholmeswasinspiredbydr

纹理物体缺陷的视觉检测算法研究--优秀毕业论文

摘要
在竞争激烈的工业自动化生产过程中，机器视觉对产品质量的把关起着举足轻重的作用，机器视觉在缺陷检测技术方面的应用也逐渐普遍起来。与常规的检测技术相比，自动化的视觉检测系统更加经济、快捷、高效与安全。纹理物体在工业生产中广泛存在，像用于半导体装配和封装底板和发光二极管，现代化电子系统中的印制电路板，以及纺织行业中的布匹和织物等都可认为是含有纹理特征的物体。本论文主要致力于纹理物体的缺陷检测技术研究，为纹理物体的自动化检测提供高效而可靠的检测算法。纹理是描述图像内容的重要特征，纹理分析也已经被成功的应用与纹理分割和纹理分类当中。本研究提出了一种基于纹理分析技术和参考比较方式的缺陷检测算法。这种算法能容忍物体变形引起的图像配准误差，对纹理的影响也具有鲁棒性。本算法旨在为检测出的缺陷区域提供丰富而重要的物理意义，如缺陷区域的大小、形状、亮度对比度及空间分布等。同时，在参考图像可行的情况下，本算法可用于同质纹理物体和非同质纹理物体的检测，对非纹理物体的检测也可取得不错的效果。在整个检测过程中，我们采用了可调控金字塔的纹理分析和重构技术。与传统的小波纹理分析技术不同，我们在小波域中加入处理物体变形和纹理影响的容忍度控制算法，来实现容忍物体变形和对纹理影响鲁棒的目的。最后可调控金字塔的重构保证了缺陷区域物理意义恢复的准确性。实验阶段，我们检测了一系列具有实际应用价值的图像。实验结果表明本文提出的纹理物体缺陷检测算法具有高效性和易于实现性。关键字: 缺陷检测；纹理；物体变形；可调控金字塔；重构
Keywords: defect detection, texture, object distortion, steerable pyramid, reconstruction
II

人脸识别外文文献

Method of Face Recognition Based on Red-BlackWavelet Transform and PCAYuqing He, Huan He， and Hongying YangDepartment of Opto-Electronic Engineering，Beijing Institute of Technology, Beijing， P.R. China, 10008120701170@。

cnAbstract。

With the development of the man—machine interface and the recogni—tion technology， face recognition has became one of the most important research aspects in the biological features recognition domain. Nowadays, PCA（Principal Components Analysis） has applied in recognition based on many face database and achieved good results. However， PCA has its limitations: the large volume of computing and the low distinction ability。

In view of these limitations, this paper puts forward a face recognition method based on red—black wavelet transform and PCA. The improved histogram equalization is used to realize image pre-processing in order to compensate the illumination. Then, appling the red—black wavelet sub—band which contains the information of the original image to extract the feature and do matching。

我知道皮肤的作文英语

我知道皮肤的作文英语Title: Understanding the Skin: An Insightful Essay。

The skin, the largest organ of the human body, is often underestimated in its complexity and significance. Beyond its superficial appearance, the skin plays multifaceted roles in maintaining homeostasis, protecting against pathogens, and communicating sensory information to the brain. In this essay, we delve into the intricate structure and functions of the skin, shedding light on its remarkable capabilities.To comprehend the skin's complexity, one must first understand its anatomy. The skin comprises three primary layers: the epidermis, dermis, and hypodermis (subcutaneous tissue). The epidermis, the outermost layer, acts as a barrier against environmental aggressors, preventing water loss and shielding the body from harmful UV radiation. It consists mainly of keratinocytes, which undergo constant renewal to maintain skin integrity. Beneath the epidermislies the dermis, a dense connective tissue rich in blood vessels, nerves, and appendages such as hair follicles and sweat glands. The dermis provides structural support, nourishment, and sensory feedback to the skin. Finally, the hypodermis serves as a cushioning layer, insulating the body and storing energy in the form of adipose tissue.Beyond its structural components, the skin hosts a diverse ecosystem of microorganisms collectively known as the skin microbiota. These commensal bacteria, fungi, and viruses play a crucial role in maintaining skin health by competing with harmful pathogens, modulating immune responses, and contributing to skin pH regulation. Disruptions in the skin microbiota can lead to various dermatological conditions, emphasizing the importance of microbial balance for skin homeostasis.Functionally, the skin serves as a dynamic interface between the body and its environment, fulfilling essential roles in thermoregulation, sensation, and immunity. Through specialized structures like sweat glands and blood vessels, the skin regulates body temperature by dissipating heatthrough evaporation or conserving warmth through vasoconstriction. Moreover, sensory receptors scattered throughout the skin enable the perception of tactile sensations, pain, pressure, and temperature, facilitating interactions with the surrounding world.One of the skin's most crucial functions is its role in immune defense. The skin's innate immune system comprises various cellular components, including dendritic cells, macrophages, and mast cells, which detect and eliminate invading pathogens. Additionally, specialized immune cells such as Langerhans cells patrol the epidermis, capturing antigens and initiating immune responses. Furthermore, the skin houses adaptive immune cells, such as T cells and B cells, which provide long-term protection against recurring threats. This intricate immune network underscores theskin's pivotal role as a frontline defense barrier against infections.Beyond its physiological functions, the skin holds profound cultural and social significance across different societies. Throughout history, skin color has been used asa marker of identity, ethnicity, and social status, leading to discriminatory practices and social inequalities. However, contemporary perspectives emphasize the beauty and diversity of all skin types, promoting inclusivity and acceptance.In conclusion, the skin represents a marvel of biological engineering, with its intricate structure and multifaceted functions essential for human health and well-being. From its role as a protective barrier to its involvement in sensory perception and immune defense, the skin exemplifies nature's ingenuity. By understanding and appreciating the complexities of the skin, we can cultivate a deeper respect for our bodies and promote holistic approaches to skin care and health maintenance.。

An Analysis of the Body Images in Oliver Twist

2018年45期总第433期ENGLISH ON CAMPUSAn Analysis of the Body Images in Oliver Twist 文/李幸【Abstract】Dickens depicts some body images in Oliver Twist which reveal the relationship between power and body. This essay attempts to discuss three kinds of body images: the disciplined body, the abused body and the resistance of body. According to the discussion, it discloses that people are under the control of power. They are disciplined by social systems and those who possess power in many ways. Being disciplined, they become the docile bodies. The disciplined bodies still try their best to get freedom by resisting the strong power. Although the power of their resistance is so weak, it also reflects their desire to challenge the evil and unjust social systems and are eager to pursue freedom.【Key words】body images; discipline; power; Oliver Twist【作者简介】李幸(1989-)，女，布依族，贵州都匀人，黔南民族师范学院，讲师，研究方向：英美文学。

基于SystemView仿真下的眼图分析

1基带传输的基本原理在实际的通信系统中，很难完全消除码之间的串扰。

这主要是由于传输过程中传输系统的信号不稳定所致，使得波形存在变形、展宽，而且之前波形会出现长的拖尾现象，到观察码元的抽样时间点上，识别器会对结果出现错误判决。

对误码率的影响现在还没有找到数学上能处理的统计规律，还无法在这方面进行针对性的计算。

码间串扰如图1所示。

为了在实验室中测量基带传输系统的性能，使用示波器观察接收信号的常用方法是将示波器连接到接收滤波器的输出端，然后调整示波器的水平扫描周期以匹配示波器的水平扫描周期，与接收的符号周期和持续效果同步。

用于扫描的示波器的波形重叠，并且示波器屏幕上显示的结果看起来像人眼，这就是将其称为“眼图”的原因。

分析码间串扰和噪声对系统性能的影响，这就是眼图分析法。

基带信号与眼图如图2所示。

眼图是通过在示波器上叠加特定的数字信号而显示的图，它包含很多信息。

噪声的影响以及符号之间的对话在眼图中可见。

这些效果反映了数字信号的一般属性，因此，评估了整个系统的优缺点。

所以眼图分析是高速互连系统信号完整性分析的核心。

此外，眼图还可以用于调整接收滤波器的属性，以减少符号之间的串扰效应，并改善整个通信系统的传输性能。

眼图张开的大小反映了码间串扰的强弱。

眼睛越大，眼图越正确，符号之间的距离越小，反之符号之间的距离越大。

如果发生噪声，则噪声会叠加在信号上，眼图的轨迹会变得模糊。

如果有拦截码，“眼睛”或多或少会张开。

与代码之间没有交集相比，原始的细轨道明显平滑且变成模糊的条纹，并且标准化程度不高。

噪声越大，轨道越宽、越深，代码之间的交点越大，眼图的校正越少。

眼睛中的图像包含大量有价值的信息，可用于衡量数字信号传输系统性能的好坏，它可以显示接收过滤器的设置，以减少代码之间的间隔。

眼图如图3所示。

①最佳采样时刻是眼睛张开最大的时间。

②对时间误差的敏感度可以通过图表的斜边斜率来确定。

③在采样期间，【作者简介】赵瑄（1983-），女，吉林磐石人，实验师，从事通信工程、电子信息工程研究。

英语翻译实例

Contents: 1. Reflection on two translation jokes 2.Apprehension of foreignization domestication 3. Influences on languages 4. Foreignization in translation 5. Domestication in translation 6. Conclusion
Reference books: 1. 连淑能. 英译汉教程. 高等教育出版社， 2006，13-15 2. 孙致礼 . 新编英汉翻译教程 . 上外教： 2004(32-47) 3. 毛荣贵. 新世纪大学英汉翻译教程. 上海交大：2003（193-201） 4. 唐玉柱. 从Milky Way的“误译”谈翻译实践中的归化和异化. 重庆交通学院学报 (社科版), 2005,5(4): 120-121
Quotations:
* “While literal translation and free translation mainly deal with linguistic reproduction, foreignizing translation and domesticating translation are concerned more with cultural, linguistic and aesthetic considerations.” (连淑能） * “归化与异化可看成是直译与意译概念的延伸，但并不完全等同于直译与意译。……如果说直译与意译是语言层次的讨论，那么，归化和异化则是将语言层次的讨论延续升格至文化、诗学和政治层面。” （王东风）
美国自封亚洲领导者缺少自谦美国支撑其在亚洲的影?salad?email?betray?cool?venus?jazz?vitamin?cocacola?media?broadway?gene?angel?marathon?poker?humor?calorie?安琪儿?马拉松?扑克?幽默?卡路里?calorie?carnation?卡路里?康乃馨?dink?ballet?carbine?jazz?tshirt?sauna?丁克族?芭蕾舞?卡宾枪?爵士乐?t恤衫?桑拿浴?sauna?aids?rally?桑拿浴?爱滋病?拉力赛transliterationcategoryword32influenceofforeignizationonenglish?纸老虎?保全面子?丢脸好久不见?papertiger?tokeepface?tolosefacelongtimenosee?好久不见?longtimenosee?alltheimperialistnationsarenothingbutpapertigers

Human Cytomegalovirus Infection Alters the Expression of Cellular

J OURNAL OF V IROLOGY,Sept.2008,p.9065–9074Vol.82,No.18 0022-538X/08/$08.00ϩ0doi:10.1128/JVI.00961-08Copyright©2008,American Society for Microbiology.All Rights Reserved.Human Cytomegalovirus Infection Alters the Expression of Cellular MicroRNA Species That Affect Its ReplicationᰔFu-Zhang Wang,1†Frank Weber,2Carlo Croce,3Chang-Gong Liu,3Xudong Liao,4and Philip E.Pellett1,5*Departments of Molecular Genetics1and Molecular Cardiology4and the Genomic Medicine Institute,2Lerner Research Institute, Cleveland Clinic,Cleveland,Ohio44195;Human Cancer Genetics Program,Comprehensive Cancer Center, Ohio State University,Columbus,Ohio432103;and Department of Immunology and Microbiology,Wayne State University School of Medicine,Detroit,Michigan482015Received8May2008/Accepted24June2008The human genome encodes over500microRNAs(miRNAs),small RNAs(19to26nucleotides[nt])thatregulate the expressions of diverse cellular genes.Many cellular processes are altered through a variety ofmechanisms by human cytomegalovirus(HCMV)infection.We asked whether HCMV infection leads tochanges in the expression of cellular miRNAs and whether HCMV-regulated miRNAs are important for HCMVreplication.Levels of most miRNAs did not change markedly during infection,but some were positively ornegatively regulated.Patterns of miRNA expression were linked to the time course of infection.Some similarlyreregulated miRNAs share identical or similar seed sequences,suggesting coordinated regulation of miRNAspecies that have shared targets.miRNAs miR-100and miR-101were chosen for further analyses based ontheir reproducible changes in expression after infection and on the basis of having predicted targets in the3؅untranslated regions(3؅-UTR)of genes encoding components of the mammalian target of rapamycin(mTOR)pathway,which is important during HCMV infection.Reporter genes that contain the3؅-UTR of mTOR(predicted targets for miR-100and miR-101)or raptor(a component of the mTOR pathway;predicted site formiR-100)were constructed.Mimics of miR-100and miR-101inhibited expression from the mTOR construct,while only miR-100inhibited the raptor construct.Together,miR-100and miR-101reduced mTOR proteinlevels.While the miR-100and miR-101mimics individually modestly inhibited production of infectiousprogeny,much greater inhibition was achieved with a combination of both(33-fold).Our keyﬁnding is thatHCMV selectively manipulates the expression of some cellular miRNAs to help its own replication.MicroRNAs(miRNAs)are small(ϳ21-nucleotide[nt]) RNA species that are expressed from specialized genes and have important roles in the regulation of cellular gene expres-sion,including the regulation of development,the differentia-tion of hematopoietic stem cells,apoptosis,and the develop-ment of cancer(reviewed in references16,25,and26). miRNA-mediated gene regulation is related to cellular de-fenses that are mediated via very similar mechanisms(small interfering RNA)but target exogenous mRNAs,such as those expressed by viruses.The human genome contains perhaps500 distinct miRNA genes(1,33).miRNAs are initially expressed as5Ј-capped and polyadenylated RNA polymerase II tran-scripts(4,31).They are expressed either as individually regu-lated genes or as clusters of miRNAs that are expressed and then processed from a single primary transcript that might contain several miRNAs(22,29,49).After cytoplasmic pro-cessing to theϳ21-nt single-stranded mature form by the en-zyme Dicer,miRNAs associate with the RNA-induced silenc-ing complex.miRNA-mediated RNA interference is manifest as reduced levels of translation from the targeted mRNAs.This translational silencing comes in two forms:(i)by inhibit-ing protein synthesis after binding via incomplete base pairing to the3Јuntranslated regions(3Ј-UTR)of target mRNAs,and (ii)by binding to mRNAs with perfect complementarity,which leads to cleavage of the targeted mRNA.One important con-sequence of miRNA-mediated inhibition of gene expression via imperfect base pairing is that individual miRNAs can po-tentially regulate many cellular targets.Further,individual genes can be targeted by multiple miRNAs.There are connections between viruses and the miRNA world.Simian virus40,human immunodeﬁciency virus type1 (HIV-1),herpes simplex virus type1,Marek’s disease virus, murine cytomegalovirus,human cytomegalovirus(HCMV), Epstein-Barr virus,and human herpesvirus8encode sets of miRNAs(3,12,18,40,42,44,48,53).HCMV encodes at least 12miRNAs that are expressed as immediate early or early viral genes in infected cells;their abundance increases for at least 72h after infection(12,18,41).Similarly,a herpes simplex virus type1viral miRNA encoded upstream of the latency-associated transcript also persisted at high levels until late in infection(9).Thus,the miRNA machinery remains opera-tional in cells infected by these viruses.While functions have yet not been ascribed to most of the virally encoded miRNAs, HCMV-miR-UL112inhibits NK cell killing by reducing major histocompatibility complex class I chain-related molecule B protein levels(46),the simian virus40miRNA plays a role in preventing the immune recognition of infected cells by reduc-ing the production of some viral proteins(47),several human*Corresponding author.Mailing address:Department of Immunology and Microbiology,Wayne State University School of Medicine,540East Canﬁeld Avenue,6225Scott Hall,Detroit,MI48201.Phone:(313)577-6494.Fax:(313)577-1155.E-mail:ppellett@.†Present address:Lineberger Comprehensive Cancer Center,Uni-versity of North Carolina at Chapel Hill,Chapel Hill,NC27599.ᰔPublished ahead of print on2July2008.9065herpesvirus 8-encoded viral miRNAs target cellular genes that may be relevant to pathogenesis (45),and an HIV-1-encoded viral miRNA suppresses nef gene expression (40).In addition to encoding miRNAs,viruses can interact with host miRNAs or affect their regulation.Nuclear export of miRNA precursors is inhibited by the adenovirus VA1non-coding RNA (35).Accumulation of primate foamy virus type 1is inhibited by a cellular miRNA (miR-32),and the virus en-codes a protein that suppresses miRNA silencing in mamma-lian and plant cells (30).In addition,a cellular miRNA (miR-122)that targets the 5Јend of the hepatitis C virus genome facilitates viral replication (24).Transfection of HeLa cells with HIV-1leads to a substantial alteration in the expression of cellular miRNAs,with many of them being downregulated (54).Moreover,HIV-1infection downregulates cellular miRNAs that repress viral growth (50).HCMV exerts diverse and profound effects on the regulation of host cell metabolism,including altering levels of cellular transcripts,perturbing the cell cycle,and inhibiting infection-triggered apoptosis (reviewed in reference 39),and the ma-chinery for miRNA biosynthesis and activity is functional in HCMV-infected cells.Thus,we asked whether HCMV infec-tion results in altered expression of cellular miRNA species.We found that levels of most miRNA species do not change dramatically during infection,but several species were mark-edly up-or downregulated,indicating that HCMV infection leads to speciﬁc changes in the expression of some cellular miRNAs.Further,we identiﬁed cellular targets for two of the virally regulated miRNAs and found that synthetic mimics of these miRNAs can inhibit viral replication.MATERIALS AND METHODSCells and virus.MRC-5,HeLa,and 293T cells were obtained from the Amer-ican Type Culture Collection (Manassas,VA)and maintained in Earle’s modi-ﬁed Eagle’s medium containing 10%fetal bovine serum (HyClone,Logan,UT),2mM L -glutamine,0.1mM nonessential amino acids,and 1mM sodium pyru-vate.T-BACwt (a gift from Hua Zhu)is a derivative of the HCMV Towne strain that has several nonessential genes (US1through half of US12)replaced with a gene encoding a green ﬂuorescence protein and was used in all experiments (38).Virus stocks were prepared in MRC-5cells infected with T-BACwt at a multi-plicity of infection (MOI)of 0.01,fresh culture medium was added when the cells showed Ͼ90%cytopathic effects,and the cells were harvested after 3to 4days.Harvested cells were resuspended in a thrice-autoclaved mixture of 50%fat-free milk and 50%culture medium and then frozen in aliquots at Ϫ80°C.Viral titers were determined by a plaque-forming assay done in triplicate in 48-well ﬂat-bottom plates (Corning Incorporated,Corning,NY).RNA preparation.MRC-5cells were infected at an MOI of 2in 100-mm cell culture dishes (Becton Dickinson,Franklin Lakes,NJ).Total RNA was extracted at 6,24,48,and 96,or 120h postinfection (hpi)with Trizol reagent (Invitrogen,Carlsbad,CA).Brieﬂy,the culture medium was removed and the cells were lysed with 6ml of Trizol reagent and then frozen at Ϫ80°C.The RNA phase was partitioned with 6ml of chloroform,precipitated with 3ml of isopropyl alcohol,washed,and resuspended in DNase/RNase-free sterile water and then stored at Ϫ80°C.RNA quality was veriﬁed by electrophoresis in agarose gels.miRNA microarray.miRNA microarray analysis was performed as described previously (34),using an updated version of the chip (human and mouse miRNA 11K version 2chip).Arrays included 40-mer oligomers with sequences corre-sponding to 250human miRNAs and their precursors known at the time this work was initiated.The chips also included a series of well-tested control probes with sequences corresponding to human tRNA and GAPDH (glyceraldehyde-3-phosphate dehydrogenase)that served as positive controls and numerous ran-dom sequences that served as negative controls.Each probe was printed in quadruplicate on activated Amersham CodeLink slides (Amersham,Piscataway,NJ).Five micrograms of total RNA was labeled with biotin by reverse transcrip-tion (RT),and the labeled cDNA was hybridized to the printed chips.Bound sequences were detected with a streptavidin-Alexa647conjugate.After process-ing,data were collected using an Axon 4000B scanner and the Genepix Pro 6.0software package (Molecular Devices,Sunnyvale,CA).miRNA data analysis.Microarray data were analyzed with BRB-ArrayTools (version 3.4_Beta_1a,National Cancer Institute,Rockville,MD).In the initial data ﬁltering,spots were excluded if the minimum ﬂuorescence intensity was Յ10or if there was insufﬁcient agreement among the quadruplicate spots.The mean ﬂuorescence intensities (MFI)of the quadruplicate spots were log 2trans-formed and then normalized using the median-centering array procedure.miRNAs were excluded if one of the expression values was less than 20%or had a 1.5-fold change in either direction from the averaged median value (34).For each miRNA for which data of sufﬁcient quality were available,array intensities from each time point after HCMV infection were compared with those of mock-infected MRC-5cells.qRT-PCR.Short RNAs (Յ200nt)were separated from total RNA prepara-tions (described above)by use of the mirVanaTM miRNA isolation kit (Ambion,Austin,TX)according to the manufacturer’s directions,beginning with 50␮g of total RNA that had been treated with 10units of DNase at 37°C for 30min.Levels of miRNA expression at different times after HCMV infection were determined by quantitative RT-PCR (qRT-PCR)from 10-ng aliquots of short RNA by using the mirVanaTM qRT-PCR miRNA detection kit (Ambion,Austin,TX)and primers for speciﬁc miRNAs (Applied Biosystems,Foster City,CA).miRNA target veriﬁcation.Fragments containing the 3Ј-UTR of mTOR and raptor that contain predicted targets of miR-100and/or miR-101were cloned from MRC-5cells by use of the following primers,each of which contains a NotI site (italicized):5Ј-GCGGCCGC AGATGTGCCCATCACGTT-3Јand 5Ј-GCG GCCGC TGATGTCATTTATTGGCACA-3Ј(mammalian target of rapamycin [mTOR],NM_004958),and 5Ј-GCGGCCGC CCTGCTACTCGCTTTTGTC-3Јand 5Ј-GCGGCCGC TTTCCCGAATTTCCAGTGTC-3Ј(raptor,NM_020761).The mTOR primers amplify a 927-bp fragment located 11to 937bp downstream of the stop codon,and the raptor primers amplify a 406-bp fragment located 143to 557bp downstream of the stop codon.Cloned sequences were conﬁrmed and the fragments were transferred to the NotI site in the 3Јmultiple cloning se-quence of pHygEGFP (BD Biosciences,Palo Alto,CA),which is located down-stream of the stop codon of a gene encoding a fusion protein of hygromycin and enhanced green ﬂuorescent protein (EGFP)to produce pHygEGFP-mTOR and pHygEGFP-raptor.HeLa cells were cotransfected with reporter genes andmim-parison of results from two independent miRNA microarray experiments.RNA was prepared from MRC-5ﬁbroblasts mock infected or infected with HCMV T-BACwt (MOI of 2)at 6,24,and 48hpi plus 96hpi (experiment 1)or 120hpi (experiment 2).These RNAs were used as probes on miRNA microarrays,as described in Materials and Methods.Data that passed the quality control criteria were compared with the BRB-ArrayTools software.9066WANG ET AL.J.V IROL .ics of miR-100,miR-101,or a negative-control miRNA mimic (Dharmacon,Lafayette,CO)by use of Lipofectamine 2000(Invitrogen,Carlsbad,California)as described previously (10).At 15to 24h after transfection,digital pictures of the cells were taken with a Leica (DM IRB)UV microscope system with a Q-Imaging camera at 100ϫmagniﬁcation,and EGFP expression levels were determined using Image-Pro Plus software (version 6.1).HCMV growth in the presence of miR-100and miR-101mimics.MRC-5cells were grown in 48-well plates (Becton Dickinson,Franklin Lakes,NJ)to 40to 50%conﬂuence and transfected using Lipofectamine 2000with miR-100,miR-101,or a negative-control miRNA mimic.Two days after transfection,the cells were serum starved for 48h and then infected with T-BACwt (MOI of 2).Four days after infection,virus titers in culture supernatants were determined by standard serial dilution on MRC-5cells.Immunoblot analyses.Cells from six-well plates were collected and lysed on ice in lysis buffer (50mM Tris-HCl [pH 8.0],150mM NaCl,1%Triton X-100[vol/vol],10%glycerol [vol/vol],and 1%protease inhibitor cocktail).After clar-iﬁcation by centrifugation,100␮g of protein lysate was separated in a gradient (4to 20%)polyacrylamide-sodium dodecyl sulfate gel and then transferred to a membrane (Immobilon-P;Millipore Corp.,Billerica,MA).The membrane was preincubated in blocking buffer containing 5%fat-free milk in Tris-buffered saline containing 0.1%Tween 20(pH 7.6).The membrane was probed with a rabbit anti-mTOR antibody (2972;Cell Signaling Technology,Inc.,Beverly,MA)overnight at 4°C and then with a horseradish peroxidase-conjugated anti-rabbit secondary antibody for 1h at room temperature with chemiluminescent detec-tion (Millipore Corp.).Bands were scanned and then quantiﬁed using Image-Pro Plus software (version 6.1).Statistical analysis.The two-tailed Student’s t test was used to analyze the effects of miRNA mimics on reporter assays and viral replication.The samples were treated as having equal variances (43,52).RESULTSBased on the abundant observations that HCMV employs a variety of mechanisms to regulate cellular processes,we hy-pothesized that (i)cellular miRNAs play a signiﬁcant role in CMV biology and (ii)HCMV regulates cellular miRNA pop-ulations,collectively or individually.miRNA microarray analysis.(i)Experimental design.To gain an overview of the impact of HCMV infection on host cellular miRNA expression,MRC-5cells that had been con-ﬂuent for 3days were infected at an MOI of 2with HCMV strain T-BACwt.In this way,we excluded the effects of cell replication and ensured the synchronous infection of the met-abolically synchronized cells (2,55).The HCMV strain we used has an integrated GFP gene so that infection efﬁciency can be conveniently monitored,as infected cells show green ﬂuorescence from 48hpi.Two independent experiments were done in which total RNA samples were extracted from mock-infected cells and cells that had been infected for 6,24,and 48h plus a 96-h time point in experiment 1and a 120-h time point in experiment 2,thus enabling analysis of early and late stages of infection.The rationale for this design was that in previous global analyses of host cellular gene expression,the number of affected cellular genes increased with time after HCMV infection,and groups of genes associated with different cellular activities,such as innate immune responses andcellFIG.2.Impact of HCMV infection on the expression proﬁle of cellular miRNA.(A)Comparison of miRNA expression levels at dif-ferent times after HCMV infection with levels in mock-infected cells.(B)Cluster analyses and heat map showing the expression levels of probes representing miRNAs with high-quality data available for every time point.The heat map is based on center-normalized MFI for each miRNA probe.Black bars denote probes for which signal intensities were below the level of quantitation.V OL .82,2008HCMV AND CELLULAR miRNA EXPRESSION 9067cycle regulation,were affected at different times after infection (2,21).The miRNA microarray chips used in this study are well established and have been used in many studies of cancer and cellular differentiation(5,7,8,13,17,20,34,51,52).The array results passed the standard blank,negative,and positive qual-ity controls.(ii)Similar results in experiments1and2.Results of two independent experiments(experiment1and experiment2)at each time point were compared to evaluate their reproducibil-ity.For the scatterplots shown in Fig.1,the analysis was re-stricted to probes that met the quality control criteria;because of the variability in the data for spots with very lowﬂuores-cence signals,signals ofՅ64ﬂuorescence units are not repre-sented in the scatterplots.These species are considered to be negative for expression in our experiments;probes for153of the250miRNAs were negative at all time points in both experiments.There was generally good agreement between the two experiments.The correlation coefﬁcients(r)wereϾ0.9at all time points.The similarity between results obtained at the 96-hpi and120-hpi time points(rϭ0.93)is consistent with the cells being in the late phase of the replication cycle.Because of the high agreement between the two experiments,further anal-yses were based on experiment2.(iii)miRNA expression after HCMV infection.The scatter-plots in Fig.2A are comparisons of the expressions of individ-ual miRNAs at each time point with mock-infected cells.Two general observations are apparent from these plots:(i)there is no global unidirectional change in miRNA expression,and(ii) the expression levels of individual miRNAs may be unaffected by infection or may be positively or negatively regulated after infection.Table1shows the changes relative to levels for mock-infected cells for a set of miRNAs for which high-quality data(as deﬁned in Materials and Methods)were available from every time point.In the comparison with mock-infected cells at the6-hpi time point,the miRNA expression proﬁles are very similar;only one cellular miRNA was upregulated,but none were downregulated by more than twofold.At24hpi, three cellular miRNA species were upregulated by more than twofold and none were downregulated(one of the miRNAs was targeted by probes for two precursor forms;thus,it shows up in the plot as two points).Especially manifest at48hpi and 120hpi is a progressive change in the expression of individual miRNAs relative to what was seen for mock-infected cells, such that the differences in expression levels of most miRNAs are readily visible.In total,2and8miRNA species wereTABLE1.Human miRNA expression patterns afterHCMV infection amiRNA ID b Mock MFI(ϮSD)cRatio of infected cell MFI tomock MFI at d:6hpi24hpi48hpi120hpimiRNAs withՆ2-folddecrease-10p3,022Ϯ2490.8**0.7***0.4**0.6*-214,286Ϯ274 1.10.9*0.5***0.4** -29p2,366Ϯ128 1.20.8*0.5***0.5** -34p1,112Ϯ3710.7*0.5***0.4*** -99p7,208Ϯ80910.8**0.4***0.5** -1009,842Ϯ63210.70.4**0.1*** -101p2,558Ϯ17110.8*0.4***0.4*** -1051,412Ϯ135 1.10.7*0.4**0.4*** -125p9,903Ϯ173 1.2*0.80.6***0.5** -1331,127Ϯ73 1.10.9*0.6**0.5** -145p2,442Ϯ184 1.20.8*0.5**0.5** -146p873Ϯ56 1.10.90.5**0.5** -155p4,035Ϯ2520.9**0.8**0.4**0.4** -181p2,445Ϯ114 1.10.7**0.4**0.3*** -1815,051Ϯ42510.6**0.3**0.3** -181p1,943Ϯ49 1.10.80.5***0.4** -1921,844Ϯ21910.7***0.5**0.4*-194963Ϯ4610.7**0.5**0.4** -213p4,350Ϯ250 1.10.6***0.4***0.3** -221p24,951Ϯ98210.90.5***0.5** -222p31,697Ϯ5890.90.7*0.4***0.4*** -223p2,847Ϯ510.90.8*0.5***0.4*** -3201,970Ϯ145 1.20.8**0.5**0.4** -3213,449Ϯ259 1.10.9*0.6**0.4** -3351,170Ϯ7010.6**0.5**0.5** miRNAs with nochanges ofϾ2-fold(up ordown)-let-7p944Ϯ53 1.9* 1.10.8*1-16856Ϯ69 1.110.9 1.4-32p927Ϯ54 1.11 1.1 1.4*** -34311Ϯ15 1.2 1.10.9*0.8*-93p955Ϯ431 1.2* 1.1 1.9** -148331Ϯ11010.90.8 1.1-191p2,137Ϯ1191 1.11 1.7*** -1961,481Ϯ272 1.5*0.80.70.7-206p1,146Ϯ91 1.8* 1.3**11-210p819Ϯ48 1.10.90.7*0.6** -212p1,537Ϯ48 1.9* 1.7 1.3 1.8-215p271Ϯ30 1.4*** 1.4* 1.6* 1.5*-220p1,896Ϯ113 1.2*10.6**0.6** -324596Ϯ58 1.30.9*0.7*0.6** miRNAs withՆ2-foldincrease-1p302Ϯ11 1.5* 2.1* 1.4* 1.1-1227Ϯ5 1.3 2.7*** 1.6 1.2-17p1,008Ϯ581 1.2* 1.3 2.3*** -20p792Ϯ781 1.1 1.2* 2.4** -33p453Ϯ24 2.2* 1.20.91-92p13,751Ϯ6160.9* 1.1 1.12**-95p1,745Ϯ1361 1.2 1.22**-106p807Ϯ56 1.1 1.2* 1.2 2.1** -132p404Ϯ38 1.4* 2.7**2** 1.7*-197p1,287Ϯ167 1.2 1.1 1.12*-2195,463Ϯ689 1.8** 1.5* 1.9*2*-203p196Ϯ15 1.1 1.7* 2.3** 3.5*** a Array data for human miRNAs with high-quality data(as deﬁned in Materials and Methods)at every time point.Data from a representative probe are shown if similar results were obtained with multiple probes for the same miRNA species.b miRNA names are for human miRNAs as listed in the Sanger miRNA registry.The sufﬁx“p”denotes species for which the array probe targeted the precursor form of the miRNA.c MFI and SD of mock-infected cells are shown to give baselines for inter-preting the change levels.d At each time point,the MFI for each probe was compared with the corre-sponding value from mock-infected cells with the two-tailed Student t test, considering the samples as paired.*,PՅ0.01;**,PՅ0.001;***,PՅ0.0001.parison of microarray and qRT-PCR results miRNA and test aFold change at b:6hpi24hpi48hpi120hpi miR-223Microarray0033 qRT-PCR0023 miR-101Microarray0034 qRT-PCR0088 miR-181Microarray0047 qRT-PCR0224 miR-100Microarray00334 qRT-PCR00416a Short RNAs(Յ200nt)isolated from the same RNA preparations used in microarray analysis were reverse transcribed and then analyzed by qRT-PCR.b The change is the ratio of threshold cycles(CT),C T mock to C T infected,for each time point.Results are shown from a subset of downregulated miRNAs.9068WANG ET AL.J.V IROL.upregulated by more than twofold at 48and 120hpi,respec-tively,and 22and 24miRNAs were downregulated by more than twofold at 48and 120hpi,respectively.In many instances,changes of 20%to 30%were statistically signiﬁcant at a P value of Ͻ0.01,and all of the changes of at least twofold were signiﬁcant at that or higher levels (Table 1).qRT-PCR was used to conﬁrm the microarray data for selected miRNAs;as shown in Table 2,the two methods were in good agreement.To compare the expression patterns of individual miRNAs over the time course of the experiment,we did a cluster anal-ysis based on the MFI for 64probes for which high-quality data were available for every time point (Fig.2B).From the den-drogram and the associated heat maps,it is apparent that changes in miRNA expression levels link to the infection time course.The heat maps reveal coherent temporal linkages from time point to time point.Thus,the mock and 6-hpi time points are closely related to each other,as are the 48-and 120-h time points,with the 24-h time point being related to the 48-and 120-h time points.From the heat maps,persistent and tran-sient effects on the expression levels of individual miRNAs are apparent.The expression levels of some species progressively changed from high to low over the experimental time course;conversely,the expression levels of some species changed from low to high.The expression levels of some species transiently increased,especially at the 6-and 24-h time points.Speciﬁc examples of the types of changes mentioned above are shown in Fig.3.Cellular miRNAs were upregulated in three general patterns:persistent increases beginning from 6hpi (Fig.3A),transient increases in expression that peaked at 24hpi (Fig.3B),and late increases (Fig.3C).Some miRNAs were downregulated,with their levels decreasing until 48hpi with little change thereafter (Fig.3D).Interestingly,some miRNAs with the same seed sequence (residues 2to 8)had similar patterns of changes in expression after HCMV infection.This included miR-99and miR-100,which have identical seed sequences and have only a single-nucleotide difference elsewhere.In addition,of the nine miRNAs upregulated at 120hpi,four (miR-17,-20,-93,and -106)have identical seed sequences and only 2-to 3-nt se-quence differences elsewhere (11).These miRNAs are en-coded on different chromosomes.Thus,some groups of unique miRNA species that have similar targets are coordinately reg-ulated after infection.(iv)Selection of miRNAs for functional analyses.After the initial data quality ﬁltering described above,we applied a series of additional ﬁlters to identify the miRNA species whose ex-pression levels were most profoundly and reproducibly changed after HCMV infection.This involved identifying miRNAs for which there was at least a twofold change in the expression level (up or down),MFI of Ն1,024ﬂuorescence units at least one time point,and concordance between exper-iments 1and 2.Species that passed these ﬁlters are miR-21,-99,-100,-101,-155,-181,-213,-222,-223,and -320(down-regulated)and miR-17,-20,-106,and -219(upregulated).Interestingly,many of these species have been identiﬁed by others as playing possible roles in cell differentiation or onco-genesis (reviewed in references 16,25,and 26),and some are known to target cellular regulators that may be important during infection.miR-100and miR-101interact with components of the mTOR pathway.For the miRNA species that had the most consistent and signiﬁcant changes in expression following HCMV infection,we examined lists of targets predicted by MIRANDA and TargetScan (19,32).miR-100and miR-101have predicted targets on the mTOR pathway,which is impor-tant in regulating the translation of capped mRNAs,cell size,cell growth,cell cycle,cell survival,and cytoskeletal organiza-tion (37)(Fig.4A).Speciﬁcally,miR-100and miR-101each have a predicted target in the mTOR 3Ј-UTR.In addition,miR-100has a predicted target in the 3Ј-UTR of raptor,whichFIG.3.Patterns of changes in cellular miRNA expression after HCMV infection.Cellular miRNAs were upregulated in three general patterns:gradual upregulation over the time course of infection (A),transient upregulation that peaked at 24hpi (B),and late increases (C).There was only one pattern for downregulated cellular miRNAs (D).V OL .82,2008HCMV AND CELLULAR miRNA EXPRESSION 9069is a partner of mTOR,and miR-101has two predicted targets in the 3Ј-UTR of another mTOR partner,rictor (Fig.4B).To determine whether miR-100and miR-101interact with components of the mTOR pathway,we constructed reporter genes containing segments of the 3Ј-UTR from the mTOR and raptor genes downstream of the EGFP reading frame (pHygEGFP-mTOR and pHygEGFP-raptor)(Fig.5A and 6A).Cells transfected with the reporter genes express EGFP,and the effects of miRNAs on EGFP expression can be as-sessed by comparing ﬂuorescence intensities in the presence or absence of synthetic miRNA mimics or a negative-control miRNA mimic (Fig.5B).The miRNA mimics are able to be processed by Dicer and produce stable and active silencing of their cognate targets.For pHygEGFP-mTOR,the cloned region contains pre-dicted targets for miR-100and miR-101(Fig.5A).At 25nM,the miR-100mimic alone inhibited EGFP expression by 35%,while the miR-101mimic alone had no effect.At 50nM,the miR-100mimic alone inhibited EGFP expression by 60%,and the miR-101mimic alone inhibited it by 38%;the combination of 25nM of each of the miR-100and miR-101mimics (50nM total)inhibited EGFP by 48%(Fig.5C).The raptor 3Ј-UTR fragment contains a predicted target for miR-100but has none for miR-101.Consistent with this,in experiments performed in parallel to those for which results are shown in Fig.5B and C,miR-100inhibited pHygEGFP-raptor by about 50%,but nei-ther miR-101nor the negative-control miRNA mimic had an effect (Fig.6B).In addition,the miR-100and miR-101mimics had no effect on the parental EGFP vector,which lacks the mTOR and raptor 3ЈUTR (Fig.6B).We extended this to an analysis of the effects of miR-100and miR-101on the expres-sion of the mTOR protein in 293T cells.As shown in Fig.5D,a combination of the two miRNAs (6.25nM of each)reduced mTOR expression by 41%relative to what was seen for the transfection control,an effect greater than that seen for either of the mimics alone at 12.5nM;the control miRNA had a negligible effect.Note that the levels of miRNA mimics re-quired to achieve silencing of the EGFP reporter constructs (Fig.5A to C)and mTOR protein expressed at its native levels (Fig.5D)are not directly comparable,due to the high level of reporter gene transcripts expressed from the HCMV immedi-ate early promoter in the transfected HeLa cells and the rel-atively low level of mTOR expressed from its native transcript in 293T cells and possibly to differences in silencing resulting from differences in the overall structure of the mTOR 3ЈUTR reporter transcript versus that of the native mTOR transcript.To summarize,two of the miRNAs reregulated after HCMV infection interact speciﬁcally with the 3Ј-UTR of at least two members of the mTOR pathway and can reduce the expression of mTOR at the protein level.miR-100and miR-101inhibit HCMV replication.We tested the hypothesis that if levels of a cellular miRNA are reduced following HCMV infection,their absence might be helpful for viral replication.Therefore,by supplementing intracellular lev-els of the miRNA by exogenous addition of the miRNA,viral replication might be suppressed.First,we directly compared miR-100and miR-101.Based on work from Kudchodkar and colleagues,subconﬂuent human ﬁbroblasts (28)were transfected with the mimics of miR-100and miR-101individually and as equimolar mixtures (10).Two days after transfection,the cells were serum starved for two more days and then infected with the T-BACwt virus at an MOI of 2.Virus titers in the culture supernatant on the fourth day after infection are shown in Fig.7A.Individual mimics reduced virus titers by about one-half of a log at concentrations from 25to 100nM.Equimolar mixes of the two mimics were much more potent (Ͼ10-fold reduction)than individual mim-ics at the same total molar concentration.Inhibition by the 25nM mixture (12.5nM of each species)was Ͼ4-fold greater than that in the presence of 100nM of either species alone.These inhibitory effects were veriﬁed and extended in an in-dependent experiment that included a commercial miRNA mimic as a negative control (Fig.7B).In this experiment,miR-100and miR-101,either alone or in combination,reduced the amount of infectious HCMV in the culture supernatants at 4days after infection to levels similar to those shown inFig.FIG.4.Predicted targets of miR-100and miR-101in the 3Ј-UTR of members of the mTOR pathway.(A)Schematic of the mTOR signaling pathway.mTOR functions by partnering with either raptor or rictor and controls mRNA translation or cell growth and survival,respectively.Rapamycin inhibits the function of the mTOR/raptor complex,while the function of Akt,the only known target of the mTOR/rictor complex,can be inhibited by caffeine.PI3K,phosphati-dylinositol 3-kinase.(B)Predicted targets.For each duplex,the upper sequence is from the predicted target and the lower sequence is the mature form of the indicated miRNA.For mTOR at least,the pre-dicted target sequences are highly conserved among diverse mamma-lian species.9070WANG ET AL.J.V IROL .。

实时人脸特征提取(外文翻译)

(4) 外文翻译译文外文标题:实时人脸特征提取作者：赵杰煜,刘箴出处：中国科学杂志（外文版）.2008年4月刊E 辑.信息科学收稿日期：2007-07-10;接受日期：2008-02-26摘要快速精确的人脸特征提取是人脸识别和表情分析的基础.文中提出了一种新型高效的视频人脸几何特征实时提取方法.视频输入图像以加权图形式表示，通过在加权图上的随机游动实现人脸像素级特征的自动提取，脸部特征包括外轮廓、眉毛、眼睛、鼻子和嘴唇.加权图釆用8-邻接结构，定义在图的边上的加权值反映随机游动通过该边的似然度.随机游动模拟了一个各向异性的扩散过程，此扩散过程在滤除图像噪声点的同时保留下脸部特征点.随机游动从一些事先通过颜色和运动信息确定的、最具人脸特征的种子点开始，通过随机游动获得的人脸特征点以其原始形式统一保存在多个链表结构中，并根据人脸各部分的相对位置聚集成对应的特征点集合.有关人脸结构的先验知识通过Bayes 方法结合到分析过程中.为了便于高层视觉计算, 釆用统计形状分析方法，将人脸特征点进一步表示成形状和配准信息, 形状是具有仿射不变特性的几何信息，用于描述人脸的全局特征.形状的距离度量釆用Procrustes 距离.实验结果表明，提出的方法快速高效，能够实时地从视频中提取出人脸特征，在一定程度的光线变化、尺度变化、头部转动、手部干扰的情形下仍可以正常工作. 人脸信息处理是一个富有挑战性的研究领域，是多学科交叉的研究热点，由于技术的进步和市场的需求，近年来引起了学术界和工业界的广泛关注^.人脸特征提取作为人脸信息处理最为重要的一步，对于后续的基于视觉的人机交互至关重要，脸部表情分析、人脸识别、生物认证、动画制作、视频会议等都依赖于高效精确的人脸特征提取.人脸特征提取过程中的细小误差很容易导致身份验证或表情分析的错误.然而，由对于视频人脸特征的分割和提取，采用求图模型最优解的方法由于计算量大做到实时运算目前尚无法实现.因此，我们采用了有限的随机游动来近似地实现一个各向异性的扩散过程该过程在滤除图像噪声点的同时保留人脸特征点，从而获得精确的人脸几何特征.本文的组织如下：第1节介绍图模型的表示并定义加权图上的随机游动；第2节给出采用随机游动实现人脸特征点分割的方法；第3节介绍统计形状分析方法；第4节给出实验结果, 第5节为小结.(4)1图模型与随机游动加权图是数字图像非常自然的一种表示形式，图的顶点对应于像素，加权边用于表示像素间的关联度.本文采用8-邻接结构的无向加权图.边上的加权反映该边所具有的脸部特征的程度，越具有典型的脸部特征，加权值越大；越不具有脸部特征的，加权值越小.这里我们简要地给出图模型以及加权图上随机游动的正式定义.一个图G = (F,五）由可数顶点集F 和边集五组成，边ee 五为顶点对：e=〈x,少>=〈少，x>, x,_yeF.如果顶点x 和y 通过一条边相连，这种相邻关系就表示为—个加权图 Gw=(G,w),其中G 是一个图，w 是一个实函数，w:五(G)R 汧>0.定义在加权图上的随机游动是通过随机选择一条当前顶点的相连边连续不断地实现对一系列相邻顶点访问的随机过程.选择一条相连边的概率由该边的加权与所有相连边的加权之和的比率决定.对边e z e 五，设其上的加权为％=W (e z ).对顶点veF,设与v 相连的所有边的集合为#(v) = {ee £(G): e=〈兄v>,少e 「}, r(v)表示与v 相连的所有边上的加权之和，即妒(v) = I eieW(v)W;.则加权图G w = (G,w)上的随机游动X 是一个采用以下转移概率的随机过程，其中/v ~Xn 是连接v 和X n 的边的集合的指示函数. 2人脸特征点分割加权图上的随机游动可以被有效地用于图像滤波和图像分割.图像滤波过程由Perona 等人M 提出的各向异性扩散方程描述；图像分割过程则由Laplace 方程表示给定一些种子像素，算法通过随机游动来标记那些最易到达种子点的像素.大多数各向异性扩散滤波算法的目标是在不穿越边界的前提下平滑图像的同质区域，而图像分割的目标是标记出同质区域. 对于图像质量不高的视频人脸特征提取，我们需要同时达到2个目标：既要滤除图像噪声，又要分割出人脸特征点.假设当前图像由实函数描述向异性扩散可以表示为如下形式：为了在滤除噪声平滑图像的同时保留人脸结构信息，传导系数一般定义为空间位置相关项，最常见的选择如下：c ( x , y , t ) = exp (|VF (x ,y ,t )||2 ^ 2l 2(3) c ( x , y , t ) = exp (|VF (x ,y ,t )||2 ^ 2l 2其中2为常数.由此可见，对于图像的一致性较好的区域，c的取值较大，达到平滑的效果；对于变化较大的人脸特征区域,c的取值较小，从而达到保留人脸结构信息的目的.上述的各向异性扩散过程可以由一组随机游动来实现.设G w =(G,W)是对应于输入图像的加权图，采用8-邻接结构，为连接顶点/和y的边上的加权，随机游动采用自回避形式, 即随机游动不重复通过同一个点，单步转移概率如下,其中V。

skull1

skull1Skull1: Unveiling the Mysteries of the Human SkullIntroduction:The human skull is an intriguing and complex part of the human anatomy. Serving as a protective casing for the brain, it consists of multiple bones that are interconnected in a unique way. This document aims to explore the various aspects of the human skull, including its structure, functions, and significance in different fields such as medicine, anthropology, and forensic science.I. Overview of the Human Skull:The human skull is composed of two main parts: the cranium and the mandible. The cranium consists of several bones, including the frontal, parietal, temporal, and occipital bones, which protect the brain. The mandible, or lower jawbone, also plays a vital role in the functionality of the skull. Together, these bones create a remarkably robust structure that provides protection and support to the brain and facilitates various essential functions.II. Structure and Composition of the Skull:A. Cranial Bones:1. Frontal Bone: This bone forms the forehead and the superior part of the eye sockets.2. Parietal Bones: Located on the sides of the cranium, these bones play a crucial role in the protection of the brain's upper surface.3. Temporal Bones: Situated on the sides and base of the cranium, temporal bones protect the middle and inner ears.4. Occipital Bone: Positioned at the rear of the skull, the occipital bone shields and supports the back of the brain and contains the foramen magnum, through which the spinal cord passes.B. Facial Bones:1. Mandible: The largest and strongest facial bone, the mandible is responsible for the movement of the lower jaw.2. Maxilla: The maxilla forms the upper jaw and supports the upper teeth.3. Zygomatic Bones: Also known as cheekbones, zygomatic bones contribute to the facial structure.4. Nasal Bones: The nasal bones shape the bridge of the nose.5. Ethmoid and Sphenoid Bones: These bones are situated deep within the skull and support the nasal cavity and eye sockets.6. Lacrimal Bones: These small bones form part of the eye sockets.III. Functions of the Skull:A. Protection of the Brain: The primary purpose of the skull is to safeguard the brain from injuries caused by external forces. The thickness and rigidity of the cranial bones create a protective shield that absorbs and distributes impact forces.B. Facilitation of Sensory Functions: The skull plays a significant role in supporting the sensory organs. For instance, the eye sockets protect the delicate eyeballs, while the nasal cavity and sinuses assist in filtering and humidifying the air we breathe.C. Attachment for Muscles: Different muscles, such as those responsible for chewing and facial expressions, attach to various points on the skull, enabling their proper functioning.IV. Importance of the Skull in Medicine:A. Craniofacial Surgery: Understanding the intricacies of the skull's structure is vital for craniofacial surgeons. Extensive knowledge of the skull's anatomy allows them to performintricate procedures, such as corrective surgeries for congenital malformations or trauma-induced injuries.B. Neurology: Neurologists depend on a thorough understanding of the skull to diagnose and treat conditions affecting the brain. The skull's composition influences neurological assessments and the interpretation of imaging tests, such as CT scans and MRIs.V. Anthropological Significance of the Skull:A. Forensic Anthropology: Human skulls play a crucial role in forensic investigations. Forensic anthropologists can determine a variety of important information through the analysis of skulls, including age, sex, and even potential causes of death.B. Evolutionary Studies: Comparative analysis of different skulls, both within and across species, provides valuable insights into the evolution of bipedalism and the development of human cognition.Conclusion:The human skull is an extraordinary structure, simultaneously protecting the brain and allowing for essential functions. Itscomplex composition, functions, and significance in various fields make it a fascinating subject of study. From medical advancements to forensic investigations and evolutionary research, the skull continues to provide invaluable information in the quest for knowledge and understanding of human biology.。

人体肤色区域的自适应模型分割方法

人体肤色区域的自适应模型分割方法一、绪论1.1 研究背景和意义1.2 国内外研究现状1.3 研究内容与目的二、人体肤色分割方法综述2.1 颜色空间2.2 阈值分割2.3 基于图像区域的方法2.4 基于机器学习的方法三、人体肤色自适应模型的建立3.1 数据集的构建与预处理3.2 颜色空间的选择和特征提取3.3 模型的建立与训练四、基于自适应模型的人体肤色分割4.1 预测人体肤色区域的方法4.2 蒙版生成与优化4.3 分割结果优化五、实验与分析5.1 实验设计与数据集介绍5.2 实验结果与分析5.3 与其他方法对比及优势评价六、结论6.1 主要研究工作总结6.2 研究成果评价6.3 未来工作展望第一章：绪论1.1 研究背景和意义图像分割是计算机视觉领域中的一个重要问题，其目的是将图像中的前景与背景进行分离，通常用来提取感兴趣的目标物并在图像处理、计算机视觉等领域中得到广泛应用。

人体肤色区域分割是图像分割领域的一个重要研究方向，其在人脸识别、医疗图像处理、虚拟试衣等应用中有着重要的作用。

对于人体肤色区域分割，传统的方法主要采用颜色阈值法和基于机器学习的方法，如支持向量机（SVM）和神经网络（NN）等。

这些方法虽然在特定场景下表现良好，但是由于颜色分布的复杂性和肤色区域的多样性，仍然存在许多问题，如准确率低、灵敏度差、易受噪声干扰等等。

1.2 国内外研究现状国内外学者在人体肤色区域分割领域进行了广泛的研究。

国外学者一般采用颜色模型，如Lab、HSI、YUV等，结合传统的阈值分割方法或基于机器学习的方法来实现人体肤色区域分割。

此外，还有一些基于区域的方法，如基于超像素分割、基于边缘检测等方法。

同时，一些新的技术也出现在这一领域，如深度学习，包括卷积神经网络（CNN）、循环神经网络（RNN）等。

国内学者在颜色模型选择、特征提取及模型优化等方面取得了很大进展。

同时，诸如人体姿态捕捉、面部表情分析、界面增强等领域的需求，也促使人们考虑将多种信息整合，构建基于多种传感器的人体肤色区域分割方法。