The unique heat radiation of infrared image is generated from subcomponents . this paper analyses the natural feature of infrared image and the relationship between the shape and orientation with eigen vectors , gives the measure to classify infrared image based on object position to improve ability of compatibility in object recognition , moreover , applies the independent component for sub - region segmentation to construct new infrared image eigen vectors . the image features are processed by a support vector machine , and transform the binary svm to multi - object classifier 文中在传统红外热图像特征量的基础之上，通过分析红外热图像自身特性以及它的形成与其自身姿态的关系，提出了对目标物体红外热图像基于目标姿态的子分类，来提高目标识别中对姿态的容忍度；同时根据红外目标子部件对图像影响的统计独立性，使用独立元的方法对目标红外热图像进行了子区域分割，形成了新的红外图像特征量。
This paper makes a good study of the face recognition , including testifying the face from an image , partitioning main area of the face , abstracting the edge of the image , locating the characteristic points , recognizing the same face , etc . i have many new ideas in the study as follows : a method partitioning facial area from an image by 24 _ bit color is described 本论文对人脸识别技术的算法作了全面的研究，包括对图像进行人脸的检测、区域分割、边缘提取、去除噪声和特征点的定位、特征向量的提取、识别等这几个部分。提出了如下几种实用和有效的算法：在对当前多种人脸检测方法的研究后，提出了一种利用肤色对面部图像进行分割的方法。
Relating to bionics , the image is transformed in order to segment and extract the region including char conveniently . considering the former methods of character exaction , we give prominence to the stroke location , run length and amount feature of character , and recognize the character based on it . the following are the major contents in the thesis 本文从仿人认知的角度着手，变换域提取图像特征达到字符区域分割目的；克服以往字符特征提取中同时将噪声、偏移等冗余信息考虑在内的缺点，以认知的角度突出最能反映字符的笔划数量、位置，游程等特征，有效的进行以认知分类为基础的字符识别。
Image segmentation techniques involving region segmentation and line feature extraction which are the foundation of our research are introduced . and the other concerning methods have been referred and exemplified . it has been well recognized that the expression and the application of knowledge play a key role in target recognition 知识驱动的目标识别方法的实现有很多种，其中图象分割中所涉及到的区域分割和线状特征提取技术是我们的研究的基础，在文章中对这两种方法中的所涉及的具体知识做了详细的介绍。
In this paper , the conception and classification of image segmentation technology are first introduced , and then , the key point , the advantages and disadvantages of different methods are proposed . with the models of video series , the motion - changed regions testing theory are introduced . after the discussion and analysis of regional growth , regional separatism and regional integration - separatism three major region - based segmentation algorithms , this paper suggests an approach for image segmentation by combining region based on tree - branches growing method 本文首先介绍了图像分割技术的概念、分类并指出了其中的关键技术及不同方法的优缺点,然后围绕视频序列的模型,介绍了分割预处理的运动变化区域检测的相关理论,接着重点研究了图像分割技术中基于区域的分割算法,在讨论和分析了区域生长、区域分裂、区域合并分裂三种主要区域分割技术的基础上,提出了基于树状结构的区域生长改进算法,实验结果证明了本文所提出算法的有效性和可行性,最后介绍了图像分割算法的评价体系并提出了算法的不足和改进方向。
In the first part of this paper , the principle of human face location is introduced : because human skin color clusters in the color space , we select a appropriate space to construct the skin color model , then skin color area of face is separated . finally , we use the construction feature of face to locate the face 本文首先详细介绍了人脸定位的原理：根据人的肤色在色度空间的聚类特性，选择合适的色彩空间建立肤色模型，利用相似度将人脸区域分割出来，再结合人脸的结构特点进行人脸定位。
