基于AAM的人脸图像描述与编码  

Description and Coding of Human Face Images Based on Active Appearance Models

作　　者：谢玉鹏[1] 吴海燕[2] 

机构地区：[1]黑龙江科技学院电气与信息工程学院,黑龙江哈尔滨150022 [2]国家开发银行,黑龙江哈尔滨150027

出　　处：《计算机仿真》2009年第6期272-276,共5页Computer Simulation

摘　　要：利用主动表观模型(AAM)来对人脸图像进行描述和编码,经过一定次数迭代,进行模型和人脸匹配,合成人脸图像。方法基于统计信息建模来实现对目标图像的描述。由于采用了优化算法,经过迭代运算使合成的模型与目标图像不断接近,最终能得到反应目标图像纹理及形状的合成模型。实验表明AAM方法进行人脸描述和编码的有效性。方法在人脸图像编码有重要的意义。Active Appearance Model （AAM） is used to describe and code human face images. After a certain number of iteration, matching the model with face images, the face images are synthesized. This statistical model reflects the facial shape and grey - level appearance. This method searches and explains the target image by the statistical information modeling. Because of using optimal algorithm, this algorithm makes synthetic model and target image more and more similar. At last, it obtains the synthetic model which represents both the shape and the texture of the target image. Experiment indicates that AAM algorithm is feasible for describing and coding human face images.

关 键 词：纹理 形状 描述和编码 

分 类 号：TP317.4[自动化与计算机技术—计算机软件与理论]