双目视觉中的一种人脸立体匹配及视差计算方法  

A Method for Face Stereo Matching and Disparity Calculation in Binocular Vision System

作　　者：雷蕴奇[1] 宋晓冰[1] 袁美玲[1] 柳秀霞[1] 欧阳江帆[1] 

机构地区：[1]厦门大学信息科学与技术学院,福建厦门361005

出　　处：《厦门大学学报（自然科学版）》2009年第1期36-41,共6页Journal of Xiamen University：Natural Science

基　　金：国家863计划(2006AA01Z129)资助

摘　　要：研究并实现了一种基于双目视觉的人脸立体匹配计算方法.采用OpenCV技术结合肤色模型检测技术得到准确人脸区域,在实验对比了3种相似性度量算子优缺点的基础上,采用互相关性度量算法作为区域匹配方法.为了缩小搜索范围及提高匹配准确率,提出结合极线约束、人脸区域约束、人脸对称性等约束条件进行立体匹配,再利用左右一致性原则进行视差后处理.实验表明,本文给出的算法在时间效率上以及对应点的匹配率上有了较大的提高,效果较好,可为后续进行人脸三维重建提供良好的基础.A method for face stereo matching and disparity calculating in binocular vision system is proposed in this paper. Combined with the skin colour model, the paper first finds the face region by the technique of OpenCV, and then, based on the experiment results of comparing three kinds of similarity measure operators, we adopt the interrelated measure operator in regional matching algorithm. In the matching stage, to decrease the searching area and increase accuracy of stereo matching,we consider the epipolar line constraint,face region constraint and face symmetry condition in matching algorithm to calculate the disparity. Finally, we use the principle of left-right consistency to accomplish the post-processing of disparity. In our experiment results, the method proposed has a satisfied performance regarding calculating time and the matching accuracy for corresponding points.

关 键 词：人脸立体匹配 人脸检测 定位 双目视觉模型 相似性度量算子 区域匹配 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]