数据采集与处理2011，Vol.26Issue(1)：15-19,5.
基于特征组合的人脸跟踪方法
Face Tracking Based on Feature Combination
摘要
Abstract
Aimed at the problem of tracking human face on complex background in color image sequences, this paper presents a robust method based on feature combination. The method uses skin color feature and motion feature to represent human face in color image sequences.Then, a set of likelihood ratio is generated for each pixel. the pixel indicates the confidence of its belonging to object or background. They are combined according to the classification performance. A particle filter algorithm is used to track the human face. The combination features are used to compute weights of particles and increase the stability on the varying lighting and complex backgrounds. An effective skin color detection algorithm is presented for varying lighting environment, and a dynamical model is used in particle filter tracking algorithm for robust face tracking. Experimental results show that the method is robust and effective compared with tracking algorithm with only one feature. The method is very hopeful for tracking free movement of the human face on complex backgrounds.关键词
人脸跟踪/特征组合/粒子滤波分类
信息技术与安全科学引用本文复制引用
夏思宇,潘泓,金立左,夏良正..基于特征组合的人脸跟踪方法[J].数据采集与处理,2011,26(1):15-19,5.基金项目
国家自然科学基金(60805002,90820009)资助项目 （60805002,90820009）
航空科学基金(20081069003)资助项目. （20081069003）
