一种结合肤色特征的人脸跟踪算法  

Face Tracking Algorithm Combined with Skin-color Characteristics

作　　者：杨德顺[1] 李丽娟[1] 

机构地区：[1]湖南大学计算机与通信学院,长沙410082

出　　处：《小型微型计算机系统》2012年第3期621-625,共5页Journal of Chinese Computer Systems

基　　金：国家"九七三"前期专项"数字媒体取证研究"基金项目(2010CB334706)资助

摘　　要：将CAMSHIFT算法用于人脸跟踪时,当跟踪目标被其它人脸遮挡时会发生跟踪失败,针对这一不足,提出一种结合肤色区域特征的连续自适应均值移动算法(SCCAMSHIFT):提取目标的肤色区域作为已知特征,通过跟踪窗口的变化来判断遮挡的发生,并在遮挡发生后利用已知特征来寻找丢失的目标.实验结果表明,SCCAMSHIFT算法能够较好的解决跟踪过程中类肤色物体的遮挡问题.The CAMSHIFT algorithm fails easily when the tracking face is occluded by other face or objects.To overcome this shortcoming,Skin-color area based Continuously Adaptive Mean Shift algorithm（SCCAMSHIFT） is proposed： the skin-color area of the target is extracted as the characteristic,the occurrence of occlusion is detected by the changes of the tracking window,and once the occlusion occurred,search for the missing targets by using the characteristic.The experimental results show that the SCCAMSHIFT algorithm can solve the occlusion problem of the skin-color like objects preferably.

关 键 词：人脸跟踪 CAMSHIFT算法 SCCAMSHIFT算法 肤色 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]