基于肤色分割的改进Camshift人脸跟踪算法  

Improved Camshift face tracking algorithm based on skin color segmentation

作　　者：王俊南 赖惠成[1] 杨东旭 WANG Junnan;LAI Huicheng;YANG Dongxu(College of Information Science and Engineering,Xinjiang University,Urumqi 830046,China)

机构地区：[1]新疆大学信息科学与工程学院,新疆乌鲁木齐830046

出　　处：《现代电子技术》2019年第10期166-170,共5页Modern Electronics Technique

基　　金：国家自然科学基金资助项目(61561048);新疆维吾尔自治区科学基金资助项目(2015211C257)~~

摘　　要：针对Camshift算法在对人脸跟踪时,容易受到肤色背景的影响,且当跟踪环境中背景较为复杂时,易失败的问题,提出一种基于肤色分割的改进Camshift人脸跟踪算法。在YCbCr空间中进行肤色分割,提取每一帧的肤色区域,排除帧图像中非肤色背景的干扰;引入人脸几何特征,对潜在的人脸肤色区域进行约束;使用Camshift算法跟踪人脸。同时为了提高算法的性能,文中引入改进的Adaboost算法。实验结果表明,该算法对视频帧中的特定人脸可以获得出色的跟踪结果,具有实时性强、跟踪稳定的优点。In allusion to the problems that the Camshift algorithm is easy to be affected by the skin color background during face tracking,and the tracking is easy to fail if the tracking environment background is complex,an improved Camshift face tracking algorithm based on skin color segmentation is proposed.The skin color segmentation is performed in the YCbCr space to extract the skin color region of each frame,so as to eliminate the interference of the non-skin color background in the frame image.The face geometric features are introduced to constrain the potential face skin color region.The Camshift algorithm is used to perform face tracking in the end.The improved Adaboost algorithm is introduced to improve the performance of the algorithm. The experimental results show that the algorithm can obtain excellent tracking results of the specific face in the video frame,and has the advantages of strong real-time performance and stable tracking ability.

关 键 词：人脸跟踪 肤色分割 形态学操作 YCBCR空间 几何特征 Adaboost算法 CAMSHIFT 

分 类 号：TN820.4-34[电子电信—信息与通信工程] TP391.41[自动化与计算机技术—计算机应用技术]