基于强跟踪Kalman滤波的鲁棒人脸跟踪算法  

Robust face tracking algorithm based on strong tracking Kalman filter

作　　者：钟锐[1,2] 吴怀宇[1] 吴若鸿[1] 

机构地区：[1]武汉科技大学信息科学与工程学院,湖北武汉430081 [2]赣南师范学院数学与计算机科学学院,江西赣州341000

出　　处：《计算机工程与设计》2016年第2期475-480,共6页Computer Engineering and Design

基　　金：国家自然科学基金项目(61075087;61203331)

摘　　要：进行人脸跟踪时,人脸出现快速移动、姿态大幅偏转及被遮挡的情况会导致人脸跟踪失败。针对以上问题,结合Haar-like特征与Adaboost算法对正面人脸进行跟踪,当人脸姿态出现大幅度偏转或被遮挡时,使用人脸肤色的直方图特征与Camshift算法对人脸进行跟踪。为提高人脸跟踪算法的运行效率,使用强跟踪Kalman滤波算法对人脸在下一帧图像中可能出现的位置进行预测,缩小人脸检测的搜索区域,提高人脸跟踪算法的实时性。实验验证了该算法在视频人脸跟踪方面具有较好的鲁棒性和实时性。A face tracking algorithm was presented to solve the problem of face tracking failure caused by fast face move,deflected face pose and occluded face.To solve the problem,Haar-like feature and Adaboost algorithm were combined to detect and track the frontal face.When face pose was greatly deflected or face was occluded,the face skin color histogram was taken as the detection feature and the Camshift algorithm was combined to detect and track the face.Strong tracking Kalman filter was used to predict the possible position of the face in the next frame,reducing the search region of face detection and improving the efficiency of the face tracking algorithm.Experimental results show that the proposed algorithm has good performance on the robustness and efficiency of face tracking in videos.

关 键 词：人脸跟踪 鲁棒性 HAAR-LIKE特征 ADABOOST算法 强跟踪Kalman滤波 CAMSHIFT算法 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]