Improved AdaBoost algorithm and CamShift algorithm for face detection and tracking in color video

摘要

Face detection and tracking is one of the core technologies of the intelligent video monitoring. This paper mainly focuses on the face detection and tracking problem based on color videos. A classification of main face identification methods is proposed and AdaBoost algorithm is discussed with emphasis. A mathematic description of is presented. As the modification of Meanshift, the CamShift is proposed to solve the face tracking problem, which is proved performing well in tracking problem. We combine these two methods and improve its performance especially in the environment which doesn't have enough light. We get good results in the experiment. The system is designed based on the VS2010 and OpenCV.