关键词: AdaBoost, Struck, 人脸检测, 实时跟踪

Abstract:

The classic Struck algorithm is incapable of automatic tracking of the human face and easily subject to the environment interference.In the paper,the classical Struck method is improved to achieve automatic real-time face tracking by the Adabboost algorithm.Firstly,the Haar feature of the face is extracted.Then the AdaBoost algorithm is used to detect face target and initialize the Struck tracker automatically.The similarity between the detected target and the last result is computed to get the best result,with Struck tracking the face continuously.Experimental results show the improved algorithm applies in such complex environment as partial occlusion,scale variation,illumination variation with good robustness and accuracy.

Key words: AdaBoost;Struck;face detection;real time tracking