【作者】 胡韦伟； 汪荣贵； 胡琼； 李守毅；

【Author】 Hu Weiwei, Wang Ronggui, Hu Qiong, Li Shouyi(Faculty of Computer and Information, Hefei University of Technology, Hefei 230009, China)

【机构】 合肥工业大学计算机与信息学院；

【摘要】 针对人脸跟踪系统中普遍存在的定位难的问题,提出了一种基于Adaboost定位的实时人脸跟踪方法。该方法分为检测和跟踪两个阶段:检测阶段用Adaboost方法快速的定位人脸,并用人脸的定位信息初始化卡尔曼滤波器。跟踪阶段用卡尔曼滤波器结合肤色特征跟踪人脸,如果跟踪失败,转入检测阶段。实验结果表明,该方法可以准确的定位人脸并且能实时的跟踪人脸。更多还原

【Abstract】 It is hard to locate the face before track it, which inspire us to develop a face tracking method using Adaboost algorithm. This method has two stages: detecting and tracking. In the detecting stage, initial the kalman filter with the size and the position of the face which located by using Adaboost algorithm. In the tracking stage, track face using kalman filter and skin-color feature, if fail to track then turn into detecting stage.Experimental result shows that the human face are exactly located and successfully tracked in real time.更多还原