基于多模型及SVM的单人脸跟踪系统

针对彩色视频中的人脸检测与跟踪问题，使用运动模型和自适应肤色模型，从图像中快速提取出人脸候选区，然后利用基于先验规则和SVM的方法进行确认。对于被确认的人脸，建立一个人脸状态记录表，通过位置预测，使用三步搜索法进行人脸区域色度特征匹配跟踪。实验表明，本文提出的方法，在复杂的环境中，能实时地、精度较高地跟踪自由运动的人脸。

Multi-Model and SVM Used in Face Tracking

As to the problem of face detection and tracking in color video frequency, we try to use the motion model and the adaptable skin model to extract the face-like region quickly at first.Then we use rule-based face verification algorithm and SVM method to verify it. After the verification of the human face, we create the face status table and track the human faces based on chromatic histograms matching through three-step-search method after the location's predicton. The experiment results show that the proposed method is more robust and powerful than the other methods.