基于脸部信息和支持向量机的人脸检测

Face detection based on facial information and SVM

人脸检测是一个复杂而又非常有意义的模式识别问题。针对目前人脸检测算法于速度和精度不能兼优的问题,提出了一种基于脸部信息及支持向量机的人脸检测方法。算法首先利用肤色模型进行人脸粗检,然后根据人脸几何特征进行筛选,最后通过奇异值分解输入支持向量机分类。实验结果表明,该方法是十分有效的。

Face detection is a complicated and significant problem of pattern recognition, which can be widely used. An algorithm for face detection with fast speed and accurate precision was presented, which integrates facial information and support vector machine. First, the skin color model is used to detect human face, then each candidate region will be evaluated by geometric feature of face, finally the appropriate features by singular value decomposition will be passed to the SVM classifier for the final decision. The experimental results demonstrate the feasibility of this approach.