一种基于AdaBoost的人脸检测算法

针对常规基于肤色检测的AdaBoost算法的不足, 提出了一种改进的AdaBoost人脸检测算法，算法包括人体肤色模型、人脸运动检测模型、改进的背景提取方法、针对人脸区域的光照增强方法。算法综合利用了人体肤色信息和人脸运动信息，能有效缩小搜索范围。实验结果表明，该方法与常规基于肤色检测的AdaBoost方法相比，在保证检测性能的基础上，有效提高了检测速度。

Face detection method based on Ada Boost algorithm

In order to overcome shortcomings of traditional skin detection based AdaBoost methods, an improved face detection method which is based on AdaBoost algorithm is proposed, including human skin model, motion detection, an optimized background extraction algorithm and an illumination enhancement method that only process face regions. This method can effectively decrease search scope using information of human skin and face motion in detected videos. Experimental results show that this approach could achieve higher speed and better detection performance compared to normal skin detection based AdaBoost algorithms.