基于新颜色空间YCgCr的人脸检测与定位

提出了一种基于肤色的复杂背景彩色图像中人脸检测的方法。首先在新颜色空间YCgCr上对输入图像进行肤色检测并应用预处理技术缩小人脸检测的搜索区域；然后在物体区域方向计算的基础上，提出了能够检测任意旋转角度人脸的方法；最后对人脸候选区域采用模板匹配方法检测人脸，去除非人脸区域，降低误检率。实验结果表明，提出的方法对不同尺寸、任意旋转的人脸有较好的检测效果。

Face Detection and Location Based on a Novel Color Space YCgCr

Presenting a method for face detection based on the skin color in color images with complex backgrounds.Firstly, skin color of the inputed images is detected in the YCgCr color space and preprocessing techniques are introduced to reduce the searching areas of the face detection.Secondly,proposing a method that can detect arbitrary rotated faces based on the compu- tation of the direction of the object.Finally,using template-matching method in the face candidate regions to detect faces,re- moving the non-face regions,reducing the false detection.The experiment results show that the proposed face.detection meth- od has more better detection results to different scales and arbitrary rotated faces.