提出一种基于肤色分割和几何特征相结合的人脸检测算法。该算法对图像进行光照补偿预处理,增强图像对比度;采用YCgCr色彩空间的人脸肤色模型,对图像进行肤色分割,得出肤色轮廓;利用人脸的几何特征筛选出人脸区域。实验结果表明,该算法检测率高,实时性好,误检率低。
We propose a human face detection algorithm which is based on the combination of complexion segmentation and geometry features. The algorithm preprocesses the image by light compensation and the contrast of image is thus enhanced. YCgCr colour space is applied to face complexion model for complexion segmentation on the face image and the complexion contour is detecled. The geometry features of face are used to filter off the non-face regions. Experimental results show that the algorithm has high detection rate, good real-time performance and low false detection rate.