一种改进的高亮度区域肤色像素检测算法

高亮度条件下肤色和非肤色具有很高的相似性，导致高亮度肤色像素的检测一直是基于肤色分割的一个难点。为了提高肤色分割的效果，提出一种能够检测出图像中高亮度肤色区域的方法。首先在YCbCr颜色空间中利用阈值自适应的单高斯模型检测出图像中具有正常亮度的肤色像素，再结合像素的邻域信息和亮度边缘信息进一步检测出图像中高亮度区域的肤色像素，有效地解决了传统肤色检测方法对高亮度肤色区域漏检的问题。实验结果表明，采用该算法可以较准确地检测出图像中的高亮度肤色区域。

An Improved High Illuminated Skin Color Detecting Algorithm

Skin color detection in a high illuminated region has been a challenging issue in skin-color- based segmentation due to the great similarity in both skin color pixel distribution and non-skin color pixel distribution. To address this problem, a novel skin color detecting algorithm for high illuminated pixels is proposed. Firstly, an adaptive threshold Single Gaussian Model in YCbCr color space is exploited to detect the normal illuminated skin color pixels in the image. Then, based on the principle of spatial and lumi- nance continuity, the skin color pixels in high illuminated regions are detected. Therefore, the undetected ratio of high illuminated skin color pixels is greatly reduced. Experimental results indicate that the pro- posed algorithm can greatly improve the skin color pixel detecting performance in high illuminated regions.