基于纹理和邻域信息的肤色分割算法

Algorithm of Skin Color Segmentation Based on Texture and Neighborhood Information

为有效分割图像中的肤色信息,提出一种基于纹理和肤色点的邻域信息肤色分割算法.首先,根据图像中的肤色点在RGB颜色空间上的分布规律,利用统计直方图模型对肤色部分进行初步分割;其次,为提取图像中的纹理信息,采用Gabor滤波器进一步过滤掉与人类皮肤颜色接近但纹理粗糙的区域;最后,采用掩码图像中的肤色点作为种子点进行邻域信息的扩散,得到最终掩码图像.实验结果表明,与统计直方图肤色模型相比,该方法肤色分割效果更好.

In order to segment the information of skin color in the image effectively,we proposed a algorithm of skin color segmentation based on texture and neighborhood information of skin point.Firstly,according to the distribution of skin color point in the image in RGB color space,we used the statistical histogram model to segment the part of skin color in the image initially.And then,in order to extract the texture information in the image,we used the Gabor filter to further filter out the part of closing to human skin color but the texture was rough.Finally,we used the skin color point in the mask image as the seed point to diffuse neighborhood information,and then got the final mask image.The experimental results show that,compared with the skin color model of statistical histogram,the method is better in skin color segmentation.