基于多色彩空间的肤色检测研究

摘要

In view of the problem that human skin color regions can not be quickly and correctly identified as well as those similar human skin color regions are mistakenly identified as skin color regions in image processing,through collecting a large number of images for sta- tistical analysis,a skin detection algorithm is proposed based on multi-color spaces. The algorithm converts RGB color space information into YUV,YIQ and HSI color space information respectively,combined with space chromaticity,hue and saturation to do image threshold segmentation,the area that contains human skin is obtained. It is shown from the experiments that the algorithm is simple,easy to realize, can quickly and accurately detect human skin area.%针对图像处理中人的肤色区域不能快速正确识别以及人的肤色相似区域误识别为肤色区域的问题，通过采集大量图像进行统计分析，提出一种基于多色彩空间的肤色检测算法。该算法把图像的RGB色彩空间信息分别转换为YUV、YIQ和HSI色彩空间信息，结合空间色度、色调和饱和度分量信息对图像进行阈值分割，得到包含人类肤色的区域。实验结果表明该算法简单，容易实现，能快速准确地检测到包含人的肤色区域。