摘要
针对复杂背景彩色图像提出了一种基于HSV色彩空间的自适应肤色检测算法。该算法首先使用阈值在HSV空间对人体肤色区域进行肤色分割,然后对分割出的肤色区域使用相对重要性滤波和自适应区域归并,最后将归并后的肤色区域使用人眼定位进行验证,将多人脸检测转化为单人脸检测。实验结果表明,该算法复杂度较小,对光照变化具有很好的鲁棒性。
In this paper,a self-adaptive skin color detection algorithm for color images based on HSV color space,which is composed of skin color segmentation using H threshold,relative significance filter and self-adaptive region merging is presented.First,a threshold of H is used for segmenting skin color regions in HSV color space,and then we use relative significance filter and self-adaptive region merging to process the skin color regions which have been seg-mented.In the end,an eye locating algorithm is used for searching faces in these skin color regions.This method con-verts multi face detection to single face detection and integrates skin color information of color images and other infor-mation of grayscale images in face detection.Experimental results demonstrate the simpleness,efficiency and robustness for variety of light of this algorithm.
出处
《计算机工程与应用》
CSCD
北大核心
2004年第14期81-85,共5页
Computer Engineering and Applications
关键词
肤色检测
相对重要性滤波
自适应区域归并
skin color detection,relative significance filter,self-adaptive region merging
