【作者】 李华胜； 王延江； 袁保宗；

【Author】 LI Hua-sheng Wang Yan-jiang YUAN Bao-zong (Institute of Information Science.NJTU.,Beijing 100044)

【机构】 北方交通大学信息所；

【摘要】 本文提出了一种利用人脸颜色、形状和纹理相结合复杂背景下彩色图像中人脸检测方法。该方法先利用HSV肤色模型进行人脸侯选区域的粗检测及分割,然后对人脸候选区域进行椭圆形状分析筛选人脸的候选集合,再利用主分量分析法(PCA)对候选区域的人脸进行确认。实验结果表明,这种方法效率高,可以检测复杂背景下不同大小和平面内旋转角度不大的正面人脸。更多还原

【Abstract】 This paper presents an effective algorithm by combining facial color,shape and texture to detect faces in color images under cluttered background.First,facial candidates are coarsely detected and segmented by using HSV color modal.Then shape analysis is applied to each detected candidate to decide the possible face-like regions.Finally PCA is used to decide whether each candidate really represents a face or not.Experimental results show that the proposed approach is effective and can detect faces with different size and small rotation angle in image plane under cluttered background.更多还原