摘要
本文提出一种在HSV空间快速进行人脸检测定位的肤色分割方法,并且与HSV空间中S分量空间特征提取方法相结合,提取人脸特征;然后对有特征点的区域,利用区域投影,包括水平投影和垂直投影,比对投影曲线,确定该区域是否符合人脸,进一步准确认证人脸。经过大量试验检验,本算法在任何背景下都具有较高的人脸检测准确率,同时在速度上可达到实时检测的要求。
This paper proposes a fast human face detection algorithm based on skin color in the HSV space. Facial features of the S component in the HSV space are used for fine face detection authentication. For the valid region, the skill of region projection is used to determine the accurate area of person face . The massive experi- ments showed that this algorithm has the high accurate rate and the real - time detection for person face examination under any background.
出处
《信息技术与信息化》
2009年第3期68-69,共2页
Information Technology and Informatization
关键词
人脸检测
肤色
脸部特征
特征提取
积分投影
Face detection Skin color Facial feature Feature abstraction Integral projection