摘要
首先利用统计原理检测出一幅图像的肤色区域,再通过哈达玛变换将这些区域从像素域变换到压缩域,最后通过改进和训练链码模板和角度模板来对候选肤色区域进行匹配和检测。实验结果显示,该方法在检测人脸的过程中速度快,漏检率较低,能正确地分辨人脸的肤色、胖瘦,算法设计简单可行,非常适合实时性要求高的图像检索系统。
In this paper, firstly, the skin area of one image was detected with the statistics theory; next, this area was converted into compressed domain through discrete Hadamard transform; lastly, the new chain code template and angle template were proposed to judge whether the face region exists or not. The experimental results show that the proposed algorithm is superior in terms of human face retrieval precision, speed. It can distinguish face color, fat or thin, and is very suitable for real-time image retrieval system.
出处
《计算机应用》
CSCD
北大核心
2010年第A01期169-171,176,共4页
journal of Computer Applications
基金
重庆航天职业技术学院科学技术项目(2010-YJKY-4)
关键词
人脸检测
链码
角度模板
哈达玛变换
图像检索
human face detection
chain code
angle template
hadamard transform
image retrieval
