摘要
人脸检测是模式识别与计算机视觉领域内一个重要的研究热点。人脸是一个常见而复杂的视觉模式,易受干扰。目前常见的人脸检测算法一般具有计算量大、运行速度慢、误检率高等弱点。基于一种新的颜色空间YCgCb的高斯肤色模型,利用颜色信息把彩色图像分割为肤色区域和非肤色区域,并对肤色区域进行去噪处理,实现脸部区域的具体定位,最后标记出来。实验表明该肤色模型可以有效地检测出人脸,误检率低。
Face detection is an important research of pattern identification and computer vision.Face is a familiar and complex vision model and been disturbed easily.At the moment,the familiar face detection algorithm has a lot shortcomings,such as large calculate,low speed and high error detection.This paper is based on a new color space YCgCb of Gaussian complexion model and divided the multicolor image into complexion regions and non-complexion regions by the color information.Then smoothen the complexion regions so as to succeed in the localization of face and marked it.The experiment expresses that complexion model can examine a person face availably,and the rate of mistake is low.
出处
《计算机安全》
2010年第12期49-51,共3页
Network & Computer Security
