摘要
人脸检测是人脸信息处理领域中的一项关键技术,当图像中具有复杂背景或多人脸时,不能够快速准确地检测人脸,成为了人脸检测领域的重要的问题。论文首先提出了一种基于双阈值的改进型Ada Boost分类器算法,使用双阈值的弱分类器代替单阈值的弱分类器,从而提高单特征的分类能力,并给出了双阈值的搜索方法;其次提出了一种基于Harr特征的改进后的Ada Boost分类器融合YCb Cr空间高斯肤色模型的人脸检测方法,首先采用基于Harr特征的改进型Ada Boost人脸检测算法对输入图片进行预检测,再用基于高斯肤色模型的检测算法剔除误检的人脸区域。实验结果表明,该文提出的方法检测率高、误检率低,具有可靠的检测性能。
Face detection is a key technology in face information processing,in the field of face detection the important problem is that can not detect face quickly and accurately when the detected image has complex background or contain more faces.In this paper,an improved AdaBoost classifier algorithm based on double threshold is proposed.Using a weak classifier with double threshold takes the place of a weak classifier with single threshold to improve the ability of classifying the single feature,and giving a search method based on double threshold.Then a detection method which combining improved AdaBoost classifier based on Harr feature and skin Gaussian model in YCbCr color space is proposed.Firstly,the input image is pre-tested by using the face detection algorithm based on Harr feature and modified AdaBoost,and then based on the Gaussian skin color model of the detection algorithm to be verified the false face.According to the result of simulation,the method proposed in this paper has high detection rate and low false detection rate,and it has reliable detection performance.
作者
姚子怡
张清勇
李雪琪
YAO Ziyi;ZHANG Qingyong;LI Xueqi(Hubei Provincial Shuiguohu High School,Wuhan 430070;School of Automation,Wuhan University of Technology,Wuhan 430070)
出处
《计算机与数字工程》
2018年第4期743-749,共7页
Computer & Digital Engineering
