一种改进的基于肤色分割和PCA人脸检测方法  

A Novel Approach of Face Detection Based on Skin Color Segmentation and PCA

作　　者：冯亚丽[1] 蒋文文[1] 刘泽光[1] 

机构地区：[1]大庆石油学院计算机与信息技术学院,大庆163318

出　　处：《科学技术与工程》2010年第10期2343-2347,共5页Science Technology and Engineering

基　　金：黑龙江省教育厅2008.1-2011.1科学技术研究项目资助

摘　　要：人脸检测在安全、信息处理等应用领域起到很重要的作用。提出了一种改进的基于肤色分割和PCA的人脸检测方法。首先,基于肤色模型的建立提出了一种基于边缘和数学形态学的肤色分割方法;其次,用基于知识的人脸检测方法验证候选人脸,缩小候选人脸的数量;最后,用PCA人脸检测算法进行检测。实验表明该方法不仅可以有效地运用于多人脸、不同尺寸和复杂背景的情况,而且对包含部分遮挡、肤色干扰等复杂环境下的图片均有很高的检测率。Face detection plays an important role in security,information processing and other applications.A novel approach of face detection based on skin color segmentation and PCA is presented.Firstly,a color segmentation method,based on edge and mathematical morphology,was proposed after the establishment of skin color model.Secondly,knowledge-based face detection method was applied to verify the face candidates in oder to reduce the number of face candidates.Finally,PCA-based face detection algorithm was applied for testing.Experimental results showed that this method can not only be applied effectively on many faces of different sizes and complex background situations,but also had a high detection rate on images under partly-blocked,color-interfered and other complex environments.

关 键 词：人脸检测 肤色分割 PCA 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]