基于肤色和图像似然度的人脸检测

【作者】 段新涛；

【导师】 王耀明；

【作者基本信息】 上海师范大学 ， 计算机应用技术， 2004， 硕士

【摘要】 人脸检测问题最初作为自动人脸识别系统的定位环节被提出，近年来由于其在安全访问控制、视觉监测、基于内容的检索和新一代人机界面等领域的应用价值，开始作为一个独立的课题受到研究者的普遍重视。本文提出了一种基于肤色分割和图像似然度的检测人脸方法，该方法能在复杂的背景下很好的检测出人脸。首先利用彩色图像的颜色信息把彩色图像分割成肤色区域与非肤色区域，然后对肤色区域做预处理后，进入图像似然度检测阶段。在图像似然度检测阶段，分为训练阶段和检测阶段。在训练阶段，从人脸图像集中选取大量人脸图像的信息矩阵的奇异值向量作为矩阵中的一列而构成的人脸图像集特征矩阵。然后，用大量的人脸图像的特征向量与人脸图像集特征矩阵比较它们的相似程度，找出值小相似度，并把这个最小相似度作为阈值；在检测阶段，求出灰度图像的待测区域的特征向量与人脸特征矩阵的相似度，若该相似度大于等于阈值，则是人脸，否则不是人脸。更多还原

【Abstract】 Human face detection problem, originated in face recognition research as its first step to locate faces in the images, has recently been intensively researched as an independent problem because of its applications in many different fields including security access control, visual surveillance, content-based information retrieval, advanced human and computer interaction. In this paper we present a new approach of face detection based on skin-color and the similarity of image sets in accordance with the cognitive law. Experiment shows that the method has good detection result even in complex background. Firstly ,human skin-color information is used to segment images into skin color region and non-skin color region, then pretreat the skin-color region and come into the phase of image similarity . The phase of image similarity is divided into training phase and detection phase. In training phase , singular value vector of a lot of face image information matrix from face image sets constitutes face image sets characteristic matrix as a column. Compare a lot of face image characteristic vector with face image sets characteristic matrix in order to get their similarity , and find the least value of similarity as threshold. In the detecting phase , Compute the similarity between characteristic vector of testing region in gray image and face image sets characteristic matrix ,if the similarity bigger or equal to threshold then the testing region is a human face ,otherwise is not.更多还原