摘要
提出了一种基于证据推理的多特征融合人脸识别算法(DSPSA)。该算法利用证据推理理论在处理不确定和冲突信息方面的优越性,融合多个面部特征的信息,有效地处理了人脸图像由于光照、旋转、表情等因素造成不确定信息,从而达到改善识别结果以及增强识别系统对训练样本库以外类别的识别能力。算法中提出了新的基本置信指派构造公式。
A face recognition algorithm using multiple feature based on evidence reasoning （DSPCA） is presented. With the advantage of evidence reasoning on dealing with the uncertain and conflict information, the algorithm manages the uncertain information of faces because of illumination, circumgyrate and expression by fusion of several facial features. The algorithm is used to improve the result of recognition and the ability of recognizing the sort beyond the training sorts. The formula of constructing basic belief assignment is given.
出处
《测控技术》
CSCD
北大核心
2009年第5期32-34,37,共4页
Measurement & Control Technology
基金
国家自然科学基金资助项目(60634030)
航空科学基金资助项目(2006ZC53037
2007ZC53037)
高等学校博士学科专项科研基金资助项目(20060699032)
关键词
证据推理
人脸识别
概率子空间
evidence reasoning
face recognition
probabilistic subspaces
