基于PFICA和模糊积分的人脸识别  

Face Recognition Based on PFICA and Fuzzy Integral

作　　者：王婵娟[1] 张强[1] 魏小鹏[1] 周昌军[1] 

机构地区：[1]大连大学先进设计与智能计算省部共建教育部重点实验室,大连116622

出　　处：《系统仿真学报》2009年第9期2578-2581,共4页Journal of System Simulation

基　　金：国家自然科学基金(60875046);教育部新世纪优秀人才支持计划(NCET-06-0298);辽宁省高等学校优秀人才支持计划(RC-05-07;2006R06);辽宁省高校创新团队项目(2008T004);辽宁省教育厅科学研究计划(05L020);大连市科学技术计划(2005A10GX106);大连大学辽宁省智能信息处理重点实验室开放课题(2006-9)

摘　　要：提出了一种基于PFICA和模糊积分的人脸识别方法,即先对原始图像数据用PCM+FCV同时进行预处理和模糊划分,再在各模糊类中用FastICA提取局部独立成分,并根据最近邻原则进行初步识别,最后运用模糊积分对各类中的识别结果进行融合,得到最终的识别结果。在ORL人脸数据库中的实验仿真证明该方法具有较好的识别效果。A face recognition method based on PFICA and fuzzy was proposed. Firstly, PCM＋FCV was applied to the original image data in order to preprocess and partition the data simultaneously. Secondly, in each fuzzy class, FastICA was applied to extract local independent components, and primary recognition was carried on based on NNP. Finally, the recognition results of each class were fused by fuzzy integral to get the final recognition result. The computer simulation illustrates the validity of this method on the ORL database.

关 键 词：人脸识别 可能性c均值 模糊c种类 模糊积分 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]