一种基于稀疏表达和神经网络的人脸识别算法  

A FACE RECOGNITION ALGORITHM BASED ON SPARSE EXPRESSION AND NEURAL NETWORK

作　　者：祁彦庆 汪烈军[1] 吴生武[1] 

机构地区：[1]新疆大学信息科学与工程学院,新疆乌鲁木齐830046

出　　处：《计算机应用与软件》2016年第10期172-175,共4页Computer Applications and Software

基　　金：国家自然科学基金项目(61471311)

摘　　要：传统的基于神经网络的人脸识别算法直接从灰度空间获取人脸图像数据,其中含有大量的噪声和冗余信息,降低了识别率且延长了识别时间。提出一种基于稀疏表达和神经网络的人脸识别算法:首先通过KSVD算法将样本变换至稀疏空间,然后运用LDA算法将稀疏编码变换至子空间,最后输至RBF神经网络进行分类。在ORL和Yale人脸库上的实验结果表明,该算法比其他算法具有更高的识别率和更快的识别速度,且具有较强的鲁棒性和泛化能力。Traditional neural network-based face recognition algorithm obtains face image data directly from gray space. It contains a lot of noise and redundant information and reduces recognition rate as well as extends recognition time. This paper presents a new face recognition method,which is based on sparse expression and neural networks. First,it transforms training samples into sparse space by KSVD algorithm,then it operates LDA algorithm to transform these sparse codes into subspace, finally they are inputted to RBF neural network for classification. The results of experiments on ORL and Yale face databases show that the proposed algorithm has higher recognition rate and faster recognition speed than other algorithms,and has strong robustness and generalisation ability as well.

关 键 词：人脸识别 KSVD 稀疏空间 LDA RBF 神经网络 

分 类 号：TP3[自动化与计算机技术—计算机科学与技术]