摘要
首先利用小波变换增强掌纹、人脸图像;然后利用一种新的子空间分析方法——对角离散余弦变换和二维主元判别分析(Diagonal,Discrete Cosine Transform and Two-Dimensional Principle Component Analysis,Dia-DCT+2DPCA)相结合的算法提出了一种掌纹、人脸特征融合的识别方法;最后运用最小距离分类器进行识别。实验结果表明,该文提出的掌纹、人脸特征融合方法实现了特征层融合,有效地提高了身份识别的正确识别率。
Firstly, palmprint and face feature are enhanced by wavelet. Secondly, combined with diagonal discrete cosine transform and two-dimensional principal component analysis (Dia-DCT+2DPCA), a novel concept for subspace analysis is presented to construct the model of palmprint and face feature fusion recognition. Finally, identity recognition can be realized according to the nearest neighbor rule. Experimental results show that feature fusion is realized. Higher correct recognition rate demonstrates the validity of this method.
出处
《电脑知识与技术》
2009年第1X期689-690,692,共3页
Computer Knowledge and Technology
基金
国家自然科学基金项目(60872057)
浙江省自然科学基金项目(Y107759
Y1080212)
浙江省科技计划项目(2009C23G2200003)
关键词
小波增强
特征融合
掌纹识别
人脸识别
身份识别
wavelet enhancement
feature fusion
palmprint recognition
face recognition
identity recognition
