基于不变矩的掌纹识别方法  

Palmprint recognition based on invariant moments

作　　者：李萍[1] 井荣枝[1,2] 尚怡君[1] 

机构地区：[1]郑州大学西亚斯国际学院,河南郑州451150 [2]华中科技大学,湖北武汉430074

出　　处：《电子设计工程》2013年第17期11-13,共3页Electronic Design Engineering

基　　金：河南省科技攻关计划项目(122102210429);西亚斯国际学院引进人才项目(2012YJRC01;2012YJRC02)

摘　　要：特征提取是掌纹识别研究中最基本的问题之一,对于掌纹图像识别而言,提取有效的分类特征是完成掌纹识别任务的一个关键问题。掌纹特征提取的本质就是把原始的高维掌纹图像样本数据投影到一个更有利于掌纹分类的低维特征子空间中。利用不变矩,对掌纹图像进行特征提取,采用最近邻分类器进行掌纹识别,在公共掌纹数据库上的实验结果表明,该方法是可行的。Feature extraction is one of the most basic problems in palmprint recognition.For palmprint image recognition,extracting the effective classifying feature of the palmprint images is a crucial step.The essence of palmprint feature extraction is to reduce the high-dimensional original palmprint image to a low-dimensional feature subspace that is benefit to classifying the palmprint.The palmprint classifying feature is extracted by using.Then the Palmprint images are recognized by adopting the K-nearest neighborhood classifier.Experiments on the public palmprint dataset validate the effectiveness and feasibility of the proposed algorithm.

关 键 词：掌纹识别 特征提取 不变矩 最近邻分类器 

分 类 号：TP931[自动化与计算机技术]