基于分类性能的掌纹特征有机融合

PALMPRINT FEATURES FUSION IN ORGANIC WAY BASED ON CLASSIFICATION PERFORMANCE

为了克服利用单一特征进行掌纹识别的局限性,提出一种基于分类性能的掌纹多特征有机融合方法。该特征融合方法通过三步来实现:特征选择、加权处理和降维处理。在分析单一特征分量聚类性能的基础上,通过构造判别准则并设定相应阈值对特征进行取舍;通过构造类内距离与类间距离之比这一函数计算得到每个特征分量的权值,然后进行加权处理;最后利用主分量分析法对特征进行降维处理。以改进的LBP算法和离散小波变换提取掌纹的两种特征,将提取的特征进行融合实验,结果表明了该方法的有效性。

In order to overcome the limitation of palmprint recognition by using single feature,a method of palmprint multi-feature fusion in an organic way based on classification performance is proposed in this paper.This features fusion method is realised in three steps： the feature selection,the weighting processing and the dimension reduction processing.Based on analysing the clustering performance of each single feature component,the features＇ trade-off is carried out by constructing the criterion and set the corresponding threshold value;the weight value of each feature component is calculated by constructing the function of within-class distance and between-class distance,and the weighting processing is then performed;at last the principal components analysis（PCA） is used to reduce the dimension of the feature.The improved Local Binary Patterns（LBP） and Discrete Wavelet Transform（DWT） are used to extract the two kinds of palmprint features,and these extracted features are used in fusion experiments,the results show the effectiveness of the method.