中文摘要：

针对目前大多数掌纹识别方法对于单训练样本系统识别性能欠佳的问题,提出一种基于小波子带融合的主成分分析方法,用于单训练样本掌纹识别.该方法将小波低频子带与水平和垂直高频子带相结合进行身份识别,使用低通滤波增强相应边缘信息的鲁棒性,以提高高频子带的识别性能,然后采用求和算子对各匹配分数进行融合.实验结果表明,对于单训练样本掌纹识别,该方法平均识别率达89.93%,较传统方法有6%~9%的性能提升.

英文摘要：

In view of the poor performance of the present most palmprint recognition for single training sample system,a principal components analysis method for single training sample palmprint recognition was presented,which combined multi-subbands of wavelet transformation.This method combined wavelet low frequency subband with horizontal and vertical subbands to identify.Low-pass filter was utilized to enhance the robustness of horizontal and vertical subbands,and the summation operator was used to fuse their matching scores.Experimental results showed that for single training sample palmprint recognition the average recognition rate of the proposed method was 89.93%,which was 6%~9% higher than some of the traditional algorithms.