An Improved Algorithm based on Palm Print Recognition

Abstract. This paper aims at the deficiency of classical PCA algorithm on only extracting linear palm print features, sufficiently uses the double matching thought of recognition technology to propose a new algorithm which relatively efficient in extracting the classified identification information. The new proposed algorithm based on PCA algorithm and by using transformation of kernel function to convert the input samples into high-dimensional space through which efficiently solving the problems of small samples and sharply rising computation quantities, while the two aforementioned problems were caused by the number of training samples after matched to the high-dimensional space that smaller than the dimensions themselves. Experimental results show that the proposed algorithm is effective and the accuracy is improved obviously