Multispectral palmprint recognition based on three descriptors: LBP, Shift LBP, and Multi Shift LBP with LDA classifier

Abstract

Local Binary Patterns (LBP) are extensively used to analyze local texture features of an image. Several new extensions to LBP-based texture descriptors have been proposed, focusing on improving noise robustness by using different coding or thresholding schemes. In this paper we propose three algorithms (LBP), Shift Local Binary Pattern (SLBP), and Multi Shift Local Binary Pattern (MSLBP),to extract features for palmprint images that help to obtain the best unique and characteristic values of an image for identification. The Principal Component Analysis (PCA) algorithm has been applied to reduce the size of the extracted feature matrix in random space and in the matching process; the Linear Discriminant Analysis (LDA) algorithm is used. Several experiments were conducted on the large multispectral database (blue, green, red, and infrared) of the University of Hong Kong. As result, distinguished and high results were obtained where it was proved that, the blue spectrum is superior to all spectra perfectly.