A Deep Learning Multifractal Texture Analysis Using Principal Line Extraction Approach for Palmprint Recognition System

Abstract

To make the methodical Palmprint Recognition System (PRS), this research proposed an unique Deep Learning classifier using the palm hand’s principal lines extraction approach and multifractal texture analysis approach. To reveal the efficient biometric security, a Deep Learning Multifractal Texture Analysis for Palmprint Recognition System (DLMTA-PRS) has been suggested. In DLMTA-PRS system, exact principal lines of Two–Dimensional Palmprint Region of Interest (2DPROI) image are extracted using morphological operations and an edge detection algorithm in a peculiar manner. Then, Feature values of 2DPROI image are fetched using multifractal texture analysis approach. To perform this approach, Box-counting and Gliding-Box algorithms are performed and the feature vector is created. The feature vector is classified using the proposed Convolution Neural Network (CNNNet) classifier approach to get the higher authentication security. The multi-spectral 2DPROI image database has been utilized for this research which is acquired from the PolyU, the Hong Kong Polytechnic University in Hong Kong. The suggested DLMTA-PRS system underwent scrutiny as well as proved the best of its by evaluating the DLMTA-PRS system using numerous criterias with the achievement of getting 99.25% accurate identification rate.