Contour Fractal Dimension Analysis using Square-Box ROI Extraction Approach with Convolution Neural Network Classifier for Palmprint Recognition System

Abstract

Contour Fractal Dimension Analysis using Square-Box ROI Extraction Approach with Convolution Neural Network Classifier for Palmprint Recognition System (CFDCNN_Net) is proposed. To bring about the originality, Contour Fractal Dimension (CFD) feature extraction approach and a Convolution Neural Network (CNN_Net) classifier approach are employed. To impart the novelty the CFD feature extraction approach, Two Dimensional-Palmprint Region of Interest (2D-PROI) is captured from five different datasets using Square-Box ROI Extraction approach and point out all the edges/contours of 2D-PROI image (CP^I) using Canny edge detection algorithm and then estimate the Fractal Dimension (FD) values using Box-Counting algorithm to create a distinctive feature vector. Classify this feature vector using Convolution Neural Network (CNN_Net) classifier approach to identify the authorized person at a higher accuracy rate. This research explores on five different datasets such as CASIA, IITD, BMPD, SMPD and multi--spectral 2D-PROI image databases. The CFDCNN_Net System model has been determined the authentication accuracy of different datasets with 98.66% of authentication accuracy.