Abstract
In this research paper, Compressed Contour Cumulative GLCM Texture Analysis Model based CNN for Palmprint Recognition (3CGLCM-CNNNet) System is proposed to make up the higher level security assurance in the biometric technology. It can be imparted by second-order statistics using a Cumulative Gray-Scale Level Co-occurrence Matrix (CGLCM) feature extraction approach and Convolution Neural Network (CNNNet) classification approach. To enact this, Two Dimensional-Palmprint Region of Interest (2D-PROI) is pre-processed and contour of 2D-PROI image (CP) is captured using canny edge detection algorithm. Linear Hybrid Conventional Compression Algorithm (LHCC) is applied to constitute the compressed contour 2D-PROI (CCPI) image. In this LHCC algorithm, conventional Discrete Wavelet Transform (DWT) and Principal Component Analysis (PCA) compression algorithms is composited in a linear manner on CP. Ideal second-order statistical texture features of CCPI are clipped using CGLCM approach. Ideal features are forwarded into 3CGLCM-CNNNet classification algorithm to match the recognized persons. Research is worked on 2D-PROI data derived from the POLYU database, Hong Kong Polytechnic University, Hong Kong. Our proposed system’s benchmark has been evaluated and tabulated with higher acceptance of 99% recognition accuracy compared with other existing approaches in biometric technology.
Keyword
Biometric Technology, Palmprint Biometric Trait, Gray-Scale Level Co-occurrence Matrix, Convolution Neural Network, Two Dimensional-Palmprint, Region of Interest, Canny edge detection, District Wavelet Transform, Principal Component analysis, Statistic texture features.
PDF Download (click here)
|