3D palmprint recognition using unsupervised convolutional deep learning network and SVM classifier. Issue 5 (20th March 2019)

• Record Type: Journal Article
• Authors: Chaa, Mourad; Akhtar, Zahid; Attia, Abdelouahab
• Is Part Of: IET image processing. Volume 13:Issue 5(2019)
• Page Start: 736
• Page End: 745
• Publication Date: 2019-03-20
• Subjects: palmprint recognition -- learning (artificial intelligence) -- image texture -- support vector machines -- image classification -- feature extraction -- convolutional neural nets; SSR images -- multiclass support vector machine classification scheme -- SVM classifier -- hand‐based automatic person recognition systems -- hand‐based biometric traits -- biometric characteristic -- academic communities -- industrial communities -- features richness -- 3D palmprint recognition -- distinctive texture patterns -- single scale retinex algorithm -- PCANet unsupervised convolutional deep learning network -- illumination‐invariant 3D palmprint images -- PCANet topology -- 3D palmprint PolyU dataset; Image processing -- Periodicals; 621.36705
• DOI: 10.1049/iet-ipr.2018.5642
• Languages: English
• ISSNs: 1751-9659

Abstract

Abstract : Since past decade, efforts are afoot to design better hand‐based automatic person recognition systems. Among the various hand‐based biometric traits, palmprint as a biometric characteristic is now gaining increased attention from both the academic and industrial communities owing to its highly distinctive texture patterns, features richness, and stability. Here, the authors propose a new 3D palmprint recognition framework based on an unsupervised convolutional deep learning network named PCANet. Specifically, the proposed framework first reconstructs illumination‐invariant 3D palmprint images using Single Scale Retinex (SSR) algorithm. Then, PCANet topology is employed to extract discriminative features from SSR images. Finally, a multi‐class support vector machine (SVM) classification scheme is utilised to determine the identity of the person. Extensive experimental analysis on publicly available 3D palmprint PolyU dataset, which is composed of 8000 range images from 200 individuals, shows that proposed method outperforms existing approaches and is also able to attain 99.98% rank‐1 accuracy.