Finger Vein Authentication Based on Wavelet Scattering Networks

• DOI: 10.32604/cmc.2022.016410
• Publication Date: 2022-03-29T06:13:38Z
• Authors: Amjad Rehman; Majid Harouni; Maedeh Omidiravesh; Suliman Mohamed Fati; Saeed Ali Bahaj
• Keywords: Gabor wavelet; Word error rate; Identification; Gabor filter

Biometric-based authentication systems have attracted more attention than traditional techniques such as passwords in the last two decades. Multiple biometrics fingerprint, palm, iris, palm vein and finger other been introduced. One of challenges is physical injury. Biometric least exposed to damage. Numerous methods proposed for with help this biometric that suffer from weaknesses high computational complexity low identification rate. This paper presents a novel method scattering wavelet-based identity identification. Scattering wavelet extracts image features Gabor filters structure similar convolutional neural networks. What distinguishes algorithm popular feature extraction deep learning methods, filter-based statistical etc., has very skill accuracy differentiating images but belongs different classes, even when subject serious damage noise, angle changes or pixel location, descriptor still generates vectors way minimizes classifier error. improves classification authentication. The evaluated using databases Finger Vein USM (FV-USM) Homologous Multi-modal Traits (SDUMLA-HMT). In addition having reasonable complexity, it recorded excellent rates rotation, transmission challenges. At best, 98.2% rate SDUMLA-HMT database 96.1% FV-USM database.