2018/01/03 by Ali Dabouei, Dabouei, Ali, Hadi Kazemi +7
Computer Science · Engineering · #Biometric Identification and Security #Computer Vision and Pattern Recognition (cs.CV) #Digital Media Forensic Detection #FOS: Computer and information sciences #Gait Recognition and Analysis
paper · pdf · doi:10.48550/arxiv.1801.01198
openalex publication_date 2018/01/03 · openalex created_date 2022/10/07 · openalex updated_date 2026/07/28
Elastic distortion of fingerprints has a negative effect on the performance\nof fingerprint recognition systems. This negative effect brings inconvenience\nto users in authentication applications. However, in the negative recognition\nscenario where users may intentionally distort their fingerprints, this can be\na serious problem since distortion will prevent recognition system from\nidentifying malicious users. Current methods aimed at addressing this problem\nstill have limitations. They are often not accurate because they estimate\ndistortion parameters based on the ridge frequency map and orientation map of\ninput samples, which are not reliable due to distortion. Secondly, they are not\nefficient and requiring significant computation time to rectify samples. In\nthis paper, we develop a rectification model based on a Deep Convolutional\nNeural Network (DCNN) to accurately estimate distortion parameters from the\ninput image. Using a comprehensive database of synthetic distorted samples, the\nDCNN learns to accurately estimate distortion bases ten times faster than the\ndictionary search methods used in the previous approaches. Evaluating the\nproposed method on public databases of distorted samples shows that it can\nsignificantly improve the matching performance of distorted samples.\n