Slim-ResCNN: A Deep Residual Convolutional Neural Network for Fingerprint Liveness Detection

Yongliang Zhang,Xiaosi Zhan,Keyi Zhu,Di Cao,Daqiong Shi,Zhiwei Li

doi:10.1109/access.2019.2927357

Abstract

Fingerprint liveness detection has gradually been regarded as a primary countermeasure for protecting the fingerprint recognition systems from spoof presentation attacks. The convolutional neural networks (CNNs) have shown impressive performance and great potential in advancing the state-of-the-art of fingerprint liveness detection. However, most existing CNNs-based fingerprint liveness methods have a few shortcomings: 1) the CNN structure used on natural images does not achieve good performance on fingerprint liveness detection, which neglects the inevitable differences between natural images and fingerprint images; or 2) a relative shallow architecture (typically several layers) has not paid attention to the capability of deep network for spoof fingerprint detection. Motivated by the compelling classification accuracy and desirable convergence behaviors of the deep residual network, this paper proposes a new CNN-based fingerprint liveness detection framework to discriminate between live fingerprints and fake ones. The proposed framework is a lightweight yet powerful network structure, called Slim-ResCNN, which consists of the stack of series of improved residual blocks. The improved residual blocks are specifically designed for fingerprint liveness detection without overfitting and less processing time. The proposed approach significantly improves the performance of fingerprint liveness detection on LivDet2013 and LivDet2015 datasets. Additionally, the Slim-ResCNN wins the first prize in the Fingerprint Liveness Detection Competition 2017, with an overall accuracy of 95.25%.

Highlights

The fingerprint recognition technology is extensively employed in border control applications and personal identification verification systems, owing to its high reliability, high generalization, and low cost
The excellent performance of Slim-ResCNN model is further confirmed in that the model wins the first place on the Fingerprint Liveness Detection Competition 2017, with an overall accuracy of 95.25%
On the basis of the improved residual blocks, we propose a novel architecture which consist of nine improved residual blocks, called SilmResCNN, which is superior for fingerprint liveness detection

Summary

INTRODUCTION

The fingerprint recognition technology is extensively employed in border control applications and personal identification verification systems, owing to its high reliability, high generalization, and low cost. It is difficult to update these additional hardware devices when the attackers improve artificial replicas with new manufacturing technology and pass the fingerprint recognition systems successfully. Software-based methods, on the other hand, have gained an increasing attention, which uses image processing technology to extract features from the captured fingerprint images so as to identify the live and fake fingerprints without additional hardware devices. In literature [13], the weber local descriptor (WLD) is utilized to prevent spoof attacks on fingerprint sensors, where the input fingerprint images are represented by extracting two-dimensional histogram features from differential excitation and square bipartite. The proposed Slim-ResCNN framework is different from the original residual network in that only nine improved residual blocks are stacked into Slim-CNN and less convolutional kernels are employed, making less training time and improving classification performance for fingerprint spoof detection.

PROPOSED METHOD

EXPERIMENTAL ENVIRONMENT AND PERFORMANCE EVALUATION METRICS

Findings

CONCLUSION