Face Spoofing Detection Based on Multiple Descriptor Fusion Using Multiscale Dynamic Binarized Statistical Image Features

Abstract

Abstract-Face recognition has been the focus of attention for the past couple of decades and, as a result, a significant progress has been made in this area. However, the problem of spoofing attacks can challenge face biometric systems in practical applications. In this work, an effective countermeasure against face spoofing attacks based on a kernel discriminant analysis approach is presented. Its success derives from three innovations. First it is shown that the recently proposed multiscale dynamic texture descriptor based on binarized statistical image features on three orthogonal planes (MBSIF-TOP) is effective in detecting spoofing attacks, showing promising performance compared to existing alternatives.Next, by combining MBSIF-TOP with a blur-tolerant descriptor, namely the dynamic multiscale local phase quantization representation (MLPQ-TOP), the robustness of the spoofing attack detector can be further improved. The fusion of the information provided by MSIF-TOP and MLPQ-TOP is realized via a kernel fusion approach based on a fast kernel discriminant analysis (KDA) technique. It avoids the costly eigenanalysis computations by solving the KDA problem via spectral regression. The experimental evaluation of the proposed system on different databases demonstrates its advantages in detecting spoofing attacks in various imaging conditions, compared to existing methods.

Secure Face Unlock: Spoof Detection on Smartphones

“…On the CASIA FASD database using the original protocol, the proposed approach achieves 5.88% EER, which is smaller than the EER reported in several other publications (6.20% [14], 7.2% [15], 12.9% [20], 14.0% [23]). On the MSU-MFSD using the original protocol, the proposed approached achieves 8.41% EER which is slightly larger than the approach in [20] (5.82%).…”

“…Idiap Replay-attack (Intra-DB, Cross-DB) [11]: (15.54%, 47.1%); [13]: (0.8%, n/a) [14]: (2.9%, 16.7%); [15]: (1.0%, n/a) CASIA FASD (Intra-DB, Cross-DB) [14]: (6.2%, 37.6%); [15]: (7.2%, 30.2% EER) Face 3D shape or depth analysis [9], [16]- [18] Effective for 2D attacks Requires multiple frames or additional devices…”

Face Liveness Detection Using Dynamic Local Ternary Pattern (DLTP)

“…The results showed better performance [22] on intra-database and cross-database as compared to the state-of-the art. Similarly, the strength of dynamic texture for face liveness detection in a video sequence was proposed in [23]. The method based on kernel discriminant analysis and utilized a multi dynamic texture descriptor based on binarized statistical image features of three orthogonal planes (MBSIF-TOP) to detect the face spoof attacks.…”

A comprehensive overview of biometric fusion

“…Description 2012 Marasco et al [12] Different frameworks for integrating a spoof detection module with a recognition system 2015 Wen et al [154] Ensemble of SVMs on reflection, blurriness, chromatic moment, and color diversity 2015 Raghavendra et al [155] Feature level concatenation with Light Field Camera based features 2015 Arashloo et al [156] Fused MBSIF-TOP and MLPQ-TOP using SR-KDA 2016 Ding et al [89] Bayesian Belief Networks for fusing match scores with liveness scores 2016 Boulkenafet et al [157] CoALBP and LPQ features in HSV and YCbCr colour space 2016 Patel et al [158] Concatenation of color moments and LBP features 2016 Siddiqui et al [159] Inter-feature and intra-feature score-level fusion of multi-scale LBP and HOOF features 2016 Ding and Ross [160] Fusion of multiple one-class SVMs to improve generalizability of a fingerprint spoof detector 2017 Toosi et al [161] Comparative study of different fusion techniques on ten fingerprint features 2017 Korshunov and Marcel [162] Studies impact of score fusion on presentation attack detection for voice 2018 Komeili et al [163] Fusion of ECG recognition and fingerprint spoof detection 2018 Yadav et al [164] Fusion of (VGG features+PCA) with (RDWT+Haralick) features and neural network 2018 Sajjad et al [165] Two-tier authentication system for recognition and spoof detection 2018 Chugh et al [166] CNN based spoof detection on fingerprint patches fused a CNN with RNN in order to extract pseudo-depth images and a remote photoplethysmography (RPPG) signal from an input face video. The extracted information were then fused for face anti-spoofing.…”

“…[156] proposed fusing Multiscale Binarized Statistical Image Features on Three Orthogonal Planes (MBSIF-TOP) and Multiscale Local Phase Quantization on Three Orthogonal Planes (MLPQ-TOP) for performing spoof detection. Fusion was performed by a kernel fusion approach, termed as Spectral Regression Kernel Discriminant Analysis (SR-KDA).…”