Face Spoofing Detection Techniques using Biometrics

Abstract

The modern biometric technologies provides us better convenience and security features. Face Recognition Biometric systems are being used and deployed in applications such as surveillance, forensic investigation etc., but it is vulnerable mostly in case of face spoofing attacks. Such spoofing can be done by means of video frames, printed photo. To detect these types of attacks, the liveness of face detection is being developed, and also being deployed in face recognition biometric systems. If these methods don't exist in the face recognition biometric systems, it may give permission to a malicious person to masquerade as authentic users to the data file system. To address these problems, it's important to develop a secure biometric recognition system. The current method and approach to detect the liveness within the facial biometrics by making use of the feature extraction methods, includes Local Binary Pattern (LBP), Color Moment Features (CMF). In the proposed system combining two or three features proposed mainly, Histogram of Oriented Gradients (HOG), Spectral Information Divergence (SID), Binarized Statistical Image Features (BSIF), Weber Local Descriptor (WLD) and Local Phase Quantization (LPQ). Support Vector Machine (SVM) classifier gives the result as whether the image is spoofed or real. Done detailed survey on face spoof detection methods, feature methods and algorithms that are existing today and being used for the detection of spoof images. Based on the facts gathered, the execution with minimum and simple use of hardware makes biometric systems better secured and robust.