Print Spoof Detection with HSI Space: Performance Evaluation with Color Scatter and GLCM Measures

The access control methods mainly employ physiological characteristics and the practical relevance to generate sophisticated spoofing attacks led to the development of liveness detection methods to increase the robustness of the identification systems. Among the largely deployed traits such as iris and fingerprint, the face is the commonly used modality in recognition systems due to its non-intrusive nature. The evolving technological progress has made the production of a wide variety of spoofing attacks such as Print attacks, Display attack, Replay, and 3D mask attacks possible and hence led to the development of various anti-spoofing methods. Here, a novel approach for face spoof detection addressing print spoof is presented. The novelty lies in distinct features derived from color scatter and variance measures on HSI color space. In an attempt to explore other features, GLCM-based measures are used. The volumetric scatter measures around the convex hull with the geometric description have yielded compact and effective features than GLCM measures and works well with naturally available acquisition devices. Spoof detection rate is highly significant with the color scatter and volumetric features and works well over inter-database protocols too.