Spoofing detection on hand images using quality assessment

Recent research on biometrics focuses achieving a high success rate of authentication and addressing the concern various spoofing attacks. Although hand geometry recognition provides adequate security over unauthorized access, it is susceptible to presentation attack. This paper presents an anti-spoofing method toward biometrics. A attack detection approach addressed by assessing visual quality genuine fake images. threshold-based gradient magnitude similarity metric proposed discriminate between real spoofed samples. The images 255 subjects from Bogazici University database are considered as original Correspondingly, each sample, we acquire forged image using Canon EOS 700D camera. Such with natural degradation for electronic screen display based detection. Furthermore, create another dataset artificial introducing additional Gaussian blur, salt pepper, speckle noises Ten metrics measured sample classification image. experiments performed k-nearest neighbors, random forest, support vector machine classifiers, well deep convolutional neural networks. similarity-based achieves 1.5% average error neighbors forest classifiers. An 2.5% obtained baseline evaluation MobileNetV2 network discriminating different types