Fuzzy local ternary pattern and skin texture properties based countermeasure against face spoofing in biometric systems

Abstract

Evaluation of facial skin texture properties is very demanding and noticeable due to its optimum resource constraint along with reduced processing expenditure. Traditionally, the local binary pattern (LBP) and the variations are accepted for analyzing the skin for liveness detection. LBP descriptor, which is the most dominant algorithm, has its own limitations of not working excellent on images with noise. Texture properties contain variations in skin features like wrinkles. LBP organizes the feature extraction for specified texture to extract the variation between normal and abnormal. In applications such as surveillance, the images are captured by remote cameras and transmitted to control room. The images shall be affected by transmission noises. Transmission noises are varied under the transmission of texture features quality, contrast based pixel coordinates. Thus, in the proposed work, the effectiveness of fuzzy local ternary pattern (FLTP) as a substitute LBP and LTP is presented. It is used to find uncertainty principles of texture image, which are trained from the local databases. Moreover, FLTP embraces Weber s law that fuses the features to wipe out the predetermined threshold setting in LTP. The proposed approach was analyzed on various publicly available databases pertaining to this domain. Furthermore the proposed system was tested with University Putra Malaysia (UPM) face spoof database. The outcomes obtained from the projected FLTP texture descriptor achieved most proper exactness and outflanked the LBP and LTP surface descriptors.