Soft Biometrics Authentication

This manuscript presents the design of a new approach human skin color authentication. Skin is one most popular soft biometric modalities. Since modality alone cannot reliably authenticate an individual, this system designed to combine results with other pure modalities increase recognition performance. In classification process, we first perform facial detection by segmentation using thresholding method in HSV space. Then, K-means algorithm clustering used determine dominant colors on pixels RGB model. Variations according R, G and B components are recorded reference model enable individual’s identity be predicted basis 30 clusters. Experimental promising give false acceptance rate (FAR) 29.47% rejection (FRR) 70.53%.