Novel Optimization of Identified Palm Geometry Using Image Segmentation

Abstract

Segmentation is one of the essential steps towards the identification of any object in the domain of image processing. In the area of hand-based biometric which is mainly deployed for a user authentication system, segmentation plays a critical role. A review of existing studies shows that there is a very less amount potential contribution in this regard. Therefore, this manuscript presents a novel optimization scheme towards palm geometry recognition system where segmentation process is the prime highlights for classification of hand and background considering a case study of finger recognition. Further, the proposed scheme uses masking operation where the Region-of-Interest section of hand is subjected to segmentation. Further proposed system uses machine learning approach (convolution neural network and Siamese Neural Network) to further assist in optimizing the segmentation performance. The experimental outcome of the study shows proposed system offers better accuracy compared to the existing system.