EFFICIENT GAIT VERIFICATION THROUGH TEMPLATE PRESERVATION

Abstract

In this paper, the effectiveness of biometric verification system has been achieved with template protection technique. Gait images of person are used; features are extracted from average gait images using Principle Component Analysis (PCA) for dimensionality reduction. At enrollment phase, three gait energy images (GEI)-upper images of person are used. The most reliable component for each person is needed to find and encoded it through Bose Chaudhuri Hocquenghem (BCH) encoding scheme to generate code word. A secrete binary string and hash value are generated for each person and stored in the database along with encoded template and position of strong reliable component. At verification phase, GEI-lower, slow and fast gait images are used. Those features are only selected from original features based on the position of strong reliable component and are used to generate code word and which is decoded using BCH decoder. The hash value is generated and compared with the hash value which was generated in the enrollment phase. If these two hash values are the same then the person is verified otherwise the person is not verified. Also comparison with other recently proposed studies is done to show the effectiveness of the proposed scheme.

Keywords

PCA, hash value, GEI.