Lip biometric template security framework using spatial steganography

Abstract

In this work, we have proposed an efficient and secured lip biometric framework. Unlike the traditional biometric frameworks, that focus on the recognition accuracy only, we focus on both recognition rate along with securing the templates stored in the biometric system. Our contribution also includes using a pre-processing step for improving the local features of the lip images. Local interest points detected by Scale Invariant Feature Transform (SIFT) are used for extracting the lip features. A spatial steganographic algorithm is employed on the lip images to ensure minimum distortion along with hiding the identity of the lip images in the images itself, thus ensuring less chance of misuse of the template. We have reported a comparative analysis of using our steganographic algorithm to secure the template management system to ensure that it does not hamper the recognition rate of the biometric system. We have validated our proposed framework on NITRLipV1 and NITRLipV2 comparing against state-of-the-art results which does not use identity hiding, and we have found the recognition along with hidden identity to yield equally satisfactory performance.