Robust Biometric Watermarking using Image Intrinsic Local Property for e-Database Applications

Mita Paunwala and Suprava Patnaik


Biometrics, fingerprint, matching, watermarking


This paper addresses a biometric watermarking technology sturdy towards image manipulations, like JPEG compression, image filtering, and additive noise. Application scenarios include information transmission between client and server, maintaining e-database and management of signatures through insecure distribution channels. Steps involved in this work are, a) generation of binary signature code for biometric, b) embedding of the binary signature to the host image using intrinsic local property, that ensures signature protection, c) host image is then made exposed to various attacks and d) signature is extracted and matched based on an empirical threshold to verify the robustness of proposed embedding method. Embedding relies on binary signature manipulating the lower order AC coefficients of Discrete Cosine Transformed sub-blocks of host image. The proposed algorithm is tested for 50 different types of host images and public data collection, DB3, FVC2002. FAR and FRR are compared with other methods to show the improvement.

Important Links:

Go Back