Biometric Data Security Using Watermarking Based on Vector Quantization

Abstract

In this paper, we present a new watermarking method applied to biometric data for the purpose of securing their transmission over a computer network. To perform this task, we combined the fingerprint modality with that of the iris. We proceeded by merging the two biometric signatures using the watermarking based on vector quantization. The proposed technique extracts the main characteristics from the fingerprint image to mask the binary iris code image. This iris code is obtained from the eye image after several transformations. This choice is found to be the obvious combination solution for the type of the manipulated standard images. The vector quantization method was implemented based on Voronoi diagram using a codebook generated from a new chaotic system. The robustness of this technique was extensively tested using several simulation scenarios handling adequate database images. The results show that the proposed method is robust enough against JPEG compression. Other tests covered also different simulated computer attacks of the watermarked image using several types of median, mean and Gaussian blur filters. In this case, the filter sizes are taken large enough i.e., (10 × 10). The overall obtained results are satisfactory and very encouraging.