Securing fingerprint images using a hybrid technique

This paper presents an efficient hybrid watermarking scheme for securing fingerprint images in the Discrete Cosine Transform (DCT) domain. The proposed method uses a NN-PSO (Neural Network-Particle Swarm Optimization) based hybrid approach to watermark host gray scale fingerprint images with their corresponding facial images in order to secure them. The algorithm divides the input image into blocks and uses a feed forward NN to determine the number of secret bits to be embedded in each block depending on the blocks' features. The output of the NN is then used as input by the PSO module to find the best DCT coefficients' locations in that block where the secret facial image data can be embedded, so that the distortion produced in the host image is minimum and the minutia predicting ability remains unaffected. The experimental results have been analyzed and compared with existing PSO based and NN based approaches. The proposed NN-PSO based hybrid approach has been found to exhibit better watermarked image quality and better robustness to possible attacks to the watermarked image. Moreover, as the proposed technique retains the feature set of the original fingerprint, the extracted facial image and the cover fingerprint image can be correctly verified at the receiver's end leading to a more secure and accurate biometric based personal authentication.