Fingerprint image protection using two watermarks without corrupting minutiae

This paper proposes an approach for embedding two watermarks into fingerprint images using the Discrete Cosine Transform (DCT) algorithm. The main aim of the proposed algorithm is to add more authentication factors based on the watermark messages and to protect the ownership of the fingerprint image. Since the information used for identification or verification of a fingerprint image mainly lies in its minutiae, the introduced watermarking algorithm does not affect fingerprint features. First watermark is constructed based on a unique identification number that can identify the user. The hash function (SHA2) is applied to generate the hash value of the user identification number to encode the watermark pattern. Then, it is embedded into the fingerprint image while avoiding the minutiae locations. The second watermark is a gray image that is inserted into first watermarked image. The extraction stage does not require the original fingerprint image. We measure the affect of the watermarks on the fingerprint features based on the comparison between the total number of extracted minutiae points before and after the embedding process. The presented schema shows a high PSNR value. The robustness of the introduced technique has been tested against Gaussian and Salt & Pepper attacks.