Robust Blind Watermarking Technique Against Geometric Attacks for Fingerprint Image Using DTCWT-DCT

Abstract

In this research paper, a new blind and robust fingerprint image watermarking scheme based on a combination of dual-tree complex wavelet transform (DTCWT) and discrete cosine transform (DCT) domains is demonstrated. The major concern is to afford a solution in reducing the consequence of geometric attacks. It is due to the fingerprint features that may be impacted by the incorporated watermark, fingerprint rotations, and displacements that result in multiple feature sets. To integrate the bits of the watermark sequence into a differential process, two DCT-transformed sub-vectors are implemented. The initial sub-vectors were obtained by sub-sampling in the host fingerprint image of both real and imaginary parts of the DTCWT wavelet coefficients. The basic difference between the relevant sub-vectors of the watermarked fingerprint image in the extraction stage directly provides the inserted watermark sequence. It is not necessary to extract watermark data from an original fingerprint image. Therefore, the technique suggested is evaluated using 80 fingerprint images from 10 persons, from both CASIA-V5-DB and FVC2002-DB2 fingerprint database. For each person, eight fingerprints are set as the template and the watermark are inserted in each image. A comparison between the obtained results with other geometric robust techniques results is performed afterwards. The comparison results show that the proposed technique has stronger robustness against common image processing processes and geometric attacks such as cropping, resizing, and rotation.