Reference no: EM131303

Fingerprint watermarking techniques

In this, we need to investigate the best method to embed watermark image into fingerprint image. We need to compare between high, middle and lower frequency coefficients and find the best band that will be modified. The aim of this work is protect fingerprint image using watermarking techniques. The embedded watermark must be invisible and robustness. In this work, the minutiae points are used to determine the fingerprint uniqueness. Based on this idea, our work can go through two different aspects:

First aspect, the watermark in some frequency levels (high, middle, low) would affect minutiae points. In this case, the watermarked fingerprint image will not match the original fingerprint image. So, we need to remove (not just extract because there is different between extraction and removal) the watermark from the watermarked image before matching process. For this aspect, we can go ahead and find advantages of this work that could be a novel work. However, we need first to study carefully which frequency level that we should use.

Second aspect, the watermark does not affect the fingerprint minutiae points. In this case, we need to find an appropriate way to embed the watermark into fingerprint image without corrupting fingerprint features. As a result, positive matching will be generated between watermarked and original fingerprint image. Only extraction process is required at the server end. In this case, also we can make good contribution and publish our work.

For both aspects, the original image is not required at the server side to extract watermark. Again, for both aspects, we need to compare our work with other good published work and claim our novel algorithm performs very well.

For this work, implementation using Matlab and journal writing is required. The code must work on Matlab. Journal paper is not in the same level of conference paper, so I need high level of work and writing. Writing must be as a journal paper not homework level.