Image Tampering Detection Using Compressive Sensing Based Watermarking Scheme

In this paper, we describe an image watermarking scheme using compressive sensing for detection of image tampering. We exploit dimensionality reduction and inherent computational security offered by compressive sensing. The frequency band containing the high energy is used to compute the hash and then its pseudo-random projections are obtained. These projections are embedded as a watermark into the frequency band containing low energy. On recovering the watermark, projections are acquired, which in turn are used to solve convex optimization problem to obtain the hash. The image tampering can be detected and integrity of watermark can be established by comparing the hash values. Index Terms – Compressive sensing, convex optimization image authentication, watermarking