Improved Capacity Reversible Watermarking

This paper continues our researches on high capacity reversible watermarking based on simple transforms. Image pixels obeying some simple constraints are transformed and then, data is embedded by simple additions. The transform induces a congruence equation. At detection, the transformed/not transformed pixels are found by checking the congruence, watermark codes are extracted and original image is recovered. The proposed method is of very low mathematical complexity and does not need additional data compression. The major novelty of the proposed technique is the use of a different transform which allows the embedding of watermark codewords into each transformed pixel. Thus, the improved version provides considerably higher embedding bit-rate. Experimental results for a single embedding level are presented.