Cropping resilient watermarking based on histogram modification

This paper proposes a watermarking scheme robust against most common geometric and signal processing operations, for applications that require an accurate detection of the owner watermark even if the digital image suffers intentional and non-intentional attacks. The proposed scheme is based on two modifications introduced in two 2D histograms. In the first modification a selected region of a 2D histogram, composed by red and green (R-G) color components, is modified according to the watermark bit sequence; while in the second modification another 2D histogram, composed by blue (B) and filtered red (R) components, is partitioned into several blocks to embed the watermark data bits. The experimental results show fairly good robustness against several geometric distortions, common signal processing operations and some quite aggressive combined attacks. A watermark robustness comparison with the related works and an analysis of the embedding method from the watermark robustness point of view are provided, in order to explain the robustness improvement of the proposed algorithm, especially against rotation and cropping attacks. The ROC curves also provided to show the desirable detection performance of the proposed method. Comparisons with the previously reported methods, based on different techniques, are also provided.