A Digital Watermarking Algorithm Based on Image Similarity Calculation

Abstract

Abstract-Normalized correlation algorithms have been used in digital watermarking evaluation system at home and abroad which still exists some shortcomings. In this paper, several common image similarity algorithms in image retrieval technology are selected and applied to watermarking technology. This paper chooses an adaptive embedding and extraction algorithm based on discrete cosine transform (DCT) and uses some different methods to calculate the similarity between the extracted watermark images and the original watermark image. Before embedding, the watermark image has been scrambled. The entire process of image watermarking system is conducted by Matlab, which demonstrates that applying the similarity calculating methods used in image retrieval into digital watermarking technology can improve the accuracy when evaluate the robustness of a watermark system.