Research On Digital Image Watermarking Based On Chaotic Sequences And DFT

The rapid development of digital services and computer networks in modern days makesimage distribution more convenient than ever. The digital representation of copyrightedimages certainly offers many advantages, but also enables illegal use of digital images, whichposes a serious threat to the rights of content owners, causing a pressing need for copyrightprotection and multimedia security. The conventional cryptographic system can preventillegal access to the encrypted data without a key, but once the data is decrypted, there is noway to track its reproduction and redistribution. One solution to this problem is digitalwatermarking technique, which embeds information about the copyright owner permanentlyin the data and remains present within the data after any decryption process.Nowadays, most researches on watermarking focus on image watermarking, and thegeneral requirements are robustness and invisibility. Achieving balance between robustnessand invisibility has always being an important issue in watermarking research, which is alsothe main object of study in this paper. The watermark should not only be robust to a certainlevel of geometric distortions, but also maintain the value of PSNR above40dB at the sametime.In this report, a watermarking method for grayscale images is proposed that is invisibleand robust to certain attacks. Chaotic maps are used to generate the watermark whichimproves the security of the method. The crown watermark is embedded in the real part ofDFT domain and the embedding position is determined by the SURF algorithm. The peaksignal-to-noise ratio is used to evaluate the perceived quality of the marked image.Normalized cross correlation is used for watermark detection. The original image is notrequired during the detection. Experiments are conducted to evaluate the robustness of theproposed method against different attacks on several images.