Digital Watermarking Algorithm Of Color Image Based On DWT

Digital watermarking is an effective technique to protect the copyright of digital media products transmitted over the network. It embeds certain amount of important information into the raw data of digital products. It can not only protect the copyright, but also can verify the content integrity of digital products, which makes it a hot research topic in the field of multimedia information security.This thesis researches on the digital watermarking techniques in wavelet domain for the copyright protection of digital images. After a brief introduction of digital watermarking techniques, two wavelet-domain watermarking approaches are proposed to overcome the existed problems in the copyright protection of digital color images.First, a region-of-interest (ROI) and chaotic mapping based multi-channel watermarking algorithm for digital color image is proposed. It utilizes the sensitivity to initial value and non-Periodicity of one-dimensional chaotic sequence to encrypt the watermark image. Watermark is embedded into the multiple channels (R, G, B) of color images by wavelet transform and taking the role of human visual system (HVS) into consideration. Experimental results show that the proposed approach is robust to common attacks such as JPEG compression, noise addition, cropping and filtering.Second, a color image watermarking algorithm is proposed based on the relationship among the discrete wavelet transform (DWT) coefficients of digital image. After three-level wavelet decomposition, the low-frequency coefficients gather most of the energies. For watermark pre-processing, the initial value of chaotic systems is obtained from those low-frequency coefficients, and then the chaotic sequence is bounded with the image content itself. The chaotic sequence is used to scramble the watermark image, which improves the security of watermark. For watermark embedding, the relationship of energy and position between the low-frequency and the high-frequency DWT coefficients are utilized. Experimental results show that the proposed approach can obtain strong robustness, satisfactory transparency and blind detection.The proposed two approaches achieve the desired goals, which can use as references for the copyright protection of digital color images.