Research On The Digital Image Watermarking Based On Wavelet Transform

With the advent of the digital information age, more and more digital products become popular in our Daily life. Although digital information technology of copying, and reproducing provides great convenience, meanwhile it brings some negative effects, such as the copyright theft and malicious tampering and illegal copying. As digital image is an important carrier of multimedia information, its copyright protection and authentication technology have important theoretical and practical value. Thus this paper conducts the research in this area and the contents are listed in the following(1) Firstly, it studies HAAR wavelet transform, Logistic chaos and Arnold scrambling in Image processing, and presents simulation results of HAAR wavelet transform, Logistic chaos and Arnold scrambling.(2) Based on non-watermark techniques, and the advantage of high energy of the original image after wavelet transform, this paper proposes a watermarking algorithm using the expandability of the copyright watermark technique and chaotic keys. Firstly, the algorithm encrypts the copyright image watermark by using chaos keys, then uses the stable features of the original image, and the characteristic matrix generated by the original image and copyright image watermark matrix with chaotic key encryption. Then, it makes some kinds of calculation for generating authentication matrix. Finally the authentication matrix, time stamp, and image information are saved to the watermark database for copyright authentication. Experiments show that the proposed scheme is robust against the conventional image attacks and the cutting attack.(3) According to the idea of embedded image authentication, this paper presents a new half fragile image watermarking algorithm based on HAAR Wavelet transform. Firstly, this algorithm determines the maximum value which the image can embed according to the needed peak signal-to-noise ratio value (PSNR) before and after image embedding. Then according to the maximum watermark capacity, the watermark image is embedded based on the HAAR wavelet-transform of the original image in the 2 * 2 low-frequency coefficients. Finally the embedded image content is extracted based on correlation comparison, detection images content integrity and authenticity. Experiments show that this algorithm is robust to image anti compress, noise, and filtering in conventional attacks, can accurately detect which region is tampered, and has the feasibility and practicability.