Invisible Digital Watermarking Algorithm And Its Transparency Evaluation

As a useful complement to the traditional copyright protection technology, digital watermarking has been studied widely by researchers at home and abroad. According to the characteristics of visual perception, digital watermarking can be divided as visible watermark and invisible watermark. The invisible watermark has two basic features, i.e. robustness and imperceptibility, which are always conflict. How to achieve a better balance is very important in invisible watermarking. Additionally, since performance evaluation of imperceptibility has been paid less attention to, it is necessary to propose some evaluation method consistent with human visual system. This paper focuses on invisible digital watermarking algorithms and imperceptibility evaluation of watermark, the main work and innovations are as follows:(1) A novel watermarking method based on Cuckoo Search Algorithm and discrete wavelet transform is suggested. In this method, the third-level horizontal detail sub-image obtained by applying3-Level Wavelet Transform to the cover image is selected as the embedded position, and the watermark is scrambled with Arnold Transform and changed into a one-dimensional bit stream. Then the Cuckoo Search Algorithm is employed to determine the best embedding coefficient. Experimental results show that this method is more robust than some existing methods based on swarm intelligence in preserving the imperceptibility.(2) A new watermarking method based on singular value decomposition (SVD) and discrete wavelet transform is proposed. In this method, after the original cover image is decomposed by a2-Level wavelet transform, the SVD technique is employed to the matrix composed by the second detail coefficients to acquire the embedded positions. Simultaneously, the original binary image watermark is scrambled with Arnold transform and then is changed into a one-dimensional bit stream. Hence, according to the different combinations of the bit stream and the elements in the singular value matrix, the bit stream of watermark is embedded into the cover image. Experimental results show that compared with some existing methods based on SVD, this method has better performance when suffering from any attack such as noise-disturbing, filtering, compressing, rotating, rescaling, cropping etc. (3) An imperceptibility evaluation method is presented based on grey relational analysis and discrete wavelet transform. In this method, after the original image and corresponding watermark-embedded images are decomposed by multilevel wavelet transform, their wavelet coefficients are divided into4categories in terms of the sensibility of human visual system, and each of them is used to construct a reference sequence and some comparative sequences. The similarities of the two kinds of sequences are calculated by grey relational analysis, which is further adjusted with the consideration of the watermark-embedded-area ratio and the watermark-embedded coefficients. A global performance of imperceptibility of watermark-embedded images is obtained after the4grey relational degrees are combined with different weights. Experimental results show that the proposed method not only is efficient and universal, but also gains better ability to distinguish tiny details than some other existing methods.