Research On High Fidelity And Robustness Of Reversible Watermarking

The recent advances in computer network technologies boosted the use of data transfer on an exponential scale.Such an environment has allowed for the realization of many fascinating digital multimedia applications related to nearly all aspects of life.However,digital products can be easily copied,processed for various purposes.Therefore,protecting the multimedia content against unauthorized access or unauthorized distribution becomes an important issue.Reversible watermarking technique has been indicated so far as a possible solution for copyright protection and data authentication especially in some applications,such as law-enforcement,medical images processing and military.It averts illegal and malevolent copying and dissemination of digital images by hiding the unremarkable ownership data in the host image.Difference expansion and histogram shifting based reversible watermarking schemes are the two main technologies in the spatial domain,but both are suffered from the low embedding capacity and image degradation.This dissertation is focused on the research on high fidelity and robustness of image reversible digital watermarking.Three high fidelity reversible watermarking schemes with high,medium and low capacity are proposed to achieve better embedding performance,and the quantization based reversible watermarking scheme is presented to provide better robustness and perceptual quality.The main research contents and innovations are as follows:(1)Reversible watermarking using dynamic block-partition strategy and pixel-valueordering.In existing PVO-based algorithms,data embedding is implemented on pixel blocks that are partitioned according to the pixel coordinates of the image.This partition manner is far from efficient especially in the texture regions.To achieve the improved block redundancy mining and the trade-off between fidelity and capacity,a noval method to incorporate the dynamic block-partition strategy into PVO algorithm is proposed.Firstly,the host image is divided into non-overlapping 4×4 blocks according to the eight-neighboringpixels search surrounding the maximum pixel value.And then each region is partitioned into different blocks,which are subsequently classified according to the local complexity and predefined threshold values for embedding.Different embedding methods should be assigned to different types of blocks.More specifically,a normal block can carry 2 bits information and a smooth block can be further divided into four sized blocks to carry 8 bits information,while a rough block remains unchanged with no data embedded.In this way,the distribution of pixel values in each block is narrowed down to a smaller range and the inherent correlation among pixels is further enhanced.Meanwhile,the smooth blocks can embed high capacity while preserve good visual quality.(2)Reversible watermarking based on pixel-value-ordering and pixel block merging strategy.Traditional pixel-value-ordering methods utilize pixel blocks with a fixed size to embed data;the smaller the pixel blocks,greater is the embedding capacity.However,it tends to result in the deterioration of the quality of the marked image.Herein,a novel reversible watermarking method is presented by combining a block merging strategy with pixel-value-ordering method,which realizes the dynamic control of block size by considering the image texture.First,the cover image is divided into non-overlapping 2 × 2 pixel blocks.Subsequently,according to their complexity,similarity,and thresholds,these blocks are employed for data embedding through the pixel-value-ordering method directly or after being emerged into 2 × 4,4 × 2,or 4 × 4 sized blocks.Hence,smaller blocks can be used in the smooth region to create a high embedding capacity and larger blocks in the texture region to maintain a high peak signal-to-noise ratio.Compared with the traditional pixel-value-ordering method,the proposed method could well exploit the advantages of block correlation by adaptively selecting blocks of different sizes according to the image characteristics,hence demonstrating excellent performance.(3)Two-layer reversible watermarking algorithm using difference expansion.In traditional difference expand technology,the difference value of two pixels in a pair is expanded to carry one data bit.This method suffered from low embedding rate and image degradation.To address this drawback,a novel two-layer embedding algorithm is introduced to extend the difference expand technology from three aspects: the pixel pair construction,embedded pair selection and embedding manner.In this work,orthogonal pairing manner is adopted in different embedding layer.Furthermore,the variance of pixel pairs is employed to select smooth pixel pairs for embedding.The proposed method can provide an embedding rate up to 1bpp and thus the visual quality of marked image is guaranteed even when the embedding rate exceeds 0.5 bpp.(4)Quantization-based robust reversible digital watermarking.Channel noise and lossy compression should be considered as unintentional attacks.However,most of the proposed lossless data hiding schemes are fragile and the watermark is not robust enough for the actual applications.Taking into consideration the need for images that has the ability to withstand lossy compression(e.g,JPEG compression),a robust lossless digital watermarking scheme based on the quantization is discussed.In the proposed method,data bits are hidden into the cover image by a reversible generalized integer transform with quantization.Specifically,for a block of n pixels,the sum of n pixels is served as the statistical magnitude.Then the integer transformation parameters of the blocks are constructed according to the embedded watermark.In this quantization embedding manner,even multiple of the quantization step represents the embedded watermark bit 1 and odd multiple 0.At the decoding end,the watermark is extracted in line with the parity of the pixel sum,and then the cover image is recovered by inverse integer transformation.The increasing robustness of the algorithm and good visual quality in the presence of lossy compression is obtained through the results.The experimental results prove that,the proposed three high fidelity reversible watermarking schemes achieve a good trade-off between the embedding distoration and embedding capacity and the quantization based reversible watermarking scheme yields reversibility and better performance of robustness to JPEG compression.