Research On Watermarking Algorithm Based On Stereoscopic Image

With the development of multimedia technology,the methods people get information are constantly increasing,the amount of information obtained from the traditional single view image cannot satisfy the consumers' expectations of visual enjoyment.Stereoscopic images,which are based on binocular disparity,combine information from different angles and directions,giving people a three-dimensional effect and enhance the perception of the scene,so the scope of application of stereo image are gradually expanding.But the stereoscopic image is faced with the problem of stealing and tampering in the process of using and transmission,therefore,effective copyright protection is need,digital watermarking is a very effective working method.But there are few algorithms designed for stereoscopic images,and the existing methods cannot fully take into account the resistance to conventional attacks and geometric attacks,and ensure that the quality of the image is not affected.So the suitable watermarking algorithms for stereoscopic image are important.In order to enrich the watermarking algorithms for stereoscopic images and improve the performance of the watermarking algorithm,this paper thoroughly studied the characteristics of stereoscopic images,first,a zero watermarking algorithm for stereoscopic images based on nonsubsampled contourlet transform(NSCT)is proposed.The watermark is embedded according to the relationship between the corresponding properties of the two viewpoints.After NSCT,the left and right view images have good frequency domain characteristics and have higher resistance to image attacks.We extract the low frequency subband containing the main information of the image as the carrier for the watermark extraction,and the redundancy of the NSCT ensures the embedding capacity of the watermark.The zero watermarking algorithm will not affect the image quality,and the contradiction between the robustness and the transparency of the watermarking algorithm is solved.When constructing the zero watermark,calculating the energy of discrete cosine transform coefficient matrix of the left and right view image blocks,and set up the standard to classify and mark each image block.The zero watermark is constructed by comparing the classification relationship of the left and right views.The verification watermark image is generated by applying the XOR operation on the zero watermark image and the logo watermark image,then register in the trusted authority.Experiments show that the proposed watermarking scheme is robust to all kinds of image attacks,and is an effective watermarking scheme.The experiment results show that the proposed watermarking scheme is robust to all kinds of image attacks,and can clearly extract the publishing right watermark image.Compared with the existing watermarking algorithms,we improve the performance of the watermarking algorithm.Based on the viewpoint watermarking algorithm,this paper deeply studied the inner relationship between the left and right viewpoint images of stereoscopic images,and proposed a disparity zero watermarking scheme based on NSCT and coding.According to the stereo image matching algorithm,we get the disparity image in the horizontal direction of the left and right viewpoints and use the stability of disparity image in the NSCT transform domain as the carrier of constructing the watermark.The disparity image is divided into blocks and the singular value decomposition is applied to each disparity image block respectively,singular value of the image block has good robustness to image interference,so we design an encoding rule to encode the relationship between the singular value of each image block,and enhance the security and stability of the algorithm.The experimental results proved that the watermarking algorithm we proposed can maintain good robustness and transparency under various attacks,the ability to resist attack has been improved compared with other algorithms.Based on this watermarking algorithm,in order to improve the ability of the algorithm against geometric attacks,we first using the SIFT feature point matching and save the scale descriptors as templates before embedding steps.And before the watermark is extracted,we correct the attacked image first to restore the synchronization between the watermark and the image.Experiment results show that the improved algorithm effectively improves the ability of watermark to resist geometric attacks,and can extract the watermark image clearly and integrally,and solves the problem that watermarking algorithm cannot resist both conventional attacks and geometric attacks.The watermarking algorithm proposed in this paper enriches the research of stereo image watermarking algorithm,and the three algorithms are all blind.Extracting watermark needs no original carrier image,which is beneficial for practical application.