Copyright Protection System Based On Digital Watermark And Image Texture

With the maturing of network technology and the rapid development of multimedia hardware technology,while facilitating people's lives,digital illegally duplicates and disseminates digital images because of their low cost and less information loss in network transmission Infringement cases.And with the continuous improvement of technology,a large number of high-fidelity,high-precision scanning equipment makes the paper printing products more common fake and shoddy.This paper studies the key technologies of copyright protection system based on digital watermarking and image texture features,design and implementation Anti-counterfeiting system.The common algorithms of image feature matching are studied,and the KL image correction algorithm is proposed based on the image feature matching algorithm.Firstly,the comprehensive robustness analysis of common feature extraction algorithms such as SIFT(Scale-invariant Feature Transform),SURF(SpeedUp Robust Feature)and ORB(Oriented Briefde)is provided,which provides the foundation for the selection of subsequent correction algorithms.In this paper,K-nearest neighbor and labor-based algorithm are used to improve the false matching points of SIFT image rectification algorithm,which effectively reduces the number of mismatched feature point pairs and improves the correctness of the perspective transformation matrix for correcting images.The semi-fragile watermarking algorithm based on discrete wavelet transform is studied to distinguish between printed images and copied images.The digital watermarking technology is analyzed and studied,and the principle and effect of wavelet transform in transform domain watermarking algorithm are analyzed emphatically.Based on the gain invariant quantization watermarking algorithm of Zareian M and Tohidypour H R,the dissertation proposes a semi-fragile watermarking algorithm based on discrete wavelet transform by changing the position of watermark embedding by discrete wavelet transform coefficient and changing the construction of watermarking embedding vector.The algorithm has good robustness to the printed image and has the fragility of the copied image,thus achieving the goal of distinguishing between the printed image and the copied image.Making the algorithm to achieve"anti-printing,anti-secondary printing" purposes.Finally,the performance of the watermarking algorithm is verified by experiments.Combined with image texture features and SVM(support vector machine)to study a watermark-free distinction between print plans and copy plans,which have a dual guarantee of anti-counterfeiting system.The paper analyzes and analyzes the change of the image texture features in the digital image printing and copying process,compares and analyzes the selected dimension texture features after printing and copying,and studies the differences among them.The texture features with obvious differences are selected and classified by SVM.Finally,the design and implementation of anti-counterfeiting system,the main design and implementation of the watermark embedding end,the model training end,watermarking and feature extraction end of the three main modules.And the watermark and feature extraction client ported to the Android platform,and thus practical application.In the migration process to solve Java and C/C++ mixed coding,SO file compatibility,application slimming,operating efficiency and other issues,which is the application has good performance.