Text Image Watermarking In Joint Transform Domain With Enhanced Encryption

Textual digital information is one of the most commonly used types of digital information.With the rapid development of computer multimedia and network technology,text images,or text documents in the form of images,such as documents,black and white pictures,signatures,contracts,and seals,have appeared.Text images are easy to collect,easy to spread,and low-cost.They have become the main form of text digital information dissemination.In order to verify that the text image is pirated,the text image watermarking technology for copyright protection has received wide attention.Algorithms for watermark encryption,embedding and extraction are the key to realize the copyright protection for text images.The watermark encryption algorithm uses scrambling encryption technology to convert the original watermark into unrecognizable chaotic information,and uses the key to reconstruct the watermark information during decryption.The role is to prevent illegal users from acquiring,identifying,restoring or tampering with watermark information,that is,to ensure the security of the watermark.At present,the watermark encryption algorithm usually adopts a single encryption algorithm.Once the encryption algorithm is disclosed,its encryption key is easily cracked,resulting in leakage of copyright information and low security.The watermark embedding and extraction algorithm adopts methods such as quantization and spreading to embed the watermark information into the spatial or transform domain of the text image for tracking and verification of copyright information.With the embedding and extraction algorithm of a single transform domain,the text image has different ability to resist various attacks with different transform domains embedded in the watermark,and it can not satisfy the robustness to all common attack types.The joint transform domain algorithm combines the advantages of different single transform domains to improve the invisibility and robustness of the watermarking algorithm.However,each transform domain algorithm focuses differently,and the compatibility against different attacks is weak,which is easy to be cut in the process of network propagation.The attack resistance,the filtering attack,the compression attack,and the noise attack are poorly resisted,and the accuracy of extracting the watermark is low.In view of the above problems,based on the analysis and summary of the theory,algorithm and technology of text image watermarking,this paper proposes a text image watermarking algorithm based on enhanced encryption on the joint transform domain to solve the above problems.In order to improve the security of text image watermarking algorithm,this paper proposes an enhanced encryption algorithm.Firstly,the Arnold scrambling encryption algorithm is used to encrypt the watermark image in one level,and the original meaningful watermark image is scrambled into a chaotic image with no obvious meaning.Then,the stable feature sequence is extracted from the text image,the chaotic image is secondarily encrypted,the encrypted watermark is obtained,the key is saved,and the risk that the watermark is recognized and maliciously falsified and erased is weakened.Combining the text image texture to present the characteristics of local uniform distribution,in order to improve the robustness of text image watermarking algorithm,this paper proposes a watermark embedding and extraction algorithm combining DWT-DCT joint transform and error control strategy.Using the error correction function of the error control strategy,the twice-encrypted watermark image is subjected to Hamming code error control coding to obtain a coded image.The DWT transforms the multi-scale,multi-resolution analysis characteristics and the DCT transform "energy concentration" and de-correlation ability,and performs DWT-DCT joint transformation on the text image to obtain a transform coefficient matrix.The redundant embedding algorithm is used to select the appropriate intermediate frequency coefficient to embed the coded image into the intermediate frequency region of the transform coefficient matrix,and reduce the influence of the watermark on the subjective vision of the text image.In the extraction process,the watermark information in multiple intermediate frequency regions is extracted for proofreading.Processing,increasing the accuracy of extracting watermarks;synthesizing the resistance and characteristics of DWT and DCT transform domain algorithms to common attacks,and improving the robustness of text image watermarking algorithms.In this paper,the above algorithm is implemented.The performance of the text image watermarking algorithm is analyzed by experimental results.The experimental results show that the proposed algorithm can improve the security of the watermarking algorithm and the robustness of the attackr as esisting JPEG compression,filtering,noise,clipping,translation,rotation etc.,under the premise of ensuring the invisibility of the watermark.