Research On Compressive Sensing Theory For Digital Image Watermarking Algorithm

With the improvement of communication network technology,the application of digital media is becoming more extensive.People's life and communication are enriched by the emerging of audio,digital images and video.However,digital media technology also arises a lot of information security issues when it provides convenience to people,e.g.malicious modification of information,attack and copyright theft,etc.To solved the security problems effectively that might be subjected by digital media information,researchers proposed digital watermark technology and had obtained some achievements.This paper takes digital image watermark technology as the object of the study,focusing on research of sparse representation theory and the blind watermarking algorithm combining Compressive Sensing theory(CS)with digital watermark technology.The main content and research results were as below:1)The research background,significance and development status of digital image watermark will be introduced briefly,at the same time,current digital watermark technology be compared and classified.Analyzing the advantages and characteristics of CS theory when it being used for digital image watermark technology,and provide some reference for the design of image watermark system.Based on the above mentioned content,we proposed the issues which need to be studied further and introduced the main content and chapters of this paper.2)For solving the issues of sparse representation theory which cause instability and high error rate when it is applied in watermark technology.We proposed the MGV-MP algorithm(minimum gray value matching pursuit)in this paper,which is based on the traditional MP algorithm(matching pursuit).MGV-MP algorithm can make sure that the carrier image be completely decomposed after finite iteration.In addition,it can avoid overflow or erasing of previous watermark information during watermark embedding and extraction process.Thus,the strong robustness can be provided for the application of sparse representation theory.3)The current color image watermark has some issues like singleness,instability and other problems.Therefore,we try to introduce the sparse representation theory into image watermark technology in the paper.First,K-SVD and learning optimization algorithm is used to obtain a binary redundancy dictionary to meet digital watermark technology.Then,sparse decomposition is conducted for the green and blue channels of the carrier image using the binary redundancy dictionary and MGV-MP algorithm.Finally,the watermark information is embedded by modifying the sparse coefficient.The experimental result shows that the algorithm we proposed having excellent performance and basically meet the demands for watermark application.Therefore,it has strong practical value.4)An image block watermark embedding and extraction algorithm is proposed in this paper which based on edge detection and CS theory,aiming at solving the invisibility and robustness problems of existing digital watermark algorithm.First,the original carrier image is divided into non-overlapping blocks by edge detection and blocking strategy.Then,target image blocks suitable for concealing of watermark information are selected in non-overlapping image blocks.Finally,the watermark information is encrypted and embedded into target image blocks by using singular value decomposition(SVD)and CS algorithm.The experimental result shows that the algorithm in the paper has excellent performance in invisibility and robustness.In addition,the algorithm is concise and has high safety.