The Self-embedding Watermarking And Its Application In Image Tamper Detection And Recovery

The development of digital multimedia technology has provided many mature,reliable,flexible,efficient,high quality and low cost multimedia information for our work and life.With the development ofvarious image processing technologies,the acquisition,processing,storage and duplication of digital products become more and more easy,and the copyright protection,integrity protection and tampering protection of digital products become more and more important.The rapid development ofinformation technology promotes the development of digital multimedia technology.In this context,with the development ofcomputer technology and various hand-held mobile intelligent terminals,multimedia data can be easily tampered with.Software installed in mobile phones,computers and tablets can tamper with multimedia data effortlessly.At the same time,it also brings problems such as illegal copying and redistribution of digital multimedia.The concept of digital watermarking is proposed to solve the problems related to copyright protection of information technology.It provides an effective solution for multimedia data authentication,copyright protection and tamper-proof.We take Image Watermarking as the main research object in this paper,discusses the relevant theories and algorithms of image watermarking,and propose three effective watermarking algorithms on this basis,which are applied to image tampering detection and restoration,and achieves good results.The main research results of this paper are summarized as follows:Firstly,an improved self-embedding digital watermarking algorithm based on hash algorithm is proposed.Based on the analysis of traditional hashing algorithm,an improved hashing algorithm is proposed,which embeds the hash value of the image itself into the original image as a watermarking for copyright protection of the image.The specific process includes three parts:pretreatment,eigenvalue extraction and hash value generation.The first step is to normalize the image by preprocessing.The second step is eigenvalue extraction.After image preprocessing,block truncation coding and central-symmetrical local binary patter are used to extract the eigenvalues of the image,and finally the image hash value is generated.The generated hash value is embedded into the original image by DCT transform domain method as a watermarking.The algorithm achieves a good balance between image perception and anti-attack,has strong robustness and good confidentiality performance.Secondly,an improved watermarking algorithm based on singular value decomposition is proposed.The eigenvalues of the image obtained by SVD are embedded into the original image as authentication watermarking for copyright protection and anti-attack.The authentication watermarking can effectively locate the tampered area of the image.At the same time,block classification algorithm is used to generate image restoration watermarking.To restore the watermarking image,it is necessary to preserve all the information of the image block as completely as possible.However,as a watermarking embedded image,there should not be too much data,otherwise,the condition of human visual threshold can not be satisfied.Therefore,on the basis of image DCT transform,according to the characteristics of image blocks,the images are divided into different categories.The compression coding of different types of image blocks is obtained to generate recovered watermarking,and the recovered watermarking of each type of image blocks are different in size.The watermarking algorithm can detect and restore the tampered image.The watermarking includes two parts:one is authentication watermarking for tamper detection and location of the image,and the other is restoration watermarking for tampered area restoration of the image.Chaotic encryption algorithm is used to embed watermarking into image blocks.If the watermarking image is tampered with,the basic information of the tampered image block will not match the extracted watermarking information.This method can be used to determine whether each image block has been tampered with.The recovered watermarking generated by the compression coding of each image block can be used to recover the tampered image block.The experimental results show that the algorithm can effectively resist image tampering,detect tampered areas,and restore the tampered blocks.Even when the tampering rate is as high as 70%,the tampered area can be located effectively and most of the tampered image blocks can be recovered.Thirdly,a watermarking algorithm for color image is proposed,which realizes tampering detection and restoration of color image based on self-embedding watermarking algorithm.In the first step,the color image is transformed into HSI and YCbCr.The eigenvalues are extracted as the hash value of the color image,which are generated as authentication bits.In the second step,on the basis of image DCT transform,the compressed coding of image blocks is obtained to generate recovery bits.In the third step,authentication bits and recovery bits are embedded into image blocks as watermarks.The basic information of the original image block is transformed into authentication bits by hash transform as part of the watermarking.According to comparing the basic information of the tampered image block with the extracted watermarking information,which image block has been tampered with can judge.The compressed coding of each image block as recovery bits are embedded in the original image.Once the original image is tampered with,these image block compression codes as reco'very bits can be used to recover the tampered image.The experimental results show that the algorithm can effectively resist image tampering,detect tampered areas,and restore the tampered blocks.It is an effective color image watermarking algorithm.