A Self-embedding Watermarking Algorithm Based On Block Classification And Coding

With the rapid development of the social information,the process of the electronic documents replacing the paper documents is accelerating gradually.Compared with the traditional paper invoice,the e-invoice can reduce the cost without paper printing.The e-invoice is more convenient in the preservation and access,which helps to reduce the waste of the paper invoice.However more and more criminal people achieve a certain illegal purpose by tampered the e-invoices.This paper designs a self-embedding watermarking algorithm based on block classification coding by using digital watermarking,which can realize the authentication of the e-invoice image authenticity and the recovery of the tampered content.At the same time,a self-embedding watermarking algorithm based on block classification coding and SPIHT compression is proposed for natural images.Firstly,this paper introduces the research significance and background,including the digital watermarking technology.At the same time,the research status of the self-embedding algorithm is analyzed and summarized by introducing the different ways to get the recovery watermark.Then the typical algorithms are analyzed and simulated to conclude the common problems existing in the self-embedding watermarking algorithms,which provides the foundation of the new algorithm.Considering the watermarking capacity and the recovery effect of the tampered image,a self-embedding watermarking algorithm based on block classification coding and SPIHT compression is proposed.According to the characteristics of the natural image block,the texture block and the smooth block generates adaptively the variable length of the recovery watermark information respectively,which can reduce the watermark capacity and improve the quality of the recovered image with a better detection capability.Selecting an appropriate size of the image block with a different bit rate of the smooth block and the texture block in SPIHT compression,the compressed code uses RS coding to generate the watermark information.The generated recovery watermark is randomly embedded in the smooth block and the texture block based on the key.Compared with the existing newest watermarking algorithms,the watermark capacity between 1.16?2.35bpp is smaller and the recovery effect which is about 40dB is improved with a better detection capability.Aiming to detect and recovery the e-invoice of the special purpose image,a self-embedding watermarking algorithm based on block classification coding is proposed.Taking the Y component of the e-invoice image and dividing it into 8 X 8 image blocks,all image blocks in the test e-invoice images are divided into four types:the important block,seal one,format one and blank one according to the characteristic of the e-invoice image.The recovery watermark is generated according to the image block types and the non-blank block binary information which are respectively coded by RS coding.Depending on the size of the watermark information,the watermark is randomly embedded in the seal block,the format one and the blank one of the image in sequence based on the key.To reduce the watermark capacity to the minimum,the recovery watermark using the type code and the binary information code can simultaneously complete the authentication certification and the image recovery without adding authentication watermark data.Compared with the existing newest self-recovery literature of natural images,this work can get a smaller watermark capacity of about 0.48bpp with a better ability of the tampered detection and recovery in both the general tampered attack and the collage attack.The quality of the recovery image is around 40dB.Finally,the algorithm simulation system is designed by MATLAB and the self-embedding watermarking algorithm proposed in this paper is simulated and verified.