Font Size: a A A

Research On Image Authenticity Identification Technology Based On Digital Watermark

Posted on:2010-11-06Degree:MasterType:Thesis
Country:ChinaCandidate:J LiFull Text:PDF
GTID:2298360275970412Subject:Communication and Information System
Abstract/Summary:PDF Full Text Request
As the important part of the National 863 Plan-“Study on Authenticity Identification Technology for Digital Image Information”, the research content of this dissertation is the image initiative anti-counterfeiting authentication, which is to identify the image authenticity by watermarking technology, to provide a reliable means for the legal usage of image and expert testimony.This dissertation first systematically introduced the watermark technology that this study mainly relies on, about its model, classification, attacking methods, evaluation index etc. Then after fully analyzing of the social needs for image authenticity identification, the digital watermark technology for image authenticity identification is proposed, discussing the watermark scheme architecture, the key technology points and typical algorithms and performance evaluations.For the three requirements of identifying the image authenticity: a) the image integrality identification; b) the tamper location; c) original image information recovering, furthermore considered the current image formats, this article will perform the research work on following points:1. Lossy Compressed Image and Anti-counterfeiting Watermarking TechnologyConsidering the popular image formats-JPEG format based on DCT domain, JPEG2000 based on wavelet domain, a semi-fragile watermark scheme based on SVD (Singular Value Decomposition) domain with higher energy concentration is proposed. By taking advantage of the robustness of the max singular value against lossy compression, the SVD domain is a good carrier for watermark to keep strong robustness against both JPEG and JPEG2000 compression which is unreachable for most watermark schemes simultaneously. The algorithm is”Semi-fragile Watermark Scheme based on the Logistic Chaos Sequence and Singular Value Decomposition”. This algorithm discussed the property of the LOGISTIC chaotic sequence and SVD, and then analyzed the scheme effectiveness both theoretically and experimentally. The experiment showed that this scheme has good robustness under high ratio compression, meanwhile providing image tamper accurately detecting and locating.2. Original Image Recovering and Anti-counterfeiting Watermarking TechnologyTwo kinds of recoverable watermark schemes are proposed based on spatial and transformed domain respectively. By taking advantage of the chaotic sequence and the relative stability of the HSB value, the algorithm-“Semi-fragile watermark scheme based on HSB and LSB”can reach the pixel-level ability for tamper locating and the high quality recovering ability by embedding the watermark into LSB of image pixel. As realized the fact that the recoverable watermarking scheme under lossy compression is very rare, this article extends the SVD to the more general energy domain and proposed the assumption that as long as there is a mathematical transform providing higher energy concentration than DCT and wavelet domain, the new energy concentration domain is robust against the lossy JPEG and JPEG2000 compression and a good watermark carrier. Here a robust recoverable watermarking scheme is proposed based on the new energy concentration domain and the error correcting ability of convolution code. Compared with the traditional method, this scheme uses the error of convolution code to detect the tampered image, while the watermark is used to recover the original image. The theoretical analysis and experiment shows that this scheme can reach the accurate block-level detecting and locating ability for tampered image and roughly recovering ability.Finally, this dissertation draws the conclusion about the research methods and contents of the study, and then makes prospects.
Keywords/Search Tags:Digital Image, Authenticity Identification, Digital Watermark
PDF Full Text Request
Related items