Segmentation-Based Watermarking Scheme For Image Tamper Detection And Recovery

The rapid advancement of multimedia technology and powerful image processing tools bring new problems to information security, which cannot be resolved by traditional encryption. Illegal alteration and tampering will cause critical harm to society and people, so how to protect the integrity and authenticity of multimedia content is becoming an urgent issue. The technology of digital watermark, which can overcome the limitations of traditional authentication methods, gives a promising solution to this problem, and becomes very hot in the digital image authentication research field.This thesis proposes a segmentation-based watermarking scheme for image tamper detection and restoration. Using the mechanism of image segmentation, an image is partitioned into irregular and interrelated regions, with the purpose of removing the block-wise independence to resist VQ attack. The authentication watermark which is produced by a feedback-based chaotic system has the capability of distinguishing the tampered regions from original image. And the recovery watermark, which is a highly compressed version of the host image, is used to recover the approximation of the tampered regions. Experimental results show that the proposed scheme succeeds in detecting and rebuilding the tampered regions with a high accuracy..Furthermore, this thesis introduces a reference sharing mechanism to improve the aforementioned algorithm. The recovery watermark to be embedded is a reference derived from the original principal content in different regions and shared by these regions for content restoration. Since the reference bit streams are scattered and dispersedly embedded into the entire image, local replacement of the content cannot completely destroy them, and the tampering coincidence problem is avoided. After identifying the tampered blocks, both the reference data and the original content in the reserved area are used to recover the principal content of the tampered area. Simulation results show that the algorithm can almost recover the content of missed portion with a high restoration quality.