| Summary: | 碩士 === 國立臺灣科技大學 === 電機工程系 === 96 === As the science and technology develop constantly in recent years, many related new theories and algorithms have been proposed in the digital image processing technique. It raises more and more attention in the development of the digital image. To meet the demand of technical development, the technique of image processing appears important.
Recently, the Internet grows rapidly. Digital multimedia information can easily be manipulated and tampered intentionally. So the intellectual property protection becomes more and more important. On the current development, there are many research proposed for watermarking, such as robust watermarking, fragile watermarking, semi-fragile watermarking, and tamper detection with recovery technique. The current image authentication researches are mostly extract watermarking or tamper detection and recovery, few of them are focusing on both of these two techniques. However, one of these techniques could not complete protect the image. Due to this reason, we propose a novel technique, not only can detect but also recover the tampering region. Embedding a digital watermark as authentication and also resists some basic attacks.
In this thesis, an image watermarking scheme with tamper detection and recovery is proposed. The main goal is to detect and recover the tampered region accurately. In addition, the proposed method is robust to against the cropping attack.
Embedding the suitable watermark in images can effectively prevent the malicious tampering to protect the primitive and integrity of the image. This can also provide the multi-purpose security service, the characteristics are as follows:
a) Undetectable:
The transparency of the embedded watermark in human vision is the most important issue. By embedding watermark, the “artifact” not only ill-flavored, but also possibly reduce or destroy the evaluation of the transcribed information. Therefore, how to design the embedding watermark method by utilizing the human visual system under the just noticeable threshold become extremely important, which cannot be detectable after watermark embedding.
b) Fragile for malicious alteration:
Any attempts to tamper data should be detected. After malicious attacks, the embedded watermark should become unable to detect, once the watermark is undetectable in the image, we may infer that the data are possibly revised.
c) Tamper regions restriction:
It can identify the tamper region, and confirm the other regions of the data that are not altered.
d) Recovery of the original image:
It can provide rough restoration for tampered data to identify the original data content.
As documented in the simulation results, the proposed method is robust to against attacks and is able to detect tampering. Moreover, the recovery ability is one of the feature of the propose method.
|
|---|
