الفهرس 
AbstractOnly 14 pages are availabe for public view
 |  | from | 113 |  |  |
| | | from | 113 | | |
Abstract
The advancement of technology leads to wide spread availability of digital data on the internet. This data includes different types of video, audio, images etc. The manipulation of images becomes very simple with the easy availability of powerful image processing tools. Digital image watermarking developed as a very efficient tool to protect the digital images from manipulation and to protect its integrity but it also suffers with many limitations . This thesis proposes a fragile watermarking approach for image authentication and self-recovery in order to locate any image tampering as well as recovery of original host image. The scheme first breaks the host image into smaller blocks and then performs SVD algorithm (Singular Value Decomposition) on it. The traces of block wise SVD is inserted into the LSB (Least Significant Bit) of image pixels to figure out any manipulation in the original image. The watermarking bits are of two types (authentication and self-recovery), which are used to be replaced with the LSB (Least Significant Bit) of the host Image. The insertion positions of self-recovery bits are determined by the help of Arnold transformation algorithm, which provides an intelligent way to recover the original image even after sufficient tampering. This sort of insertion ensures the different insertion positions of Neighborhood block’s recovery information. As neighborhood blocks of any image pixel contains almost similar information, so this randomize insertion increases the chance of recovery of approximate value of pixels in case of original value get lost or changed by the attacker. SVD based watermarking information not only increases the security of scheme but also provide a way to find out the attacked area in the host Image. The scheme is tested using different types of attacks and shows a better performance towards authentication of host image and self-recovery. The use SVD based authentication provides us very good tamper localization with accuracy. The use of grouped (neighborhood) block method offer more chances of recovery as well as supplementary way to locate the attacked area into the watermarked image. Two authentication bits namely block authentication and self-recovery bits were used to survive the vector quantization attack. The using of Arnold transform makes it possible to recover the tampered region from the neighborhood blocks also. The results showed that proposed scheme was highly reliable and had ability to locate.