Large scale image tamper detection and restoration

Article Type

Research Article

Publication Title

Multimedia Tools and Applications

Abstract

Detection and restoration of greatly tampered images is an important albeit difficult problem. Two schemes for the detection of tampered areas in images and their restoration have been presented. One of them works in the spatial domain and implements a quadruple watermarking approach. A watermark is constructed from four different parts of the image such that a filtered version of the image intensities contained in each of these parts is embedded identically in four other regions of the image. These regions are decided by a mapping algorithm. The advantage of this approach is that even if three of the four regions are tampered, the watermark from the untampered region can be used to reconstruct the three tampered regions. The chief motivation behind the quadruple scheme is the restoration of an image which has suffered tampering on a really large scale, upto 75%. The other proposed algorithm uses wavelet decomposition, based on which, two different watermarks are embedded. These serve two different purposes, one being tamper detection, while the second is restoration of the tampered area. They are embedded in non-overlapping regions of the wavelet transformed image. This algorithm is designed to obtain very good quality restoration and works well for tampering less than 50% of the total image area. The performance of both these algorithms has been examined using images from the entire USC-SIPI database. Comparison has been made with a well known existing approach. The superiority of the proposed approaches is evident from the plots, figures and tables presented.

First Page

17761

Last Page

17791

DOI

10.1007/s11042-020-08669-0

Publication Date

7-1-2020

This document is currently not available here.

Share

COinS