A Block Feature Correlation Based Image Watermarking for Tamper Detection Using Linear Equation

Abstract

A new watermark method for detecting image tampering is proposed in this paper. First, an authentication number is created by using a pair of watermark pixels as the coefficients of a linear function. The central pixel of a host block as the input data to the linear function. The authentication information of that block is then embedded into the host image using the LSB matching revisited embedding procedure. In the detection stage, ISB matching revisited will be applied to extract the authentication information and to verify whether the image blocks have been tampered with. The authentication information establishes the block feature correlation. This kind of correlation helps resist VQ attack.