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

: This paper proposes a new watermarking method for detecting image tampering. First, an authentication number is created using a pair of watermark pixels as the coefficients of a linear function and the central pixel of a host block as the input data. The authentication information of that block is then embedded into the host image using the LSB matching revisited embedding procedure. In the detection stage, LSB matching revisited algorithm will be applied to extract the establishes the block feature correlation, which helps to resist the VQ attack and to verify whether the image blocks have been manipulated. The correlation will raise the fault negative rate when not all three blocks in a group are damaged. This problem is fixed by embedding block features into corresponding masking blocks using the self-embedding technique. In the final stage, as the unauthentication area is detected, we apply a prediction scheme to restore the tampered area to its visual image.

• Type: Conference
• Year: 2009
• Where: IEEEIAS
• Authors: Chin-Feng Lee, Kuo-Nan Chen, Chin-Chen Chang, Meng-Cheng Tsai
• Keywords: Block Feature Correlation, Block Features, Host Block, IEEEIAS 2009, Information Technology