Wavelet Based Region Duplication Forgery Detection

With the assistance of editing tools, digital images can be easily tampered without leaving many clues. As a result, the authenticity and integrity of digital images are no longer reliable, which is critical when the images are used as evidence in court or widely spread in public. Region duplication is a commonly used tampering technique for digital images, which a region is copied from an image and pasted to another region of the same image. This technique is used to hide important objects or for decoration. In this paper, we propose a wavelet-based approach to detect region duplication forgery. The image is divided into overlapped blocks with fixed size. The multilevel 2D discrete wavelet transform is then applied to each block. We extract the discriminative features from the wavelet coefficients of a block. The feature vectors of all blocks are lexicographically sorted and the block matching step is applied to find the duplicated blocks. Experimental results demonstrate that the proposed wavelet-based method can successfully detect the duplicated regions. It can also achieve better performance than the DCT-based method, especially when the images are heavily distorted with noises.