Coarse-to-fine copy-move image forgery detection method based on discrete cosine transform

Abstract

Digital image forgery has become a serious problem in the present society. As the world is advancing in the information and communication technology, it has become more crucial for researchers to take part in overcoming the wide-spreading digital image forgery to prove an image authenticity especially when the legislative field is involved. Copy-move forgery is a type of image forgery where one part of an image is copied and pasted in other regions of the same image, and it is one of the most common image forgeries to conceal some information in the original image. There are numerous techniques available to detect copy-move forgeries which each of them have their own advantages and drawbacks. Discrete Cosine Transform (DCT) is a powerful algorithm developed as a method to detect copy-move forgery which is well known for its detection efficiency. However, the detection rate relies intensely on the size of block used. Small block size is known for its ability to detect fine cloned objects, but the drawback is it produces too many false positive and requires high execution time. In this research, a method to overcome the weaknesses of using small block size by applying the coarse-to-fine approach with the two-tier process is proposed. The proposed method is evaluated on fifteen forged images on the CoMoFoD dataset. The results demonstrated that the proposed method is able to achieve high precision and recall rate of over 90% as well as improves the computation time by reducing the overall duration of forgery detection up to 73% compared to the traditional DCT method using small block size. Therefore, these findings validate that the proposed method offers a trade-off between accuracy and runtime.