A Comparative Study of Copy-Move Forgery Detection Techniques

In the modern world, the popularity of image editing software has risen sharply, making it easy for non-experts to manipulate images. This poses significant challenges regarding image authenticity and integrity. Consequently, a lot of attention has been aroused by researchers working on this topic and many methods have been developed to discover image forgeries. Copy-move is a manipulation technique that involves copying a region in an image and pasting it in a different location on the same image, leading to misinterpretation. In this paper, a comparative study between conventional and deep learning-based methods is presented. More specifically a statistical-based method is compared against a transfer learning method. Various experiments have been conducted on six different publicly available datasets for different copy-move scenarios. The experiments revealed the superior performance of the deep learning approach, surpassing the statistical method in most cases. This study highlights the effectiveness of deep learning techniques in the field of copy-move forgery detection. Furthermore, the comparative analysis presented in this paper serves as a valuable reference for researchers to accomplish further advancements in image forensics.