Copy–Move Forgery Detection Algorithm: A Machine Learning-Based Approach to Detect Image Forgery

The development of technology and the widespread availability of editing software promote the forging of digital material. Anyone with a smartphone is able to edit photographs and modify their attributes. Copy and move forgery is the most prevalent sort of picture counterfeiting. An approach based on machine learning is proposed in this work to identify copy–move picture forgery. The suggested technique detects and localizes copy–move patches. The experiment is conducted with the CoMoFoD picture collection. This dataset contains both the original and ground truth pictures for each photograph. This collection includes photos with various manipulations, including scaling, JPEG compression, and rotation. The photos are scaled based on various scaling parameters. The suggested methodology utilized supervised learning to classify forgery patch pixels. The experiment obtains an accuracy of 94.37, a recall of 97.03, and an F1 value of 94.03. The performance of the suggested algorithms is compared to the state-of-the-art approach and found to be superior.