Copy-Move Image Forgery Detection using Deep Learning Approaches

The advancement of digital image processing software has reached a stage where it is effortless to manufacture forgeries by using numerous manipulating approaches on authentic photos. Occupations such as law, healthcare, and education are susceptible to manipulated imagery. Therefore, it is crucial to ascertain the authenticity of an image. Single type of image manipulation is identified as ‘copy move,’ when significant item or objects are concealed by duplicating and inserting them into the original image. It is crucial to determine whether an image is genuine. The current techniques employed to identify image forgeries relied on conventional feature extraction algorithms, such as key point and block-based algorithms. These conventional approaches yield a subpar outcome. Deep learning approaches have established superior performance in image processing applications. This paper presents a highly effective method for identifying instances of copy-move forgery in images using a hybrid approach that combines deep learning (DL) such as (VGG19 and ResNet50) for feature extraction and Machine Learning (ML) algorithms including (Random Forests (RF) and Support Vector Machine (SVM)) to classify digital images. The proposed models (VGG19-RF), (VGG19-SVM), (ResNet50-RF), and (ResNet50-SVM) have been compared using two standard datasets. The results indicate that ((VGG19-RF) achieves higher recall, accuracy, precision and F1-score of 97% on MICC-F220 dataset, On the other hand (VGG19-RF) and (ResNet50-RF) achieve higher recall, precision, accuracy and F1-score of 97% on the MICC-F2000 dataset.