Optimizing CNN Hyperparameters for Copy-Move Tampered Images Detection

Digital image manipulation is becoming very common due to the widespread and easy use of digital image manipulation software. The development of easy-to-use image editing tools has increased the need for image forensics. The copy-move technique is one of the most commonly used methods for tampering with images. This study proposes an optimized Convolution Neural Network (CNN) model to automatically detect original or fake images because a copy-move has been carried out. In this article, hyperparameter optimization will be carried out to improve the performance of the CNN model, which consists of batch size, learning rate and optimizer. The dataset we used consists of MICC-F220, MICC-F600, MICCF2000 and a combination of the three datasets. We evaluate the results using various performance measures, such as accuracy, f1-score and area under the receiver operating characteristic curve (AUC-ROC). The final results show that the proposed CNN model achieves a test accuracy of 90.91% for MICC-F220, 91.67% for MICC-F600, 98.50% for MICC-F2000 and 94.23% for mixed MICC datasets.