Image Manipulation Detection Using Error Level Analysis
Bhavya Shah, Divya Shah, Shubham Thakar, Samkit Shah, Sudhir Dhage · 2023
Image editing is now easy because there are so many user-friendly multimedia tools available. The photographs can be altered using a variety of methods. The process of combining two or more photos to produce a single composite image is known as “image splicing,” one of numerous methods for altering images. These edited images might be used to trick others. A deep learning-based method to detect picture splicing in the photographs is suggested in this work. The input image must first be preprocessed using a technique called “ELA” in order to extract the noise remnant from the input image by suppressing the image content. Second, the well-known MobileNet network is used as a feature extractor. The obtained features are then tested for authenticity using the Voting Classifier classifier. Research on the CASIA dataset shows that the suggested approach performs better than competing approaches already in use. The proposed method achieves an average classification accuracy of 94.24.