Analysis of image forgery detection using convolutional neural network. (28th June 2022)

• Record Type: Journal Article
• Authors: Gnaneshwar, Chiluveru; Singh, Manish Kumar; Yadav, Satyendra Singh; Balabantaray, Bunil Kumar
• Is Part Of: International journal of applied systemic studies. Volume 9:Number 3(2022)
• Page Start: 240
• Page End: 260
• Publication Date: 2022-06-28
• Subjects: image forgery detection -- convolutional neural network -- CNN -- error level analysis -- ELA -- machine learning -- ML -- deep learning -- DL; System analysis -- Periodicals; 003
• Languages: English
• ISSNs: 1751-0589

Abstract

Prior to the age of cameras, if someone wanted to see/verify any incident or document, then one must go to that place and verify. The fact is that no one ever questions once someone has verified something with their own eyes. Nowadays, with the rapid development of new technologies, one cannot be sure of an image, which one is a copy of the sight or not a sight itself. Such types of verifications are not possible in the current time due to the development of varieties of advanced image editing tools like Corel draw, Photoshop, GIMP, etc. These are low cost and open-source tools for the users and frequently used to make memes on social media websites. This paper presents an image forgery detection using convolutional neural networks (CNNs/ConvNet). The error level analysis (ELA) method is discussed in detail for image forgery detection. The binary decision of CNN-based model helps in declaration of an image aptness for official uses. The CNN model has been trained for the Kaggle dataset and detailed simulations have been carried out to validate the accuracy and precision of the proposed model.