FORGERY DETECTION IN DIGITAL IMAGES USING CNN

Abstract

With the widespread availability of smartphones equipped with digital cameras and other inexpensive imaging equipment, as well as the low cost of internet connectivity, there is a growing interest in capturing and sharing images online. Many image-editing and enhancement applications are freely available for smartphones and computers, and while they are frequently used to improve and enhance image quality, they are also occasionally used to make alterations to the image in order to propagate false information or slander someone. Image forging is the process of concealing details or adding something that was not o riginally part of the image, causing the integrity of the image to be compromised. For example, a forged driving licence provides false information about a person's ability to drive. Also, digital images are now often utilised as evidence in the media and courts. Therefore, it becomes extremely important to spot these tampered and forged images. In this work, the various types of image forgeries are explored. This study includes a review and comparison of several detection methods for image forgeries. To solve this issue, a CNN network has been built based on previous research and its performance was compared on two different datasets. A data augmentation strategy, as well as multiple hyperparameter tuning, are also evaluated in terms of classification accuracy. Our results suggest that the outcomes are highly dependent on the difficulty of the dataset. In this research, we proposed a Convolutional Neural Network (CNN) model for detecting image forgery.