Digital Integrity: Enhancing Image Forensics with Error Level Analysis and Deep Learning

The widespread use of digitally altered images highlight the need for trustworthy methods to differentiate between real and altered content. This study introduces ELA as a more effective and economical method for detecting changes in digital images, diverging from the original method described in the base work, which employed Fourier transform of residual images. This study offers a reliable method for differentiating between authentic and altered photos using error level analysis (ELA) in combination with a lightweight convolutional neural network (CNN), which is MobileNet. The proposed methodology uses ELA to detect faint aberrations suggestive of picture alteration, from which MobileNet extracts significant features in an effective manner. An SVM is then used to classify the collected features, thereby drawing a line between genuine and artificial intelligence (AI)-generated images. The results of the experiments show that the suggested design performs robustly in picture classification tasks, achieving a high accuracy of 94.18%. This research adds to the field of image forensics by improving the dependability of fake image recognition and providing a scalable solution that can be used in a variety of real-world circumstances, such as cybersecurity and content verification. This integration of ELA with SVM and MobileNet is a major step forward in the ongoing actions taken to maintain the integrity of digital media and fight false information.