Author : Subrato Kumar 1
Date of Publication :30th May 2023
Abstract: This study presents a novel approach to detect image forgery using Error Level Analysis (ELA) with Sequential Convolutional Neural Network (SCNN) model named as error level sequential convolution neural network (ELSCNN). ELA is a widely-used technique in detecting image manipulation that exploits the variance in compression artifacts between authentic and manipulated images. The proposed method enhances the effectiveness of ELA by integrating it with SCNN models, which learn to classify authentic and manipulated images using a large set of training data. This models are trained to detect forgery in various image manipulation scenarios. Experimental results demonstrate that the proposed method outperforms than existing ELA-based methods in respecting of accuracy, robustness, and computational ability. The proposed method has potential applications in forensic investigation, media authentication, and content verification
Reference :