A NOVEL FRAMEWORK FOR ANTI-FORENSICS IN DIGITAL FORGERY LOCALIZATION

Abstract

Digital image forensics has become an important research area for identifying manipulated or forged images. Modern forensic techniques focus on detecting image tampering by analyzing inconsistencies in pixel patterns, compression artifacts, and statistical features. However, with the rapid advancement of image editing tools and artificial intelligence, attackers are increasingly developing anti-forensic techniques designed to hide traces of manipulation and evade forensic detection systems. This creates significant challenges for reliable forgery localization. This study proposes a novel anti-forensics framework aimed at bypassing digital forgery localization methods by intelligently modifying manipulated regions to mimic authentic image characteristics. The framework utilizes advanced image processing and machine learning techniques to alter forensic traces such as noise patterns, compression artifacts, and edge inconsistencies that are typically used by forensic detection algorithms. By integrating adaptive feature manipulation and learning-based optimization, the proposed system reduces the effectiveness of existing forgery localization models. The framework evaluates its performance against common image forensic detection methods and demonstrates its capability to obscure manipulation evidence while maintaining high visual quality of the altered images. Experimental results show that the proposed approach significantly decreases the detection accuracy of traditional forensic models while preserving perceptual realism. The research highlights the growing need for more robust and resilient digital forensic systems capable of handling advanced anti-forensic attacks. The proposed framework contributes to understanding the limitations of current forensic techniques and encourages the development of stronger detection mechanisms for securing digital media authenticity.