Machine Learning-Based Image Forgery Detection Using Combinatorial Mapping with Hybrid Distance Measure

Image forgery detection (IFD) is essential for upholding trust and credibility in our digital world, preventing harm, ensuring the accuracy of information, and safeguarding various aspects of society and technology. Hence the research presents a novel contribution to the field of IFD by developing a robust model using Combinatorial Mapping (CM). This model seamlessly integrates K-Nearest Neighbors (KNN), Logistic Regression (LR), and Decision Tree (DT) algorithms, employing Bhattacharya distance (BD), Euclidean distance (ED), and Manhattan Distance (MD) as effective distance measures. Our innovative approach is designed to enhance the accuracy and effectiveness of detecting forged images, offering a promising solution in image authenticity verification. The results demonstrate the effectiveness of our approach, with accuracy rates of 92.86% for Dd1 and 94.50% for Dd2. Sensitivity values of 94.83% for Dd1 and 94.50% for Dd2 underscore the robustness of our system. Notably, the specificity values for Dd1 (94.76%) and Dd2 (94.40%) surpass those achieved by prior methodologies, highlighting the superior performance of our proposed approach.