Gabor Filter and Centre Symmetric-Local Binary Pattern based technique for forgery detection in images

This paper presents a new approach for the detection of copy-move image forgery, a commonly employed technique for image manipulation. The proposed method combines a modified version of the Gabor Filter and Centre Symmetric Local Binary Pattern (CS-LBP) for feature extraction, aiming to meet the growing demand for accurate forgery detection. The process involves pre-processing the image and extracting features using the Gabor filter and CS-LBP at varying scales and orientations. Key points are matched using the Manhattan distance to identify forged regions. Classification of the forged images is achieved using Hybrid Neural Networks with Decision Tree (HNN-DT). In order to assess the performance of the presented method, diverse image datasets are used and compared to existing feature extraction techniques. The results demonstrate the effectiveness of the modified Gabor filter with CS-LBP in accurately classifying forged images. Specifically, the HNN-DT method with Gabor filter CS-LBP feature extraction surpasses HNN-DT with SURF and PCA feature extraction in terms of classification accuracy and overall performance. Evaluation on the CoMoFoD database confirms superior results compared to existing techniques, establishing the proposed method as a reliable approach for distinguishing between authentic and forged images. Consequently, it serves as a robust solution for image classification and forgery detection. The presented work focuses on detecting mainly three types of image forgery, namely retouching, splicing, and cloning. It is applicable to various image formats, including JPEG and BMP, and is designed specifically for forensic applications in image forgery detection.