Copy-move Image Forgery Detection Based on Gabor Descriptors and K-Means Clustering

At present, popularity of using image as the fundamental media of information is growing. Rapid development of technology brings effective image processing tools available and makes image forgery very easy. As an outcome, it turns into a complicated issue in late time. In that case, validating the legitimacy and integrity of digital images is ending up progressively vital issue. The most challenging region-duplication forgery is made by copying some portion of an image and pasting on different region of the same image. This study proposes an efficient region-duplication forgery detection technique. This research is categorized into segment-based region duplication forgery detection method. The design of the algorithm based on image segmentation and using Gabor descriptors and K-Means clustering. Initially, the image is segmented using normalized cut (NCut) segmentation technique. Then, applied Gabor Filters to extract image features and cluster similar features using KMeans clustering algorithm. Finally, comparing the clustering regions with the given threshold value will decide image authenticity. Experiment results proves the strength of the proposed method against various post-processing attacks such as rotation, scale, blurring and JPEG compression. A comparison with existing image forgery detection algorithms demonstrates that the proposed algorithm gives better performance.