Copy–move image tampering is one of the most frequently tampering types that contaminate images authenticity due to its ease. Tampering detection becomes more difficult when the tampered parts are subjected to post operations like scaling, rotation, compression or noise. In this paper, a blind copy-move tampering detection and localization method is proposed. Its novelty lies in the combination of SIFT, PCA and DBSCAN techniques. The proposed method shows its potential to disclose and localize tampered regions of different sizes and shapes. Furthermore, our method requires no prior information about the image or the manipulation operations carried on it. A comparative analysis between the proposed method and other tampering detection methods is evaluated based on various performance measures. Our experimental results show that the proposed method outperforms the previous reported techniques and is quite reliable in copy-move tampering detection and localization. |
