Robust Image Forgery Detection. Detecting the tampered images from the original ones and stating the type of forgery using Feature Analysis & ML

      Аннотация: In this thesis, we present forgery image detection techniques for two of the most common image tampering techniques; copy-move and splicing. We used match points technique after feature extraction process using SIFT and SURF for detecting the copy-move forgery. For splicing detection, we extracted the edges if integral images of Y. Cb and Cr image components. Using Machine Learning techniques in classifying the images used to test our model that determines whether an image was fake or not, plus pointing out to the type of forgery used. We compared the results of detecting both the forgeries. We developed a GUI that detects both of the forgeries and gives detailed resultswhether a certain image was tampered or authenticated.Finally, we proposed an approach for image verification, using invisible watermark approach by embedding certaindata in an image.

Ключевые слова: Image Forgery, copy-move Forgery Detection, Image splicing, SIFT, SURF, Support vector machine (SVM), Machine Learning, digital forensics