The Localization of Deleted Frames in the Forged Videos

The manipulation processes made by deleting frames on videos are the fastest and laborless method of video forgery. The process of misleading people is done by cutting the frames of the scene or object to be hidden from the related video. With this study, designing a fast-running algorithm to detect in which location the frame deletion forgery is aimed for localization the forgery location. The test of the algorithm's resistance against different compression formats and different compression qualities was carried out. Frame deletion forgeries were performed in random sizes and locations on the videos obtained from SULFA and TRACE LIBRARY databases which used frequently in the literature. Forgery location was detected with 100% accuracy at uncompressed videos. In MPEG4 and H264 compressed videos, accuracies are %85 and %90 respectively. Average processing time is 0.0007 second for frame. Experimental results show that the proposed method can detect the forgery location with high accuracy.