Identifying Video Forgery using Object Trajectory and Optical Flow

With the rapid growth in video editing software, digital video has become increasingly vulnerable to tampering which can be used to mislead the viewer with wrong information. This paper develops an algorithm to resolve the issue of most frequently practiced copy-move tampering or duplication of frames using optical flow. It extracts the foreground and estimates the object trajectory using its optical flow values. These motion paths are then refined to achieve a smooth trajectory of every motion in a video. Such motion data are analyzed to identify the forged area. Experimental evaluation on a public video data set is presented to effective and efficient in detecting and localizing duplication forgery.