Over the years, video surveillance systems have been used for indisputable evidence of a crime. Unfortunately, videos of the surveillance systems can be forged through adding (deleting) an object to (from) a video scene (i.e., object-based forgery) with invisible traces and little
effort. In this paper, we propose a novel approach that uses spatial decomposition, temporal filtering, and sequential analysis to detect object-based video forgery and estimate a movement of removed objects. The results show that our approach not only outperforms a previous approach in detecting
forged videos but it is also more robust against compressed and lower resolution videos. Also, our approach can effectively estimate a movement of different sizes of removed objects.