A robust video synchronization method based on hierarchical shot detection
Srinivasa Rao Chalamala, Krishna Rao Kakkirala, Jyoti Dhillon · 2014
We propose an video synchronization method using hierarchical video shot detection process to match two videos when one of them is modified at unknown locations. The shots include both abrupt and gradual transitions (dissolves and fade-groups). The philosophy of this method is to use a light weight method to pre-process the video to remove those frames which strictly does not comprise a normal scene and may contribute to fade-in, fade-out or dissolve areas. These frames are further processed by a complex and robust shot detection method to find the exact shot boundaries. We used object (contour) matching as a pre-processing step and SURF features on top of it for robust shot detection even in fade-in, fade-out or dissolve areas of the scene. Once all the shots are correctly found then the shot-boundaries are matched to identify the point of videos synchronization.