Synchronization of cameras from human motion using feature points

We propose an approach to synchronize the videos recorded by consumer level cameras, using feature points which extracted from markerless human motion video sequences. In this paper, we extend the work on computing rank constraints of videos, to find the temporal offset between two video sequences and achieve the goal of video synchronization. Our method uses the feature points´ information to compose fundamental matrices, and then compute the singular values of matrices and finally find out the temporal offset we need.