A Fast Simple Optical Flow Computation Approach Based on the 3-D Gradient

作者:Li, Yuanwei; Zhu, En*; Zhao, Jianmin; Yin, Jianping; Zhao, Xiangfu
来源:IEEE Transactions on Circuits and Systems for Video Technology, 2014, 24(5): 842-853.
DOI:10.1109/TCSVT.2013.2283648

摘要

Optical flow estimation is a fundamental task of many computer vision applications. In this paper, we propose a fast simple algorithm to compute optical flow based on the 3-D gradient in video sequences. Although the algorithm does not provide highly accurate results, it is computationally simple and fast, and the output is applicable for many applications. The basic idea is that points will form trajectories in video sequences, and the trajectory between two frames of each point is approximated as a straight line, which is the tangent of the trajectory in our algorithm. Therefore, the optical flow of each point is the projecting line of the straight line, which represents its trajectory, in the image plane. Experimental results show that the proposed algorithm is efficient and effective, and is of satisfying accuracy on angle. It is able to provide effective optical flow results for real-time applications.