Rain removal in videos based on optical flow and hybrid properties constraint
Yuan Zhou, Yusheng Han, Pucheng Zhou · 2015
The raindrops decrease the quality of outdoor surveillance system, and it brings difficulty for computer vision algorithm, especially in using feature information. This paper proposed a rain removal algorithm based on optical flow and hybrid properties constraint. To identify the candidate rain pixels, the optical flow is adopted. And then, the hybrid properties constraint of raindrops is used to refine the rain streaks. When the rain streaks are accurately detected, the scene can be restored by a weighted composing method. Experiments on rain-affected videos show the feasibility and validity of our algorithm.