Video Deblurring via Semantic Segmentation and Pixel-Wise Non-Linear Kernel

TL;DR

This paper introduces a novel pixel-wise non-linear kernel model for video deblurring that leverages semantic segmentation and non-linear optical flow to better handle complex motion blur in challenging scenes.

Contribution

It proposes a new non-linear kernel model based on non-linear optical flow and scene understanding to improve deblurring accuracy in complex videos.

Findings

Outperforms state-of-the-art deblurring methods on challenging videos.

Effectively models complex motion blur trajectories.

Utilizes semantic segmentation to guide optical flow estimation.

Abstract

Video deblurring is a challenging problem as the blur is complex and usually caused by the combination of camera shakes, object motions, and depth variations. Optical flow can be used for kernel estimation since it predicts motion trajectories. However, the estimates are often inaccurate in complex scenes at object boundaries, which are crucial in kernel estimation. In this paper, we exploit semantic segmentation in each blurry frame to understand the scene contents and use different motion models for image regions to guide optical flow estimation. While existing pixel-wise blur models assume that the blur kernel is the same as optical flow during the exposure time, this assumption does not hold when the motion blur trajectory at a pixel is different from the estimated linear optical flow. We analyze the relationship between motion blur trajectory and optical flow, and present a novel pixel-wise non-linear kernel model to account for motion blur. The proposed blur model is based on the non-linear optical flow, which describes complex motion blur more effectively. Extensive experiments on challenging blurry videos demonstrate the proposed algorithm performs favorably against the state-of-the-art methods.