Unfolding a blurred image

TL;DR

This paper introduces a novel method to reconstruct a sequence of sharp, temporally consistent frames from a single blurred image by learning motion representations and employing a guided training approach.

Contribution

It proposes an unsupervised learning framework for motion representation from videos and applies it to generate sharp video sequences from blurred images, outperforming existing methods.

Findings

Outperforms state-of-the-art in accuracy, speed, and compactness

Capable of real-time single image deblurring

Successfully generates plausible, temporally consistent sharp frames

Abstract

We present a solution for the goal of extracting a video from a single motion blurred image to sequentially reconstruct the clear views of a scene as beheld by the camera during the time of exposure. We first learn motion representation from sharp videos in an unsupervised manner through training of a convolutional recurrent video autoencoder network that performs a surrogate task of video reconstruction. Once trained, it is employed for guided training of a motion encoder for blurred images. This network extracts embedded motion information from the blurred image to generate a sharp video in conjunction with the trained recurrent video decoder. As an intermediate step, we also design an efficient architecture that enables real-time single image deblurring and outperforms competing methods across all factors: accuracy, speed, and compactness. Experiments on real scenes and standard datasets demonstrate the superiority of our framework over the state-of-the-art and its ability to generate a plausible sequence of temporally consistent sharp frames.