Bringing Alive Blurred Moments

TL;DR

This paper introduces a novel method to reconstruct a sequence of sharp video frames from a single blurred image by learning motion representations and employing a trained autoencoder, outperforming existing techniques in accuracy and speed.

Contribution

It presents a new framework combining unsupervised motion learning and guided training to generate temporally consistent videos from blurred images, with an efficient real-time deblurring architecture.

Findings

Outperforms state-of-the-art methods in accuracy and speed.

Successfully generates plausible, temporally consistent sharp video sequences.

Demonstrates effectiveness on real scenes and standard datasets.

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.