Zurück zur Übersicht

Deblurring photographs of characters using deep neural networks.

full text: html PDF
author/s: Thomas Germer, Stefan Harmeling, Tobias Uelwer
type:Article
journal:Inverse Problems and Imaging
month:January
year:2022
ISSN:1930-8337
language:English
keywords:Helsinki Deblur Challenge 2021, deblurring, deconvolution, point spread function, neural network, U-Net
Abstract

In this paper, we present our approach for the Helsinki Deblur Challenge (HDC2021). The task of this challenge is to deblur images of characters without knowing the point spread function (PSF). The organizers provided a dataset of pairs of sharp and blurred images. Our method consists of three steps: First, we estimate a warping transformation of the images to align the sharp images with the blurred ones. Next, we estimate the PSF using a quasi-Newton method. The estimated PSF allows to generate additional pairs of sharp and blurred images. Finally, we train a deep convolutional neural network to reconstruct the sharp images from the blurred images. Our method is able to successfully reconstruct images from the first 10 stages of the HDC 2021 dataset. Our code is available at https://github.com/hhu-machine-learning/hdc2021-psfnn.

Heinrich Heine Universität

Datenbanken und Informationssysteme

Lehrstuhlinhaber

Prof. Dr. Stefan Conrad


Universitätsstr. 1
40225 Düsseldorf
Gebäude: 25.12
Etage/Raum: 02.24
Tel.: +49 211 81-14088

Sekretariat

Lisa Lorenz



Universitätsstr. 1
40225 Düsseldorf
Gebäude: 25.12
Etage/Raum: 02.22
Tel.: +49 211 81-11312
Verantwortlich für den Inhalt:  E-Mail senden Datenbanken & Informationssysteme