Deep Image Deblurring: A Survey

Abstract

Image deblurring is a classic problem in lowlevel computer vision, which aims to recover a sharp image from a blurred input image. Recent advances in deep learning have led to significant progress in solving this problem, and a large number of deblurring networks have been proposed. This paper presents a comprehensive and timely survey of recently published deep-learning based image deblurring approaches, aiming to serve the community as a useful literature review. We start by discussing common causes of image blur, introduce benchmark datasets and performance metrics, and summarize different problem formulations. Next we present a taxonomy of methods using convolutional neural networks (CNN) based on architecture, loss function, and application, offering a detailed review

A Comprehensive Survey on Deep Neural Image Deblurring

“…• Image deblurring Datasets The image deblurring datasets currently available in the literature include synthetic pairs of blurry/sharp images. As a consequence, trained networks often perform poorly on some real blurry images [165]. To address this issue, some efforts need to follow to study real-world blurring effects and blurring sources and capture massive realistic images based on the understanding.…”

“…Another brief survey conducted by Li [77] reviews the conventional prior-based optimization methods along with deep neural image deblurring methods. Meanwhile, as the most recent and significant survey, Zhang et al [165] provide more extensive reviews of deep neural image deblurring approaches. They discuss various blur types, image quality assessment methods, general network architectures with their corresponding loss functions.…”

Learnable Blur Kernel for Single-Image Defocus Deblurring in the Wild

“…GAN usually works well on details and textures. In the field of image deblurring, GAN has a very wide range of applications (Zhang et al 2022), De-blurGAN (Kupyn et al 2018) and DeblurGAN-V2 (Kupyn et al 2019) are the most famous methods of them. While these current works were not suitable for defocus deblurring task, which achieved weak performance.…”

Using Deep Learning to Improve Detection and Decoding Of Barcodes

“…In terms of barcode image deblurring, Shi et al [6] provide an excellent review of state-of-the-art deblurring approaches, which are generally based on traditional image restoration techniques with focus on utilization of image statistics as priors (e.g., image distribution and edge information) to estimate the blur kernel (i.e., transformation) and restore the sharp image through an iterative process [6,7]. According to a survey by Zhang et al [8], a variety of architectures have been explored for deblurring in other domains, including deep autoencoders (DAE), cascaded networks, multi-scale networks, GANs, etc. In an attempt to explore deep learning for barcode deblurring, Wang et al [9] trained a DeblurGAN model [10] to perform motion deblur of QR codes.…”