Fast removal of non-uniform camera shake

By Michael Hirsch, Christian J. Schuler, Stefan Harmeling, Bernhard Schölkopf. 2011

356 citations recorded by the source corpus.

Subfields: Computer Vision and Pattern Recognition

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Vision and Pattern Recognition

Citations in the corpus

Learning a convolutional neural network for non-uniform motion blur removal (2015, background): “In [7, 8, 25, 26, 31], nonuniform motion blur is modeled as a global camera motion, which basically estimates an uniform kernel in the camera motion space.”
Unnatural L0 Sparse Representation for Natural Image Deblurring (2013, methodology): “For comparison, previously most efﬁcient approach [9] takes 1567 seconds for the same image.”
Blind Image Deblurring Using Dark Channel Prior (2016, methodology): “We employ the fast forward approximation [11] to estimate the latent image I and the weight k.”
A Comparative Study for Single Image Blind Deblurring (2016, background): “The progress in this field can be attributed to the advancement of efficient inference algorithms [2] [5] [17] [35] [44], various natural image priors [15] [21] [29] [38] [45], and more general motion blur models [8] [9] [11] [43].”
Deblurring by Realistic Blurring (2020, background): “C V ] 7 M a y 2020 instant and sharp snapshots [10].”
From Motion Blur to Motion Flow: A Deep Learning Solution for Removing Heterogeneous Motion Blur (2016, methodology): “Some recent methods [40, 12, 14, 18, 26, 32] have been proposed to remove spatially-varying blur caused by camera panning, and/or object movement, with some restrictive assumptions on the types of blur, image prior, or both.”
Deep Multi-scale Convolutional Neural Network for Dynamic Scene Deblurring (2016, background): “Most of these approaches are based on following blur model [29, 10, 13, 11]. where B , S and n are vectorizd blurry image, latent sharp image, and noise, respectively.”
DeblurGAN: Blind Motion Deblurring Using Conditional Adversarial Networks (2017, methodology): “Following [14][25][27] we take a random between 5 and 25 frames taken by 240fps camera and compute the blurred version of a middle frame as an average of those frames.”
Learning to Deblur (2014, methodology): “conjunction with the recently proposed fast forward model of [31] to restore images with spatially-varying blur.”
Image Deblurring via Extreme Channels Prior (2017, methodology): “In addition, we also test the competing methods [20, 19, 23, 12, 7, 22, 17] on the benchmark dataset by [11], which includes 4 images and 12 blur kernels.”
Human-Aware Motion Deblurring (2019, methodology): “33]. Furthermore, rather than simply assuming the whole image is associated with a uniform blur kernel, some methods estimate a global blur kernel descriptor [45,10], or predict a set of blur kernels [13,11]. However, they are limited by assuming that the sources of blurs are camera motions. For mor”
Segmentation-Free Dynamic Scene Deblurring (2014, methodology): “Our method does not require any complex model to handle specific types of motion blur [10, 23, 27] and does not depend on the accurate motion segmentation.”
Real-World Blur Dataset for Learning and Benchmarking Deblurring Algorithms (2020, background): “To deal with more realistic blur, several approaches with extended blur models have been proposed, but their performance is still limited due to the inherent ill-posedness of the inverse problem [18, 44, 8, 14, 17, 19].”
Learning to Estimate and Remove Non-uniform Image Blur (2013, unknown): “In summary, previous approaches to our deblurring problem either (i) fall short in the estimation or the deblurring step, (ii) require multiple images, or (iii) consider a limited set of blur types (e.g., uniform, constant velocity motion, camera rotation).”
Deep Image Deblurring: A Survey (2022, unknown): “Numerous approaches have been developed to address non-uniform blur by modeling the blur kernel from 3D camera motion (Hirsch et al. 2011; Whyte et al. 2012).”