Analyzing spatially-varying blur

By Ayan Chakrabarti, Todd Zickler, William T. Freeman. 2010

218 citations recorded by the source corpus.

Blur is caused by a pixel receiving light from multiple scene points, and in many cases, such as object motion, the induced blur varies spatially across the image plane. However, the seemingly straight-forward task of estimating spatially-varying blur from a single image has proved hard to accomplish reliably. This work considers such blur and makes two contributions: a local blur cue that measures the likelihood of a small neighborhood being blurred by a candidate blur kernel; and an algorithm that, given an image, simultaneously selects a motion blur kernel and segments the region that it affects. The methods are shown to perform well on a diversity of images.

Subfields: Computer Vision and Pattern Recognition, Media Technology

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

In Media Technology

Citations in the corpus

Non-uniform Deblurring for Shaken Images (2010, background): “Much of the previous work has relied on the local uniformity of the blur, for example, modeling blur due to moving objects as piecewise uniform (Levin 2006; Cho et al. 2007; Chakrabarti et al. 2010), or approximating a continuously varying blur by a spatially varying combination of localized uniform”
$L_0$ -Regularized Intensity and Gradient Prior for Deblurring Text Images and Beyond (2017, background): “In addition to camera motion, blurred images caused by different object motions are analyzed [28], [29], [30].”
Deep Image Deblurring: A Survey (2022, background): “Although these approaches can model out-of-plane camera shake, they cannot handle dynamic scenes, which motivated the use of blur ﬁelds of moving objects (Chakrabarti et al. 2010; Gast et al. 2016; Hyun Kim et al. 2013).”
Sharpness-Aware Low-Dose CT Denoising Using Conditional Generative Adversarial Network (2017, background): “number of reconstructed atoms [32]. There is research that tries to directly estimate the blur kernel but the estimated maps tend to be very coarse and the optimization process is very time consuming [76, 12]. There are other works that can produce a sharpness map, such as in depth map estimation [7”
A Blur-Robust Descriptor with Applications to Face Recognition (2012, background): “A common assumption made to overcome this condition is to assume the blur to be locally uniform [6], which is valid in most practical cases.”
Learning a convolutional neural network for non-uniform motion blur removal (2015, methodology): ““BlurSpect” is the approach proposed in [1].”
From Motion Blur to Motion Flow: A Deep Learning Solution for Removing Heterogeneous Motion Blur (2016, background): “Recently, learning-based discriminative methods [4, 7] have been proposed to learn blur image patterns and avoid the heavy computational cost of blur estimation.”
Learning to Estimate and Remove Non-uniform Image Blur (2013, methodology): “We show in this section that estimating the non-uniform blur of an image can be cast as a segmentation problem, where uniform regions correspond to homogeneous blur strength.”