Abstract | ||
---|---|---|
This paper presents an approach for automatically estimating the spatial bound of the blur kernel in a motionblurred image, based on the statistics of multi-level image gradients. We observe that blur has a significant impact on the Histogram of Oriented Gradients (HOG) at higher levels of an image pyramid, but has much less impact at coarser levels. Based on this fact we estimate the spatial bound of the unknown blur kernel using a learning-based approach. We first learn a generic pyramid HOG model from natural sharp images, then given a HOG pyramid of a blurry image, we predict the corresponding model of its latent sharp image. Finally, we learn another model to predict the spatial kernel bound from the difference between the observed and the predicted HOG pyramids. Experimental results show that the proposed method can estimate accurate blur kernel sizes, enabling existing blind deconvolution methods to achieve best possible results. |
Year | DOI | Venue |
---|---|---|
2016 | 10.1109/TCSVT.2015.2418585 | IEEE Trans. Circuits Syst. Video Techn. |
Keywords | Field | DocType |
Blur-Kernel Bound Estimation,Image Deblur,Motion Deblur,Motion Prior,Pyramid Statistics | Histogram,Blind deconvolution,Computer science,Pyramid (image processing),Artificial intelligence,Pyramid,Image restoration,Computer vision,Pattern recognition,Support vector machine,Histogram of oriented gradients,Statistics,Variable kernel density estimation | Journal |
Volume | Issue | ISSN |
PP | 99 | 1051-8215 |
Citations | PageRank | References |
3 | 0.37 | 14 |
Authors | ||
4 |
Name | Order | Citations | PageRank |
---|---|---|---|
Shiyuan Liu | 1 | 17 | 8.28 |
Hsin-min Wang | 2 | 1201 | 129.62 |
Jue Wang | 3 | 2871 | 155.89 |
Pan, C. | 4 | 3 | 0.37 |