English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

The defocus deblurring raised from the finite aperture size and exposure timeis an essential problem in the computational photography. It is verychallenging because the blur kernel is spatially varying and difficult toestimate by traditional methods. Due to its great breakthrough in low-leveltasks, convolutional neural networks (CNNs) have been introduced to the defocusdeblurring problem and achieved significant progress. However, they apply thesame kernel for different regions of the defocus blurred images, thus it isdifficult to handle these nonuniform blurred images. To this end, this studydesigns a novel blur-aware multi-branch network (BaMBNet), in which differentregions (with different blur amounts) should be treated differentially. Inparticular, we estimate the blur amounts of different regions by the internalgeometric constraint of the DP data, which measures the defocus disparitybetween the left and right views. Based on the assumption that different imageregions with different blur amounts have different deblurring difficulties, weleverage different networks with different capacities (\emph{i.e.} parameters)to process different image regions. Moreover, we introduce a meta-learningdefocus mask generation algorithm to assign each pixel to a proper branch. Inthis way, we can expect to well maintain the information of the clear regionswhile recovering the missing details of the blurred regions. Both quantitativeand qualitative experiments demonstrate that our BaMBNet outperforms thestate-of-the-art methods. Source code will be available athttps://github.com/junjun-jiang/BaMBNet.

BaMBNet: A Blur-aware Multi-branch Network for Defocus Deblurring