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

Image alignment by mesh warps, such as meshflow, is a fundamental task whichhas been widely applied in various vision applications(e.g., multi-frameHDR/denoising, video stabilization). Traditional mesh warp methods detect andmatch image features, where the quality of alignment highly depends on thequality of image features. However, the image features are not robust inoccurrence of low-texture and low-light scenes. Deep homography methods, on theother hand, are free from such problem by learning deep features for robustperformance. However, a homography is limited to plane motions. In this work,we present a deep meshflow motion model, which takes two images as input andoutput a sparse motion field with motions located at mesh vertexes. The deepmeshflow enjoys the merics of meshflow that can describe nonlinear motionswhile also shares advantage of deep homography that is robust againstchallenging textureless scenarios. In particular, a new unsupervised networkstructure is presented with content-adaptive capability. On one hand, the imagecontent that cannot be aligned under mesh representation are rejected by ourlearned mask, similar to the RANSAC procedure. On the other hand, we learnmultiple mesh resolutions, combining to a non-uniform mesh division. Moreover,a comprehensive dataset is presented, covering various scenes for training andtesting. The comparison between both traditional mesh warp methods and deepbased methods show the effectiveness of our deep meshflow motion model.

DeepMeshFlow: Content Adaptive Mesh Deformation for Robust Image Registration