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

This paper proposes GradientSurf, a novel algorithm for real time surfacereconstruction from monocular RGB video. Inspired by Poisson SurfaceReconstruction, the proposed method builds on the tight coupling betweensurface, volume, and oriented point cloud and solves the reconstruction problemin gradient-domain. Unlike Poisson Surface Reconstruction which finds anoffline solution to the Poisson equation by solving a linear system after thescanning process is finished, our method finds online solutions from partialscans with a neural network incrementally where the Poisson layer is designedto supervise both local and global reconstruction. The main challenge thatexisting methods suffer from when reconstructing from RGB signal is a lack ofdetails in the reconstructed surface. We hypothesize this is due to thespectral bias of neural networks towards learning low frequency geometricfeatures. To address this issue, the reconstruction problem is cast ontogradient domain, where zeroth-order and first-order energies are minimized. Thezeroth-order term penalizes location of the surface. The first-order termpenalizes the difference between the gradient of reconstructed implicitfunction and the vector field formulated from oriented point clouds sampled atadaptive local densities. For the task of indoor scene reconstruction, visualand quantitative experimental results show that the proposed methodreconstructs surfaces with more details in curved regions and higher fidelityfor small objects than previous methods.

GradientSurf: Gradient-Domain Neural Surface Reconstruction from RGB Video