Show simple item record

dc.contributor.authorZhang, Xianyaoen_US
dc.contributor.authorOtt, Melvinen_US
dc.contributor.authorManzi, Marcoen_US
dc.contributor.authorGross, Markusen_US
dc.contributor.authorPapas, Mariosen_US
dc.contributor.editorGhosh, Abhijeeten_US
dc.contributor.editorWei, Li-Yien_US
dc.date.accessioned2022-07-01T15:36:51Z
dc.date.available2022-07-01T15:36:51Z
dc.date.issued2022
dc.identifier.issn1467-8659
dc.identifier.urihttps://doi.org/10.1111/cgf.14587
dc.identifier.urihttps://diglib.eg.org:443/handle/10.1111/cgf14587
dc.description.abstractWe propose a method for constructing feature sets that significantly improve the quality of neural denoisers for Monte Carlo renderings with volumetric content. Starting from a large set of hand-crafted features, we propose a feature selection process to identify significantly pruned near-optimal subsets. While a naive approach would require training and testing a separate denoiser for every possible feature combination, our selection process requires training of only a single probe denoiser for the selection task. Moreover, our approximate solution has an asymptotic complexity that is quadratic to the number of features compared to the exponential complexity of the naive approach, while also producing near-optimal solutions. We demonstrate the usefulness of our approach on various state-of-the-art denoising methods for volumetric content. We observe improvements in denoising quality when using our automatically selected feature sets over the hand-crafted sets proposed by the original methods.en_US
dc.publisherThe Eurographics Association and John Wiley & Sons Ltd.en_US
dc.subjectCCS Concepts: Computing methodologies --> Ray tracing; Feature selection; Neural networks
dc.subjectComputing methodologies
dc.subjectRay tracing
dc.subjectFeature selection
dc.subjectNeural networks
dc.titleAutomatic Feature Selection for Denoising Volumetric Renderingsen_US
dc.description.seriesinformationComputer Graphics Forum
dc.description.sectionheadersVolume Rendering
dc.description.volume41
dc.description.number4
dc.identifier.doi10.1111/cgf.14587
dc.identifier.pages63-77
dc.identifier.pages15 pages


Files in this item

Thumbnail
Thumbnail

This item appears in the following Collection(s)

  • 41-Issue 4
    Rendering 2022 - Symposium Proceedings

Show simple item record