{"title":"神经八面体场:同时进行平滑和锐边正则化的八面体先验","authors":"Ruichen Zheng, Tao Yu","doi":"arxiv-2408.00303","DOIUrl":null,"url":null,"abstract":"Neural implicit representation, the parameterization of distance function as\na coordinate neural field, has emerged as a promising lead in tackling surface\nreconstruction from unoriented point clouds. To enforce consistent orientation,\nexisting methods focus on regularizing the gradient of the distance function,\nsuch as constraining it to be of the unit norm, minimizing its divergence, or\naligning it with the eigenvector of Hessian that corresponds to zero\neigenvalue. However, under the presence of large scanning noise, they tend to\neither overfit the noise input or produce an excessively smooth reconstruction.\nIn this work, we propose to guide the surface reconstruction under a new\nvariant of neural field, the octahedral field, leveraging the spherical\nharmonics representation of octahedral frames originated in the hexahedral\nmeshing. Such field automatically snaps to geometry features when constrained\nto be smooth, and naturally preserves sharp angles when interpolated over\ncreases. By simultaneously fitting and smoothing the octahedral field alongside\nthe implicit geometry, it behaves analogously to bilateral filtering, resulting\nin smooth reconstruction while preserving sharp edges. Despite being operated\npurely pointwise, our method outperforms various traditional and neural\napproaches across extensive experiments, and is very competitive with methods\nthat require normal and data priors. Our full implementation is available at:\nhttps://github.com/Ankbzpx/frame-field.","PeriodicalId":501174,"journal":{"name":"arXiv - CS - Graphics","volume":null,"pages":null},"PeriodicalIF":0.0000,"publicationDate":"2024-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"Neural Octahedral Field: Octahedral prior for simultaneous smoothing and sharp edge regularization\",\"authors\":\"Ruichen Zheng, Tao Yu\",\"doi\":\"arxiv-2408.00303\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"Neural implicit representation, the parameterization of distance function as\\na coordinate neural field, has emerged as a promising lead in tackling surface\\nreconstruction from unoriented point clouds. To enforce consistent orientation,\\nexisting methods focus on regularizing the gradient of the distance function,\\nsuch as constraining it to be of the unit norm, minimizing its divergence, or\\naligning it with the eigenvector of Hessian that corresponds to zero\\neigenvalue. However, under the presence of large scanning noise, they tend to\\neither overfit the noise input or produce an excessively smooth reconstruction.\\nIn this work, we propose to guide the surface reconstruction under a new\\nvariant of neural field, the octahedral field, leveraging the spherical\\nharmonics representation of octahedral frames originated in the hexahedral\\nmeshing. Such field automatically snaps to geometry features when constrained\\nto be smooth, and naturally preserves sharp angles when interpolated over\\ncreases. By simultaneously fitting and smoothing the octahedral field alongside\\nthe implicit geometry, it behaves analogously to bilateral filtering, resulting\\nin smooth reconstruction while preserving sharp edges. Despite being operated\\npurely pointwise, our method outperforms various traditional and neural\\napproaches across extensive experiments, and is very competitive with methods\\nthat require normal and data priors. Our full implementation is available at:\\nhttps://github.com/Ankbzpx/frame-field.\",\"PeriodicalId\":501174,\"journal\":{\"name\":\"arXiv - CS - Graphics\",\"volume\":null,\"pages\":null},\"PeriodicalIF\":0.0000,\"publicationDate\":\"2024-08-01\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"arXiv - CS - Graphics\",\"FirstCategoryId\":\"1085\",\"ListUrlMain\":\"https://doi.org/arxiv-2408.00303\",\"RegionNum\":0,\"RegionCategory\":null,\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"\",\"JCRName\":\"\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"arXiv - CS - Graphics","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/arxiv-2408.00303","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
Neural Octahedral Field: Octahedral prior for simultaneous smoothing and sharp edge regularization
Neural implicit representation, the parameterization of distance function as
a coordinate neural field, has emerged as a promising lead in tackling surface
reconstruction from unoriented point clouds. To enforce consistent orientation,
existing methods focus on regularizing the gradient of the distance function,
such as constraining it to be of the unit norm, minimizing its divergence, or
aligning it with the eigenvector of Hessian that corresponds to zero
eigenvalue. However, under the presence of large scanning noise, they tend to
either overfit the noise input or produce an excessively smooth reconstruction.
In this work, we propose to guide the surface reconstruction under a new
variant of neural field, the octahedral field, leveraging the spherical
harmonics representation of octahedral frames originated in the hexahedral
meshing. Such field automatically snaps to geometry features when constrained
to be smooth, and naturally preserves sharp angles when interpolated over
creases. By simultaneously fitting and smoothing the octahedral field alongside
the implicit geometry, it behaves analogously to bilateral filtering, resulting
in smooth reconstruction while preserving sharp edges. Despite being operated
purely pointwise, our method outperforms various traditional and neural
approaches across extensive experiments, and is very competitive with methods
that require normal and data priors. Our full implementation is available at:
https://github.com/Ankbzpx/frame-field.