{"id":{"repo_id":"york","oai_identifier":"oai:yorkspace.library.yorku.ca:10315/41338"},"canonical_url":"https://search.dev.ndltd.org/etd/york/oai:yorkspace.library.yorku.ca:10315/41338","repository":{"repo_id":"york","name":"York University","base_url":"https://yorkspace.library.yorku.ca/oai/request"},"display":{"title":"Volumetric Attribute Compression for 3D Point Clouds using Feedforward Network with Geometric Attention","abstract":"We study 3D point cloud attribute compression using a volumetric approach: given a target volumetric attribute function $f : \\mathbb{R}^3 \\rightarrow \\mathbb{R}$, we quantize and encode parameter vector $\\theta$ that characterizes $f$ at the encoder, for reconstruction $f_{\\hat{\\theta}}(\\x)$ at known 3D points $\\x$'s at the decoder, where $\\hat{\\theta}$ is a quantized version of $\\theta$. Extending a previous work Region Adaptive Hierarchical Transform (RAHT) that employs piecewise constant functions to span a nested sequence of function spaces, we propose a feedforward linear network that implements higher-order B-spline bases spanning function spaces without eigen-decomposition. Feedforward network architecture means that the system is amenable to end-to-end neural learning. The key to our network is space-varying convolution, similar to a graph operator, whose weights are computed from the known 3D geometry for normalization. We show that the number of layers in the normalization at the encoder is equivalent to the number of terms in a matrix inverse Taylor series. Experimental results on real-world 3D point clouds show up to 2-3 dB gain over RAHT in energy compaction and 20-30\\% in bitrate reduction.","abstract_html":"We study 3D point cloud attribute compression using a volumetric approach: given a target volumetric attribute function <span class=\"etd-inline-math\">f : \\mathbb{R}<sup>3</sup> \\rightarrow \\mathbb{R}</span>, we quantize and encode parameter vector <span class=\"etd-inline-math\">&theta;</span> that characterizes $f$ at the encoder, for reconstruction <span class=\"etd-inline-math\">f<sub>\\hat{&theta;}</sub>(\\x)</span> at known 3D points $\\x$&#x27;s at the decoder, where <span class=\"etd-inline-math\">\\hat{&theta;}</span> is a quantized version of <span class=\"etd-inline-math\">&theta;</span>. Extending a previous work Region Adaptive Hierarchical Transform (RAHT) that employs piecewise constant functions to span a nested sequence of function spaces, we propose a feedforward linear network that implements higher-order B-spline bases spanning function spaces without eigen-decomposition. Feedforward network architecture means that the system is amenable to end-to-end neural learning. The key to our network is space-varying convolution, similar to a graph operator, whose weights are computed from the known 3D geometry for normalization. We show that the number of layers in the normalization at the encoder is equivalent to the number of terms in a matrix inverse Taylor series. Experimental results on real-world 3D point clouds show up to 2-3 dB gain over RAHT in energy compaction and 20-30\\% in bitrate reduction.","abstract_has_math":true,"creators":["Do, Viet Ho Tam Thuc"],"institution":null,"degree_name":null,"degree_level":null,"degree_discipline":null,"degree_department":null,"school":null,"contributors":[],"advisors":["Cheung, Gene"],"committee_chairs":[],"committee_members":[],"year":2023,"date_issued":"2023-08-04","date_published":"2023-08-04","updated_at":"2026-07-24T06:33:34Z","subjects":["Engineering","Computer science","Applied mathematics"],"languages":["en"],"rights":["Author owns copyright, except where explicitly noted. Please contact the author directly with licensing requests."],"rights_urls":[],"identifier_entries":[]},"links":{"outbound_url":"https://hdl.handle.net/10315/41338","outbound_label":"Handle","outbound_source":"dc:identifier.uri"},"metadata_groups":[{"id":"people","label":"People","entries":[{"key":"dc:contributor.advisor","label":"Advisor","values":["Cheung, Gene"]},{"key":"dc:creator","label":"Author","values":["Do, Viet Ho Tam Thuc"]}]},{"id":"academic_context","label":"Academic Context","entries":[{"key":"dc:date.accessioned","label":"Dc Date Accessioned","values":["2023-08-04T15:12:34Z"]},{"key":"dc:date.available","label":"Dc Date Available","values":["2023-08-04T15:12:34Z"]},{"key":"dc:date.issued","label":"Date","values":["2023-08-04"]},{"key":"dc:type","label":"Dc Type","values":["Electronic Thesis or Dissertation"]}]},{"id":"subjects_keywords","label":"Subjects and Keywords","entries":[{"key":"dc:subject","label":"Dc Subject","values":["Engineering","Computer science","Applied mathematics"]}]},{"id":"language_rights","label":"Language and Rights","entries":[{"key":"dc:language","label":"Dc Language","values":["en"]},{"key":"dc:rights","label":"Dc Rights","values":["Author owns copyright, except where explicitly noted. Please contact the author directly with licensing requests."]}]},{"id":"identifiers","label":"Identifiers","entries":[{"key":"dc:identifier.uri","label":"Identifier URI","values":["https://hdl.handle.net/10315/41338"]}]},{"id":"additional","label":"Additional Metadata","entries":[{"key":"dc:description.abstract","label":"Abstract","values":["We study 3D point cloud attribute compression using a volumetric approach: given a target volumetric attribute function $f : \\mathbb{R}^3 \\rightarrow \\mathbb{R}$, we quantize and encode parameter vector $\\theta$ that characterizes $f$ at the encoder, for reconstruction $f_{\\hat{\\theta}}(\\x)$ at known 3D points $\\x$'s at the decoder, where $\\hat{\\theta}$ is a quantized version of $\\theta$. Extending a previous work Region Adaptive Hierarchical Transform (RAHT) that employs piecewise constant functions to span a nested sequence of function spaces, we propose a feedforward linear network that implements higher-order B-spline bases spanning function spaces without eigen-decomposition. Feedforward network architecture means that the system is amenable to end-to-end neural learning. The key to our network is space-varying convolution, similar to a graph operator, whose weights are computed from the known 3D geometry for normalization. We show that the number of layers in the normalization at the encoder is equivalent to the number of terms in a matrix inverse Taylor series. Experimental results on real-world 3D point clouds show up to 2-3 dB gain over RAHT in energy compaction and 20-30\\% in bitrate reduction."]},{"key":"dc:title","label":"Title","values":["Volumetric Attribute Compression for 3D Point Clouds using Feedforward Network with Geometric Attention"]}]}],"canonical_facts":{"dc:contributor.advisor":["Cheung, Gene"],"dc:creator":["Do, Viet Ho Tam Thuc"],"dc:date.accessioned":["2023-08-04T15:12:34Z"],"dc:date.available":["2023-08-04T15:12:34Z"],"dc:date.issued":["2023-08-04"],"dc:description.abstract":["We study 3D point cloud attribute compression using a volumetric approach: given a target volumetric attribute function $f : \\mathbb{R}^3 \\rightarrow \\mathbb{R}$, we quantize and encode parameter vector $\\theta$ that characterizes $f$ at the encoder, for reconstruction $f_{\\hat{\\theta}}(\\x)$ at known 3D points $\\x$'s at the decoder, where $\\hat{\\theta}$ is a quantized version of $\\theta$. Extending a previous work Region Adaptive Hierarchical Transform (RAHT) that employs piecewise constant functions to span a nested sequence of function spaces, we propose a feedforward linear network that implements higher-order B-spline bases spanning function spaces without eigen-decomposition. Feedforward network architecture means that the system is amenable to end-to-end neural learning. The key to our network is space-varying convolution, similar to a graph operator, whose weights are computed from the known 3D geometry for normalization. We show that the number of layers in the normalization at the encoder is equivalent to the number of terms in a matrix inverse Taylor series. Experimental results on real-world 3D point clouds show up to 2-3 dB gain over RAHT in energy compaction and 20-30\\% in bitrate reduction."],"dc:identifier.uri":["https://hdl.handle.net/10315/41338"],"dc:language":["en"],"dc:rights":["Author owns copyright, except where explicitly noted. Please contact the author directly with licensing requests."],"dc:subject":["Engineering","Computer science","Applied mathematics"],"dc:title":["Volumetric Attribute Compression for 3D Point Clouds using Feedforward Network with Geometric Attention"],"dc:type":["Electronic Thesis or Dissertation"]},"updated_at":"2026-07-24T06:33:34Z"}