3DRMF: No-reference 3D point cloud quality assessment with Rotation Mamba Fusion.
Saved in:
| Title: | 3DRMF: No-reference 3D point cloud quality assessment with Rotation Mamba Fusion. |
|---|---|
| Authors: | Li, Tiansong1 (AUTHOR) tiansongli@cqnu.edu.cn, Yang, Hao1 (AUTHOR), Li, Fei1 (AUTHOR), Cui, Shaoguo1 (AUTHOR), Wang, Hongkui2 (AUTHOR), Yu, Li3 (AUTHOR) |
| Source: | Displays. Dec2025, Vol. 90, pN.PAG-N.PAG. 1p. |
| Subjects: | Point cloud, Multisensor data fusion, Virtual reality, Quality (Philosophy) |
| Abstract: | With the rapid development of 3D data acquisition technologies, point clouds have been widely applied in fields such as virtual reality, augmented reality, and autonomous driving, etc. However, the distortions are inevitably introduced during the acquisition, transmission, and processing of point clouds, which not only degrade visual quality but also impair the performance of downstream tasks. The quality degradation of colored point clouds typically involves distortions in geometric attributes, color attributes, and their fusion. How to accurately quantify the perceptual quality of point clouds to achieve application-driven optimization remains a challenging problem. To address this, we propose a no-reference point cloud quality assessment (NR-PCQA) model, named 3DRMF, which consists of feature embedding and feature fusion. First, PointNet++ is utilized to extract geometric and color features separately. Then, in the feature embedding stage, a Group Channel Attention (GCA) module and a Multi-Level Feature Interaction (MLFI) module are designed to efficiently aggregate geometric and color features semantically through an attention mechanism. In the feature fusion stage, a Rotational Mamba Fusion (RMF) module based on Mamba is proposed to enable the interactive fusion between high-level global features. Finally, a multilayer perceptron (MLP) is employed to predict the quality score for point cloud assessment. Experimental results demonstrate that the proposed 3DRMF model outperforms the state-of-the-art NR-PCQA methods, achieving superior performance. The code of the proposed model is publicly available at https://github.com/yh-dyb/3DRMF/. • A novel 3DRMF model is proposed for no-reference point cloud quality assessment. • The GCA module is designed to achieve aggregation of geometric and color features. • The MLFI module is designed to realize cross-level feature interaction. • The RMF module is proposed to enable interactive fusion between high-level global features. [ABSTRACT FROM AUTHOR] |
| Copyright of Displays is the property of Elsevier B.V. and its content may not be copied or emailed to multiple sites without the copyright holder's express written permission. Additionally, content may not be used with any artificial intelligence tools or machine learning technologies. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.) | |
| Database: | Engineering Source |
| FullText | Text: Availability: 0 |
|---|---|
| Header | DbId: egs DbLabel: Engineering Source An: 188445422 AccessLevel: 6 PubType: Academic Journal PubTypeId: academicJournal PreciseRelevancyScore: 0 |
| IllustrationInfo | |
| Items | – Name: Title Label: Title Group: Ti Data: 3DRMF: No-reference 3D point cloud quality assessment with Rotation Mamba Fusion. – Name: Author Label: Authors Group: Au Data: <searchLink fieldCode="AR" term="%22Li%2C+Tiansong%22">Li, Tiansong</searchLink><relatesTo>1</relatesTo> (AUTHOR)<i> tiansongli@cqnu.edu.cn</i><br /><searchLink fieldCode="AR" term="%22Yang%2C+Hao%22">Yang, Hao</searchLink><relatesTo>1</relatesTo> (AUTHOR)<br /><searchLink fieldCode="AR" term="%22Li%2C+Fei%22">Li, Fei</searchLink><relatesTo>1</relatesTo> (AUTHOR)<br /><searchLink fieldCode="AR" term="%22Cui%2C+Shaoguo%22">Cui, Shaoguo</searchLink><relatesTo>1</relatesTo> (AUTHOR)<br /><searchLink fieldCode="AR" term="%22Wang%2C+Hongkui%22">Wang, Hongkui</searchLink><relatesTo>2</relatesTo> (AUTHOR)<br /><searchLink fieldCode="AR" term="%22Yu%2C+Li%22">Yu, Li</searchLink><relatesTo>3</relatesTo> (AUTHOR) – Name: TitleSource Label: Source Group: Src Data: <searchLink fieldCode="JN" term="%22Displays%22">Displays</searchLink>. Dec2025, Vol. 90, pN.PAG-N.PAG. 1p. – Name: Subject Label: Subjects Group: Su Data: <searchLink fieldCode="DE" term="%22Point+cloud%22">Point cloud</searchLink><br /><searchLink fieldCode="DE" term="%22Multisensor+data+fusion%22">Multisensor data fusion</searchLink><br /><searchLink fieldCode="DE" term="%22Virtual+reality%22">Virtual reality</searchLink><br /><searchLink fieldCode="DE" term="%22Quality+%28Philosophy%29%22">Quality (Philosophy)</searchLink> – Name: Abstract Label: Abstract Group: Ab Data: With the rapid development of 3D data acquisition technologies, point clouds have been widely applied in fields such as virtual reality, augmented reality, and autonomous driving, etc. However, the distortions are inevitably introduced during the acquisition, transmission, and processing of point clouds, which not only degrade visual quality but also impair the performance of downstream tasks. The quality degradation of colored point clouds typically involves distortions in geometric attributes, color attributes, and their fusion. How to accurately quantify the perceptual quality of point clouds to achieve application-driven optimization remains a challenging problem. To address this, we propose a no-reference point cloud quality assessment (NR-PCQA) model, named 3DRMF, which consists of feature embedding and feature fusion. First, PointNet++ is utilized to extract geometric and color features separately. Then, in the feature embedding stage, a Group Channel Attention (GCA) module and a Multi-Level Feature Interaction (MLFI) module are designed to efficiently aggregate geometric and color features semantically through an attention mechanism. In the feature fusion stage, a Rotational Mamba Fusion (RMF) module based on Mamba is proposed to enable the interactive fusion between high-level global features. Finally, a multilayer perceptron (MLP) is employed to predict the quality score for point cloud assessment. Experimental results demonstrate that the proposed 3DRMF model outperforms the state-of-the-art NR-PCQA methods, achieving superior performance. The code of the proposed model is publicly available at https://github.com/yh-dyb/3DRMF/. • A novel 3DRMF model is proposed for no-reference point cloud quality assessment. • The GCA module is designed to achieve aggregation of geometric and color features. • The MLFI module is designed to realize cross-level feature interaction. • The RMF module is proposed to enable interactive fusion between high-level global features. [ABSTRACT FROM AUTHOR] – Name: AbstractSuppliedCopyright Label: Group: Ab Data: <i>Copyright of Displays is the property of Elsevier B.V. and its content may not be copied or emailed to multiple sites without the copyright holder's express written permission. Additionally, content may not be used with any artificial intelligence tools or machine learning technologies. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract.</i> (Copyright applies to all Abstracts.) |
| PLink | https://search.ebscohost.com/login.aspx?direct=true&site=eds-live&db=egs&AN=188445422 |
| RecordInfo | BibRecord: BibEntity: Identifiers: – Type: doi Value: 10.1016/j.displa.2025.103137 Languages: – Code: eng Text: English PhysicalDescription: Pagination: PageCount: 1 StartPage: N.PAG Subjects: – SubjectFull: Point cloud Type: general – SubjectFull: Multisensor data fusion Type: general – SubjectFull: Virtual reality Type: general – SubjectFull: Quality (Philosophy) Type: general Titles: – TitleFull: 3DRMF: No-reference 3D point cloud quality assessment with Rotation Mamba Fusion. Type: main BibRelationships: HasContributorRelationships: – PersonEntity: Name: NameFull: Li, Tiansong – PersonEntity: Name: NameFull: Yang, Hao – PersonEntity: Name: NameFull: Li, Fei – PersonEntity: Name: NameFull: Cui, Shaoguo – PersonEntity: Name: NameFull: Wang, Hongkui – PersonEntity: Name: NameFull: Yu, Li IsPartOfRelationships: – BibEntity: Dates: – D: 01 M: 12 Text: Dec2025 Type: published Y: 2025 Identifiers: – Type: issn-print Value: 01419382 Numbering: – Type: volume Value: 90 Titles: – TitleFull: Displays Type: main |
| ResultId | 1 |
