Topological Feature Fusion for Multi-Channel No-Reference Point Cloud Quality Assessment
IEEE Signal Processing Letters · 2026 · 33 · 3207-3211
DOI: 10.1109/lsp.2026.3718778 ↗Kurz gefasst
Fachartikel in der Zeitschrift IEEE Signal Processing Letters, 2026. Verfasst von Zexi Li und 4 weiteren Autor:innen. Die Arbeit nennt 37 Literaturangaben. Gefördert von: National Natural Science Foundation of China, Basic and Applied Basic Research Foundation of Guangdong Province.
Abstract (Originalsprache)
Point Cloud Quality Assessment (PCQA) plays a crucial role in various applications, including autonomous driving and virtual reality, by providing important data evaluation metrics. However, existing PCQA methods have largely overlooked the application of topological features in point clouds. In this paper, we propose a novel Topological Feature Fusion-based Multi-Channel no-reference PCQA model (TFMC-PCQA) that integrates point, image, and topological features based on simplicial complexes. Specifically, we extract topological features of point clouds using simplicial complexes and integrate them with high-level semantic information to form a multi-channel feature extraction module. We then apply a cross-modal multi-head cross-attention module to effectively align local geometric and topological cues from point clouds with texture and semantic cues from projection images, achieving cross-modal feature complementarity and representation enhancement. Moreover, we design a structural adaptive fusion module to dynamically integrate the enhanced image and point-cloud representations. Experimental results on three public datasets demonstrate that our method significantly outperforms 16 state-of-the-art PCQA approaches, validating the effectiveness of embedding simplicial-complex features and high-level semantic information to enhance PCQA performance.
Förderung
- National Natural Science Foundation of China
- Basic and Applied Basic Research Foundation of Guangdong Province
Zitieren
APA
Li, Z., Xu, D., Peng, A., Zhong, Q., & Zheng, X. (2026). Topological Feature Fusion for Multi-Channel No-Reference Point Cloud Quality Assessment. IEEE Signal Processing Letters, 33, 3207-3211. https://doi.org/10.1109/lsp.2026.3718778
Vancouver
Li Z, Xu D, Peng A, Zhong Q, Zheng X. Topological Feature Fusion for Multi-Channel No-Reference Point Cloud Quality Assessment. IEEE Signal Processing Letters. 2026;33:3207-3211. doi:10.1109/lsp.2026.3718778
BibTeX
@article{Li2026,
author = {Li, Zexi and Xu, Dianjun and Peng, Aizhong and Zhong, Qiman and Zheng, Xianwei},
title = {Topological Feature Fusion for Multi-Channel No-Reference Point Cloud Quality Assessment},
journal = {IEEE Signal Processing Letters},
year = {2026},
volume = {33},
pages = {3207--3211},
doi = {10.1109/lsp.2026.3718778}
}
EUVIMED