Deep learning-based point cloud upsampling: A survey of methodologies, performance comparisons, and noise robustness analysis

Yin, Yihang, Yu, Li, Zhou, Wei and Gabbouj, Moncef 2026. Deep learning-based point cloud upsampling: A survey of methodologies, performance comparisons, and noise robustness analysis. Neurocomputing 681 , 133316. 10.1016/j.neucom.2026.133316

Full text not available from this repository.

Abstract

Point cloud upsampling has emerged as a pivotal technology in 3D computer vision, addressing the essential need to reconstruct dense, uniform point distributions from sparse inputs for applications ranging from autonomous systems to digital content creation. This paper presents a comprehensive review of deep learning-based point cloud upsampling algorithms. We methodically categorize existing approaches into supervised and unsupervised learning frameworks, discussing representative architectures and analyzing their core designs. A key contribution lies in our systematic experimental comparisons of state-of-the-art methods, particularly emphasizing their noise robustness, an aspect critical for real-world applications but underexplored in prior reviews. By synthesizing insights from methodological innovations and empirical evaluations, this review provides guidelines for algorithm selection under specific application constraints and identifies open issues that should be addressed in the future.

Item Type:	Article
Date Type:	Publication
Status:	Published
Schools:	Schools > Computer Science & Informatics
Additional Information:	License information from Publisher: LICENSE 1: Title: This article is under embargo with an end date yet to be finalised.
Publisher:	Elsevier
ISSN:	0925-2312
Date of Acceptance:	10 March 2026
Last Modified:	20 Mar 2026 10:00
URI:	https://orca.cardiff.ac.uk/id/eprint/185888

Actions (repository staff only)

Edit Item