| Guo, Zhihao, Su, Jingxuan, Qian, Chenghao, Wang, Shenglin, Fan, Jinlong, Zhang, Jing, Zhou, Wei, Amirpour, Hadi, Zhao, Yunlong, Han, Liangxiu and Wang, Peng 2026. GP-GS: Gaussian processes densification for 3D Gaussian Splatting. Presented at: 22nd International Conference on Intelligent Computing, Toronto, ON, Canada, 22-26 July 2026. Published in: Li, Gang, Filipe, Joaquim and Xu, Zhiwei eds. Communications in Computer and Information Science. , vol.3037 Singapore: Springer, pp. 139-150. 10.1007/978-981-92-3548-3_12 |
Abstract
3D Gaussian Splatting (3DGS) enables photorealistic rendering but suffers from artefacts due to sparse Structure-from-Motion (SfM) initialisation. To address this limitation, we propose GP-GS, a Gaussian Process (GP) based densification framework for 3DGS optimisation. GP-GS formulates point cloud densification as a continuous regression problem, where a GP learns a local mapping from 2D pixel coordinates to 3D position and colour attributes. An adaptive neighbourhood-based sampling strategy generates candidate pixels for inference, while GP-predicted uncertainty is used to filter unreliable predictions, reducing noise and preserving geometric structure. Extensive experiments on synthetic and real-world benchmarks demonstrate that GP-GS consistently improves reconstruction quality and rendering fidelity, achieving up to 1.12 dB PSNR improvement over strong baselines.
| Item Type: | Conference or Workshop Item - published (Paper) |
|---|---|
| Date Type: | Publication |
| Status: | Published |
| Schools: | Schools > Computer Science & Informatics |
| Publisher: | Springer |
| ISSN: | 1865-0929 |
| Last Modified: | 27 Jul 2026 13:30 |
| URI: | https://orca.cardiff.ac.uk/id/eprint/188482 |
Actions (repository staff only)
 |
Edit Item |
Dimensions
Dimensions