EUVIMEDEUVIMED entdecken
Crossref

DNGaussian++: Improving Sparse-View Gaussian Radiance Fields With Depth Normalization

Jiahe Li, Jiawei Zhang, Xiaohan Yu, Xiao Bai, Jin Zheng, Xin Ning, Lin Gu

IEEE Transactions on Pattern Analysis and Machine Intelligence · 2026-07 · 48 (7) · 7622-7639

DOI: 10.1109/tpami.2026.3664307 ↗

Kurz gefasst

Fachartikel in der Zeitschrift IEEE Transactions on Pattern Analysis and Machine Intelligence, 2026-07. Verfasst von Jiahe Li und 6 weiteren Autor:innen. Die Arbeit wurde bisher 2-mal zitiert (Crossref) und nennt 96 Literaturangaben. Gefördert von: National Natural Science Foundation of China, R&D Program of Jiangxi Province, China, JST Moonshot.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Synthesizing novel views from sparse views has achieved impressive advances with radiance fields, yet prevailing methods suffer from high consumption or insufficient refinement capability. This paper introduces DNGaussian, a depth-regularized framework based on 3D Gaussian Splatting, offering real-time and high-quality few-shot novel view synthesis at low costs. Our motivation stems from the remarkable advancement of recent 3D Gaussian Splatting, despite it will encounter a geometry degradation when input views decrease. In the Gaussian radiance fields, we find this degradation in scene geometry primarily lined to the positioning of Gaussian primitives and can be mitigated by depth constraint. Consequently, we propose a Hard and Soft Depth Regularization to restore accurate scene geometry under coarse monocular depth supervision while maintaining a fine-grained color appearance. To further refine detailed geometry, we introduce Global-Local Depth Normalization, enhancing the focus on small local depth changes. Although DNGaussian shows impressive performance, its patch-wise regularization obscures the inconsistency in cross-patch errors. Additionally, primitives can still be irreversibly trapped in local minima under sparse views, even if depth regularization is applied. In this paper, we propose an extended version, DNGaussian++. First, a Geometry Instance Regularizer is developed to enable depth regularization for continuous consistency by exploiting reliable instance-level depth cues. Leveraging the depth gradient guidance, we then propose a Depth-Guided Geometry Reorganization to address the aforementioned local minima problem with high representation efficiency. Extensive experiments show that DNGaussian++ exhibits state-of-the-art performance in multiple datasets and scenarios with high efficiency, and the broad applicability and effectiveness are verified on various backbones and tasks.

Förderung

  • National Natural Science Foundation of China
  • R&D Program of Jiangxi Province, China
  • JST Moonshot

Zitieren

APA

Li, J., Zhang, J., Yu, X., Bai, X., Zheng, J., Ning, X., & Gu, L. (2026). DNGaussian++: Improving Sparse-View Gaussian Radiance Fields With Depth Normalization. IEEE Transactions on Pattern Analysis and Machine Intelligence, 48(7), 7622-7639. https://doi.org/10.1109/tpami.2026.3664307

Vancouver

Li J, Zhang J, Yu X, Bai X, Zheng J, Ning X, et al. DNGaussian++: Improving Sparse-View Gaussian Radiance Fields With Depth Normalization. IEEE Transactions on Pattern Analysis and Machine Intelligence. 2026;48(7):7622-7639. doi:10.1109/tpami.2026.3664307

BibTeX
@article{Li2026,
  author = {Li, Jiahe and Zhang, Jiawei and Yu, Xiaohan and Bai, Xiao and Zheng, Jin and Ning, Xin and Gu, Lin},
  title = {DNGaussian++: Improving Sparse-View Gaussian Radiance Fields With Depth Normalization},
  journal = {IEEE Transactions on Pattern Analysis and Machine Intelligence},
  year = {2026},
  volume = {48},
  number = {7},
  pages = {7622--7639},
  doi = {10.1109/tpami.2026.3664307}
}
Kostenlos starten →