CityGaussianV2：針對大規模場景的高效且幾何精確重建

CityGaussianV2: Efficient and Geometrically Accurate Reconstruction for Large-Scale Scenes

November 1, 2024

作者: Yang Liu, Chuanchen Luo, Zhongkai Mao, Junran Peng, Zhaoxiang Zhang

cs.AI

摘要

最近，3D 高斯飛灑（3DGS）已經在輝度場重建方面帶來革命性變革，展現出高效率和高保真度的新視角合成。然而，在大型和複雜場景中準確表示表面，特別是由於 3DGS 的非結構化特性，仍然是一個重大挑戰。本文提出了 CityGaussianV2，這是一種針對大規模場景重建的新方法，解決了與幾何精度和效率相關的關鍵挑戰。借鑒 2D 高斯飛灑（2DGS）良好的泛化能力，我們解決了其收斂性和可擴展性問題。具體來說，我們實現了一種基於分解梯度的致密化和深度回歸技術，以消除模糊的藝術品並加速收斂。為了擴展規模，我們引入了一種延伸濾波器，以減輕 2DGS 退化引起的高斯計數爆炸。此外，我們對 CityGaussian 流程進行了優化，實現了高效的平行訓練，達到最多 10 倍的壓縮，至少節省 25% 的訓練時間，以及 50% 的記憶體使用減少。我們還在大型場景下建立了標準幾何基準。實驗結果表明，我們的方法在視覺質量、幾何精度以及存儲和訓練成本之間取得了有希望的平衡。項目頁面位於 https://dekuliutesla.github.io/CityGaussianV2/。

English

Recently, 3D Gaussian Splatting (3DGS) has revolutionized radiance field reconstruction, manifesting efficient and high-fidelity novel view synthesis. However, accurately representing surfaces, especially in large and complex scenarios, remains a significant challenge due to the unstructured nature of 3DGS. In this paper, we present CityGaussianV2, a novel approach for large-scale scene reconstruction that addresses critical challenges related to geometric accuracy and efficiency. Building on the favorable generalization capabilities of 2D Gaussian Splatting (2DGS), we address its convergence and scalability issues. Specifically, we implement a decomposed-gradient-based densification and depth regression technique to eliminate blurry artifacts and accelerate convergence. To scale up, we introduce an elongation filter that mitigates Gaussian count explosion caused by 2DGS degeneration. Furthermore, we optimize the CityGaussian pipeline for parallel training, achieving up to 10times compression, at least 25% savings in training time, and a 50% decrease in memory usage. We also established standard geometry benchmarks under large-scale scenes. Experimental results demonstrate that our method strikes a promising balance between visual quality, geometric accuracy, as well as storage and training costs. The project page is available at https://dekuliutesla.github.io/CityGaussianV2/.