BlockGaussian: Efficient Large-Scale Scene Novel View Synthesis via Adaptive Block-Based Gaussian Splatting

2025-04-12

AI Summary

To address severe tiling artifacts, low optimization efficiency, and supervision mismatch in 3D Gaussian Splatting (3DGS) reconstruction for large-scale scenes, this paper proposes a content-aware tiling and visibility-driven optimization framework. Our method introduces three key innovations: (1) a scene partitioning strategy adaptive to local content complexity; (2) an auxiliary point interpolation supervision alignment mechanism to mitigate supervision mismatch; and (3) pseudo-view geometric consistency constraints to effectively suppress spatial floating-point artifacts. The framework enables city-scale scene modeling on a single 24 GB GPU. Extensive evaluations on multiple large-scale datasets demonstrate an average PSNR improvement of 1.21 dB and a 5× speedup in optimization convergence, significantly enhancing both rendering quality and training efficiency.

Technology Category

• Computer Vision: 3D Computer Vision
• Constraint Satisfaction and Optimization: Distributed CSP/Optimization
• Search and Optimization: Learning to Search

Application Category

• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

The recent advancements in 3D Gaussian Splatting (3DGS) have demonstrated remarkable potential in novel view synthesis tasks. The divide-and-conquer paradigm has enabled large-scale scene reconstruction, but significant challenges remain in scene partitioning, optimization, and merging processes. This paper introduces BlockGaussian, a novel framework incorporating a content-aware scene partition strategy and visibility-aware block optimization to achieve efficient and high-quality large-scale scene reconstruction. Specifically, our approach considers the content-complexity variation across different regions and balances computational load during scene partitioning, enabling efficient scene reconstruction. To tackle the supervision mismatch issue during independent block optimization, we introduce auxiliary points during individual block optimization to align the ground-truth supervision, which enhances the reconstruction quality. Furthermore, we propose a pseudo-view geometry constraint that effectively mitigates rendering degradation caused by airspace floaters during block merging. Extensive experiments on large-scale scenes demonstrate that our approach achieves state-of-the-art performance in both reconstruction efficiency and rendering quality, with a 5x speedup in optimization and an average PSNR improvement of 1.21 dB on multiple benchmarks. Notably, BlockGaussian significantly reduces computational requirements, enabling large-scale scene reconstruction on a single 24GB VRAM device. The project page is available at https://github.com/SunshineWYC/BlockGaussian

Problem

• Efficient large-scale scene reconstruction using adaptive block partitioning
• Improving rendering quality via visibility-aware block optimization
• Reducing computational requirements for single-device large-scale reconstruction

Innovation

• Content-aware scene partition strategy
• Visibility-aware block optimization
• Pseudo-view geometry constraint