Incremental Online Scene Reconstruction by 3D Gaussian Triangulation
2026-07-12 • Computer Vision and Pattern Recognition
Computer Vision and Pattern Recognition
AI summaryⓘ
The authors developed a new way to build and update detailed 3D models of scenes step-by-step without needing extra processing steps. Their method creates clear, detailed mesh surfaces directly from a set of 3D points called Gaussians, which helps with both high-quality images and accurate 3D shapes. They also introduced a way to keep the mesh accurate by aligning these points to flat surfaces and save computing power by locking parts of the model that are already done. Tests show their approach works better than older methods in both how the scenes look and how accurately they reconstruct the environment.
Incremental Scene Reconstruction3D Gaussian SplattingMesh ExtractionTriangulationImplicit FieldExplicit MeshSurface ReconstructionPlane-based Pulling ConstraintRendering QualityComputational Efficiency
Authors
Yanjin Zhu, Shaofan Liu, Jianke Zhu
Abstract
Incremental scene reconstruction is essential for real-world applications. Although 3D Gaussian Splatting shows strong potential, most existing approaches require offline conversion of the optimized Gaussians into an intermediate implicit field for explicit mesh extraction, which hinders seamless integration with downstream tasks. To address this limitation, we propose a novel online framework that incrementally reconstructs and updates high-fidelity explicit meshes by directly triangulating a dense geometric Gaussian representation, which supports both high-quality rendering and incremental surface reconstruction. Moreover, we present a direct meshing algorithm that efficiently extracts and updates the mesh from the Gaussian set. To ensure mesh accuracy, we enforce a plane-based pulling constraint that dynamically aligns 3D Gaussian primitives to the approximated local surface. Furthermore, our framework significantly reduces memory and computational overhead during long-sequence processing by dynamically freezing fully optimized historical regions. Experiments on public datasets demonstrate that our method outperforms conventional Gaussian-based methods on both rendering quality and reconstruction accuracy.