Bai Lizhi, Tian Chunqi, Yang Jun, Zhang Siyu, Suganuma Masanori, Okatani Takayuki
IEEE Trans Vis Comput Graph. 2026 Feb;32(2):1452-1466. doi: 10.1109/TVCG.2025.3616173.
3D Gaussian Splatting has emerged as a promising technique for high-quality 3D rendering, leading to increasing interest in integrating 3DGS into realism SLAM systems. However, existing methods face challenges such as Gaussian primitives redundancy, forgetting problem during continuous optimization, and difficulty in initializing primitives in monocular case due to lack of depth information. In order to achieve efficient and photorealistic mapping, we propose RP-SLAM, a 3D Gaussian splatting-based vision SLAM method for monocular and RGB-D cameras. RP-SLAM decouples camera poses estimation from Gaussian primitives optimization and consists of three key components. Firstly, we propose an efficient incremental mapping approach to achieve a compact and accurate representation of the scene through adaptive sampling and Gaussian primitives filtering. Secondly, a dynamic window optimization method is proposed to mitigate the forgetting problem and improve map consistency. Finally, for the monocular case, a monocular keyframe initialization method based on sparse point cloud is proposed to improve the initialization accuracy of Gaussian primitives, which provides a geometric basis for subsequent optimization. The results of numerous experiments demonstrate that RP-SLAM achieves state-of-the-art map rendering accuracy while ensuring real-time performance and model compactness.
3D高斯点渲染已成为一种用于高质量3D渲染的有前途的技术，这使得人们对将3D高斯点渲染集成到真实感SLAM系统中的兴趣与日俱增。然而，现有方法面临着诸如高斯基元冗余、连续优化过程中的遗忘问题以及在单目情况下由于缺乏深度信息而难以初始化基元等挑战。为了实现高效且逼真的映射，我们提出了RP-SLAM，这是一种基于3D高斯点渲染的用于单目和RGB-D相机的视觉SLAM方法。RP-SLAM将相机位姿估计与高斯基元优化解耦，并由三个关键组件组成。首先，我们提出一种高效的增量映射方法，通过自适应采样和高斯基元滤波来实现对场景的紧凑且准确的表示。其次，提出了一种动态窗口优化方法来减轻遗忘问题并提高地图一致性。最后，对于单目情况，提出了一种基于稀疏点云的单目关键帧初始化方法，以提高高斯基元的初始化精度，这为后续优化提供了几何基础。大量实验结果表明，RP-SLAM在确保实时性能和模型紧凑性的同时，实现了领先的地图渲染精度。
