科技创新与应用2025，Vol.15Issue(11)：152-155,4.DOI:10.19981/j.CN23-1581/G3.2025.11.037
基于SFM与高斯渲染优化的建筑物三维重建方法
摘要
Abstract
To address the increasing demand for high-precision 3D modeling,this study proposes an enhanced reconstruction framework.The method focuses on optimizing camera pose estimation,point cloud processing,mesh generation,and texture mapping.During point cloud generation,a rendering model based on continuous Gaussian distribution is introduced to adjust point-wise weights and control spatial density.This approach improves reconstruction performance,particularly in edge regions and low-texture areas.During mesh modeling,a global reconstruction algorithm using surface orientation is combined with a multi-view texture fusion method.This integration enhances both the geometric continuity and visual coherence of the 3D models.Experimental results demonstrate that the proposed method achieves higher accuracy and visual quality than conventional techniques in complex scenarios.It is especially suitable for high-quality 3D modeling in architectural environments.关键词
三维重建/高斯渲染/点云优化/泊松表面重建/纹理融合Key words
3D reconstruction/gaussian rendering/point cloud optimization/poisson surface reconstruction/texture fusion分类
信息技术与安全科学引用本文复制引用
周健,罗玉恒..基于SFM与高斯渲染优化的建筑物三维重建方法[J].科技创新与应用,2025,15(11):152-155,4.基金项目
2025年度甘肃省级科技计划自然科学基金(25JRRE030) （25JRRE030）
2023年甘肃省科技厅自然科学基金(23JRRE0731) （23JRRE0731）
2023年甘肃省高校教师创新基金(2023A248) （2023A248）
2022年度甘肃林业职业技术学院院级基金(GSLY2022-A04) （GSLY2022-A04）
