现代信息科技2026，Vol.10Issue(8)：49-52,4.DOI:10.19850/j.cnki.2096-4706.2026.08.009
基于特征匹配的无相机姿态三维场景重建方法
Pose-free 3D Scene Reconstruction Method Based on Feature Matching
邓永衡 1严华1
作者信息
- 1. 四川大学 电子信息学院,四川 成都 610065
- 折叠
摘要
Abstract
3D scene reconstruction has significant practical value.However,most reconstruction methods rely on known camera poses,while some pose-free approaches require ordered image inputs.To address the problem of 3D scene reconstruction under pose-free and unordered input conditions,a pose-free reconstruction method based on feature matching is proposed.The method computes image proximity through feature matching,constructs a weighted graph,and determines the sequence of camera pose estimation using a maximum spanning tree algorithm,thereby achieving progressive scene reconstruction.Experiments on the Tanks and Temples dataset demonstrate that the proposed method maintains high reconstruction accuracy and stability under unordered input conditions.The novel view synthesis quality and pose estimation accuracy are comparable to those of CF3DGS under ordered inputs,and significantly superior to its performance with unordered inputs,achieving an average PSNR of 31.409 across multiple test scenes.关键词
三维场景重建/特征匹配/姿态估计Key words
3D scene reconstruction/feature matching/pose estimation分类
信息技术与安全科学引用本文复制引用
邓永衡,严华..基于特征匹配的无相机姿态三维场景重建方法[J].现代信息科技,2026,10(8):49-52,4.
