软件导刊2025，Vol.24Issue(7)：13-21,9.DOI:10.11907/rjdk.251045
卫星可见光影像视差信息的表面重构综述
Review of Surface Reconstruction of Disparity Information from Satellite Visible Light Imagery
摘要
Abstract
Satellite visible light disparity information is essential for surface reconstruction through stereo matching to extract 3D information.Common methods include triangulation,which handles complex shapes but is sensitive to noise;voxel-based methods,suitable for dense point clouds but computationally expensive;and template prior methods,which use geometric templates for efficient reconstruction.Modeling prior methods generate complex surfaces using geometric rules but are inefficient for complex scenes.Deep learning methods,including seman-tic and local geometric priors,can handle complex scenes but require large datasets and are less robust to noise.With improved satellite image accuracy,semantic sensing technology,neural network-based adaptive reconstruction,and 3DGS technology hold significant potential for fu-ture development in satellite-based 3D reconstruction.关键词
卫星影像/可见光/立体匹配/表面重构/深度学习Key words
satellite imagery/visible light/stereo matching/surface reconstruction/deep learning分类
信息技术与安全科学引用本文复制引用
邓毅,解文彬,殷宏,张京晶,白玮..卫星可见光影像视差信息的表面重构综述[J].软件导刊,2025,24(7):13-21,9.基金项目
"十四五"某部科研项目(BHJ22SS1R048) （BHJ22SS1R048）
