基于深度学习的多视图立体匹配三维重建算法分析  

Analysis of Multi View Stereo Matching 3D Reconstruction Algorithm Based on Deep Learning

作　　者：黄洋 HUANG Yang(Nanning Vocational and Technical College,Guangxi 530028,China)

机构地区：[1]南宁职业技术学院,广西530028

出　　处：《电子技术（上海）》2024年第8期302-303,共2页Electronic Technology

摘　　要：阐述基于深度学习的多视图立体匹配三维重建算法,包括特征提取、微分单应性变换计算代价体、代价体正则化、深度图与重建点云。三维重建有利于学者们对场景做出观察、感受、学习和研究。This paper expounds a multi view stereo matching 3D reconstruction algorithm based on deep learning,including feature extraction,differential homography transformation to calculate cost volume,cost volume regularization,depth map and reconstructed point cloud.3D reconstruction is beneficial for scholars to observe,feel,learn,and study the scene.

关 键 词：深度学习 多视图 三维重建算法 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术] TP18[自动化与计算机技术—计算机科学与技术]