ubuntu20.04中使用openMVG和openMVS进行三维重建

需要指定相机焦距-f，大致可以由图像长乘以1.2估计得到，例如图像长宽为1280×720，则焦距估计为1280×1.2=1536。（或者可以点击图像属性看，一般自己拍的能显示焦距，网图的话可能就没有这个信息）

openMVG_main_SfMInit_ImageListing -i /home/zll/restruct/picture -o /home/zll/restruct/match -f 4800

openMVG_main_ComputeFeatures -i /home/zll/restruct/match/sfm_data.json -o /home/zll/restruct/match -m SIFT

openMVG_main_PairGenerator -i /home/zll/restruct/match/sfm_data.json -o /home/zll/restruct/match/pairs.bin

openMVG_main_ComputeMatches -i /home/zll/restruct/match/sfm_data.json -p /home/zll/restruct/match/pairs.bin -o /home/zll/restruct/match/matches.putatives.bin

openMVG_main_GeometricFilter  -i /home/zll/restruct/match/sfm_data.json -m /home/zll/restruct/match/matches.putatives.bin -o /home/zll/restruct/match/matches.f.bin

openMVG_main_SfM --sfm_engine INCREMENTAL --input_file /home/zll/restruct/match/sfm_data.json --match_dir /home/zll/restruct/match --output_dir /home/zll/restruct/reconstruction

openMVG_main_ComputeSfM_DataColor -i /home/zll/restruct/reconstruction/sfm_data.bin -o /home/zll/restruct/reconstruction/colorized.ply

openMVS可以很好地与openMVG对接，其主要的作用是进行稠密点云重建、网格重建与纹理映射。-d 4与–resolution-level=4的作用是为了简化运算，减小了重建的网格数量与精细程度

 openMVG_main_openMVG2openMVS -i /home/zll/restruct/reconstruction/sfm_data.bin -o /home/zll/restruct/mvs/scene.mvs -d /home/zll/restruct/picture

/usr/local/bin/OpenMVS/DensifyPointCloud scene.mvs --dense-config-file Densifu.ini --resolution-level 1 --number-views 7 -w /home/zll/restruct/mvs

/usr/local/bin/OpenMVS/ReconstructMesh scene_dense.mvs -w /home/zll/restruct/mvs

 /usr/local/bin/OpenMVS/RefineMesh --resolution-level=4 scene_dense_mesh.mvs -w /home/zll/restruct/mvs

/usr/local/bin/OpenMVS/TextureMesh scene_dense_mesh_refine.mvs --decimate 0.5 -w /home/zll/restruct/mvs

/usr/local/bin/OpenMVS/Viewer /home/zll/restruct/mvs/scene_dense_mesh_refine_texture.mvs 

 效果：