1. MVE
项目主页 https://www.gcc.tu-darmstadt.de/home/proj/mve/
Github地址 https://github.com/simonfuhrmann/mve
#!/bin/bash workspace_path=/root/test_result/mve_result image_dir=${workspace_path}/${1} scene_dir=${workspace_path}/${2} mve=/root/misc_codes/mve/apps maxpixel=20000000 intrinsic_fp="2759.48,0,0,0.4950,0.4916,0.9983" # fountain-p11 intrinsic_tp="1520.40,0,0,0.4724,0.5143,0.9964" # temple intrinsic_eth_pipe="3430.27,0,0,0.5015,0.4969,1.0003" # eth3d pipes intrinsic_dtu="2892.33,0,0,0.5145,0.5159,1.0032" # dtu dataset intrinsic_tanks="" intrinsic=${intrinsic_tanks} # --init-intrinsics=${intrinsic} ${mve}/makescene/makescene --original --images-only ${image_dir} --max-pixels=${maxpixel} ${scene_dir} && # --fixed-intrinsics ${mve}/sfmrecon/sfmrecon --max-pixels=${maxpixel} --verbose-ba ${scene_dir} && ${mve}/dmrecon/dmrecon --neighbors=9 --scale=0 --max-pixels=${maxpixel} --local-neighbors=6 --keep-dz --progress=fancy ${scene_dir} && ${mve}/scene2pset/scene2pset -F0 ${scene_dir} ${scene_dir}/pset-L0.ply && ${mve}/fssrecon/fssrecon ${scene_dir}/pset-L0.ply ${scene_dir}/surface-L0.ply && ${mve}/meshclean/meshclean --threshold=8.0 --delete-scale ${scene_dir}/surface-L0.ply ${scene_dir}/surface-clean.ply
2. SMVS
项目主页 https://www.gcc.tu-darmstadt.de/home/proj/smvs/smvs.en.jsp
Github地址 https://github.com/flanggut/smvs
#!/bin/bash workspace_path=/root/test_result/smvs_result image_dir=${workspace_path}/${1} scene_dir=${workspace_path}/${2} mve=/root/misc_codes/mve/apps smvs=/root/misc_codes/smvs/smvsrecon maxpixel=2000000 intrinsic_fp="2759.48,0,0,0.4950,0.4916,0.9983" # fountain-p11 intrinsic_tp="1520.40,0,0,0.4724,0.5143,0.9964" # temple intrinsic_eth_pipe="3430.27,0,0,0.5015,0.4969,1.0003" # eth3d pipes intrinsic_dtu="2892.33,0,0,0.5145,0.5159,1.0032" # dtu dataset intrinsic_tanks="2304.00,0,0,0.5,0.5,1.0000" # Manually set intrinsic intrinsic=${intrinsic_fp} ${mve}/makescene/makescene --original --images-only ${image_dir} --max-pixels=${maxpixel} --init-intrinsics=${intrinsic} ${scene_dir} && ${mve}/sfmrecon/sfmrecon --max-pixels=${maxpixel} --fixed-intrinsics --verbose-ba ${scene_dir} && ${smvs} ${scene_dir} && ${mve}/fssrecon/fssrecon ${scene_dir}/pset-L0.ply ${scene_dir}/surface-L0.ply && ${mve}/meshclean/meshclean --threshold=8.0 --delete-scale ${scene_dir}/surface-L0.ply ${scene_dir}/surface-clean.ply
3. openMVG+openMVS
多视图立体几何基础库 openMVG https://github.com/openMVG/openMVG
稠密重建库 openMVS https://github.com/cdcseacave/openMVS
#!/bin/bash workspace_path=/root/test_result/openmvs_result/${1} image_dir=${workspace_path}/images recon_dir=${workspace_path}/reconstruct match_dir=${recon_dir}/matches openmvg=/root/misc_codes/openMVG/openmvg-bin/bin openmvs=/root/misc_codes/openMVS/openmvs-build/bin maxres=6400 minres=480 intrinsic_fp="2759.48;0;1520.69;0;2764.16;1006.81;0;0;1" # fountain-p11 intrinsic_tp="1520.40;0;302.32;0;1525.90;246.87;0;0;1" # temple intrinsic_eth_pipe="3430.27;0;3119.2;0;3429.23;2057.75;0;0;1" # eth3d pipes intrinsic_dtu="2892.33;0;823.21;0;2883.17;619.07;0;0;1" # for all dtu datasets intrinsic_tanks="2304.00;0;960;0;2304.00;540;0;0;1" # Manually set intrinsic intrinsic=${intrinsic_dtu} mkdir ${recon_dir} && mkdir ${match_dir} ${openmvg}/openMVG_main_SfMInit_ImageListing -i ${image_dir} -o ${match_dir} --camera_model 1 --intrinsics ${intrinsic} --group_camera_model 1 ${openmvg}/openMVG_main_ComputeFeatures -i ${match_dir}/sfm_data.json --outdir ${match_dir} --describerPreset HIGH ${openmvg}/openMVG_main_ComputeMatches -i ${match_dir}/sfm_data.json --out_dir ${match_dir} --nearest_matching_method ANNL2 ${openmvg}/openMVG_main_IncrementalSfM -i ${match_dir}/sfm_data.json --matchdir ${match_dir} --outdir ${recon_dir} --camera_model 1 --refineIntrinsics NONE # --refineIntrinsics "ADJUST_FOCAL_LENGTH|ADJUST_PRINCIPAL_POINT" ${openmvg}/openMVG_main_ComputeSfM_DataColor -i ${recon_dir}/sfm_data.bin -o ${recon_dir}/colorized.ply ${openmvg}/openMVG_main_ComputeStructureFromKnownPoses -i ${recon_dir}/sfm_data.bin --match_dir ${match_dir} --match_file ${match_dir}/matches.f.bin --output_file ${recon_dir}/robust.bin ${openmvg}/openMVG_main_ComputeSfM_DataColor -i ${recon_dir}/robust.bin -o ${recon_dir}/robust_colorized.ply # outfile is the file name to save converted result # outdir is the path to save undistorted images ${openmvg}/openMVG_main_openMVG2openMVS --sfmdata ${recon_dir}/sfm_data.bin --outfile ${recon_dir}/scene.mvs --outdir ${recon_dir} ${openmvs}/DensifyPointCloud --working-folder ${recon_dir} -i ${recon_dir}/scene.mvs --max-resolution=${maxres} --min-resolution=${minres} --number-views=6 # free-space-support is for textureless region ${openmvs}/ReconstructMesh --working-folder ${recon_dir} -i ${recon_dir}/scene_dense.mvs --free-space-support 1 ${openmvs}/RefineMesh --working-folder ${recon_dir} -i ${recon_dir}/scene_dense_mesh.mvs --min-resolution ${minres} --max-views 9 --scales 5 --planar-vertex-ratio 5 ${openmvs}/TextureMesh --working-folder ${recon_dir} -i ${recon_dir}/scene_dense_mesh.mvs --min-resolution ${minres} --cost-smoothness-ratio 0.3
4. COLMAP
Github地址 https://github.com/colmap/colmap
#!/bin/bash colmap=/root/misc_codes/colmap/colmap-bin/bin/colmap workspace=/root/test_result/colmap_result/${1} images=${workspace}/images database_path=${workspace}/database.db sparse_path=${workspace}/sparse dense_path=${workspace}/dense maxsize=2000 maxfeature=8192 intrinsic_fp="2759.48,2764.16,1520.69,1006.81" # fountain-p11 intrinsic_tp="1520.40,1525.90,302.32,246.87" # temple intrinsic_dtu="2892.33,2883.17,823.21,619.07" # all dtu dataset intrinsic_tanks="" intrinsic=${intrinsic_dtu} # --ImageReader.camera_params ${intrinsic} ${colmap} feature_extractor --database_path ${database_path} --image_path ${images} --ImageReader.camera_model PINHOLE --ImageReader.camera_params ${intrinsic} --ImageReader.single_camera 1 --SiftExtraction.max_image_size ${maxsize} --SiftExtraction.max_num_features ${maxfeature} ${colmap} exhaustive_matcher --database_path ${database_path} --SiftMatching.guided_matching 0 mkdir ${sparse_path} ${colmap} mapper --database_path ${database_path} --image_path ${images} --output_path ${sparse_path} --Mapper.ba_refine_principal_point false mkdir ${dense_path} && ${colmap} image_undistorter --image_path ${images} --input_path ${sparse_path}/0 --output_path ${dense_path} --output_type COLMAP --max_image_size ${maxsize} && ${colmap} patch_match_stereo --workspace_path ${dense_path} --workspace_format COLMAP --PatchMatchStereo.max_image_size ${maxsize} --PatchMatchStereo.window_radius 9 --PatchMatchStereo.geom_consistency 1 --PatchMatchStereo.filter_min_ncc 0.07 && ${colmap} stereo_fusion --workspace_path ${dense_path} --input_type geometric --output_path ${dense_path}/fused.ply && ${colmap} poisson_mesher --input_path ${dense_path}/fused.ply --output_path ${dense_path}/meshed-poisson.ply ${colmap} delaunay_mesher --input_path ${dense_path} --input_type dense --output_path ${dense_path}/meshed-delaunay.ply
colmap recon script(from tanks and temples)
#!/bin/bash # ---------------------------------------------------------------------------- # - TanksAndTemples Website Toolbox - # - http://www.tanksandtemples.org - # ---------------------------------------------------------------------------- # The MIT License (MIT) # # Copyright (c) 2017 # Arno Knapitsch <arno.knapitsch@gmail.com > # Jaesik Park <syncle@gmail.com> # Qian-Yi Zhou <Qianyi.Zhou@gmail.com> # Vladlen Koltun <vkoltun@gmail.com> # # Permission is hereby granted, free of charge, to any person obtaining a copy # of this software and associated documentation files (the "Software"), to deal # in the Software without restriction, including without limitation the rights # to use, copy, modify, merge, publish, distribute, sublicense, and/or sell # copies of the Software, and to permit persons to whom the Software is # furnished to do so, subject to the following conditions: # # The above copyright notice and this permission notice shall be included in # all copies or substantial portions of the Software. # # THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR # IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, # FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE # AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER # LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, # OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN # THE SOFTWARE. # ---------------------------------------------------------------------------- # # This script generates a COLMAP reconstruction from a numbe rof input imagess # Usage: sh get_colmap_reconstruction.sh <COLMAP-exe-directory> <image-set-directory> <project-directory> colmap_folder=$1/ iname=$2/ outf=$3/ DATABASE=${outf}sample_reconstruction.db PROJECT_PATH=${outf} mkdir -p ${PROJECT_PATH} mkdir -p ${PROJECT_PATH}/images cp -n ${iname}*.jpg ${PROJECT_PATH}/images ${colmap_folder}/colmap feature_extractor --database_path ${DATABASE} --image_path ${PROJECT_PATH}/images --ImageReader.camera_model RADIAL --ImageReader.single_camera 1 --SiftExtraction.use_gpu 1 ${colmap_folder}/colmap exhaustive_matcher --database_path ${DATABASE} --SiftMatching.use_gpu 1 mkdir ${PROJECT_PATH}/sparse ${colmap_folder}/colmap mapper --database_path ${DATABASE} --image_path ${PROJECT_PATH}/images --output_path ${PROJECT_PATH}/sparse mkdir ${PROJECT_PATH}/dense ${colmap_folder}/colmap image_undistorter --image_path ${PROJECT_PATH}/images --input_path ${PROJECT_PATH}/sparse/0/ --output_path ${PROJECT_PATH}/dense --output_type COLMAP --max_image_size 1500 ${colmap_folder}/colmap patch_match_stereo --workspace_path $PROJECT_PATH/dense --workspace_format COLMAP --PatchMatchStereo.geom_consistency true ${colmap_folder}/colmap stereo_fusion --workspace_path $PROJECT_PATH/dense --workspace_format COLMAP --input_type geometric --output_path $PROJECT_PATH/dense/fused.ply
上述几个脚本都放到了我的github上，地址是：https://github.com/philleer/program_test/tree/mvs_script
Ubuntu 18.04 LTS 亲自测试有效，欢迎补充