Skip to content

Latest commit

 

History

History
83 lines (53 loc) · 3.14 KB

README.md

File metadata and controls

83 lines (53 loc) · 3.14 KB
Raw

MonocularSfM : Monocular Structure from Motion

Introuction

MonocularSfm是一个三维重建的程序, 可以对有序或者无序的图片进行三维重建.

程序的输入是图片和相机参数(包括相机内参fx, fy, cx, fy和畸变参数k1, k2, p1, p2[可选]).

程序的输出是三维稀疏点云和已注册图像的投影矩阵.

south-building

Image text

person-hall

Image text

东北大学

Image text Image text

Number points3D : 542084

Number images : 1329

Mean reprojection error : 0.33772 [px]

Dependencies

Building

mkdir build && cd build
cmake ..
make -j3

How to Run

# step1 : 提取特征
./FeatureExtraction ../config/NEU.yaml

# step2 : 计算匹配(根据数据集的不同，决定使用**顺序匹配**或者是**暴力匹配**, 通过修改config中配置文件的参数来实现)
./ComputeMatches ../config/NEU.yaml

# step3 : 检查匹配, 通过显示不同图像之间的匹配对, 来确认前两步是否正确(可跳过).
./CheckMatches  ../config/NEU.yaml

# step4 : 重建
./Reconstruction ../config/NEU.yaml

更多细节，请查看config文件夹下的yaml文件

或者直接使用python脚本文件pipeline.py

./pipeline ./config/NEU.yaml

Dataset

You can download the datasets from here, which provided by COLMAP

  • Gerrard Hall: 100 high-resolution images of the “Gerrard” hall at UNC Chapel Hill, which is the building right next to the “South” building. The images are taken with the same camera but different focus using a wide-angle lens.
  • Graham Hall: 1273 high-resolution images of the interior and exterior of “Graham” memorial hall at UNC Chapel Hill. The images are taken with the same camera but different focus using a wide-angle lens.
  • Person Hall: 330 high-resolution images of the “Person” hall at UNC Chapel Hill. The images are taken with the same camera using a wide-angle lens.
  • South Building: 128 images of the “South” building at UNC Chapel Hill. The images are taken with the same camera, kindly provided by Christopher Zach.

Knowledge

See the wiki page

Citations

[1] Snavely N, Seitz S M, Szeliski R. Photo Tourism: Exploring Photo Collections In 3D[J]. Acm Transactions on Graphics, 2006, 25(3):págs. 835-846.

[2] Wu C. Towards Linear-Time Incremental Structure from Motion[C]// International Conference on 3d Vision. IEEE Computer Society, 2013:127-134.

[3] Schönberger J L, Frahm J M. Structure-from-Motion Revisited[C]// Computer Vision and Pattern Recognition. IEEE, 2016.