https://github.com/nebula-beta/MonocularSfM
Monocular Structure from Motion
https://github.com/nebula-beta/MonocularSfM
cpp linux opencv sfm structure-from-motion
Last synced: 3 months ago
JSON representation
Monocular Structure from Motion
- Host: GitHub
- URL: https://github.com/nebula-beta/MonocularSfM
- Owner: nebula-beta
- Created: 2018-11-15T06:44:20.000Z (over 7 years ago)
- Default Branch: master
- Last Pushed: 2020-10-11T12:06:49.000Z (over 5 years ago)
- Last Synced: 2024-05-03T08:03:41.349Z (almost 2 years ago)
- Topics: cpp, linux, opencv, sfm, structure-from-motion
- Language: C
- Homepage:
- Size: 4.73 MB
- Stars: 151
- Watchers: 5
- Forks: 32
- Open Issues: 2
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
- awesome-computer-vision-resources - [github
README
# 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
### person-hall
### 东北大学
Number points3D : `542084`
Number images : `1329`
Mean reprojection error : `0.33772 [px]`
## Dependencies
* [Eigen](http://eigen.tuxfamily.org) version 3.2
* [OpenCV](http://opencv.org) version 3.x or higher
* [Ceres](http://ceres-solver.org) version 1.10 or higher
## 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](https://onedrive.live.com/?authkey=%21AAQumsDDwZBIW3w&id=C58A258D760E1B58%2146879&cid=C58A258D760E1B58), which provided by [COLMAP](https://colmap.github.io/datasets.html#datasets)
* **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](https://github.com/nebula-beta/MonocularSfM/wiki) page
## Citations
[1] Snavely N, Seitz S M, Szeliski R. [Photo Tourism: Exploring Photo Collections In 3D](http://phototour.cs.washington.edu/Photo_Tourism.pdf)[J]. Acm Transactions on Graphics, 2006, 25(3):págs. 835-846.
[2] Wu C. [Towards Linear-Time Incremental Structure from Motion](http://ccwu.me/vsfm/vsfm.pdf)[C]// International Conference on 3d Vision. IEEE Computer Society, 2013:127-134.
[3] Schönberger J L, Frahm J M. [Structure-from-Motion Revisited](https://demuc.de/papers/schoenberger2016sfm.pdf)[C]// Computer Vision and Pattern Recognition. IEEE, 2016.