这个dockerfile built的时候，有几个地方需要注意一下： 1，每一个RUN命令都是一层，执行‘docker-compose built'的时候，从修改的地方开始，之后每一行代码都会重新执行，而之前的会使用缓存。所以要把那些稳定的，耗时长的RUN命令放到前面。因为并不是built一次就成功了，重复下载会浪费大量的时间。。。

2，有很多种方式获得可执行程序？ 源码编译（这个一般从github上面） docker镜像（通过docker命令获取）。现在的经验是简单的（依赖少的）建议通过docker镜像（如果有点话），复杂的尽量还是通过源码编译的方式，因为有大量的依赖，使用docker镜像无法正常运行

This repository contains a Docker-based 3D reconstruction pipeline using COLMAP and OpenMVS.

1. Clone this repository:

git clone <repository-url>
cd <repository-directory>

2. Build the Docker container:

docker-compose build

3. Start the Docker container:

docker-compose up -d

To perform 3D reconstruction from images:

1. Place your images in the images/ directory.

2. Run the reconstruction script inside the Docker container:

docker-compose exec mvs bash -c "cd /workspace && ./scripts/reconstruct.sh"

3. Check the results:

docker-compose exec mvs bash -c "cd /workspace && ./scripts/check_results.sh"

The reconstruction pipeline produces the following outputs:

• Sparse Reconstruction (COLMAP):

  • output/colmap/sparse/ - Sparse point cloud and camera poses

• Dense Reconstruction (COLMAP):

  • output/colmap/dense/ - Dense point cloud
  • output/colmap/dense/fused.ply - Fused point cloud

• Mesh Reconstruction (OpenMVS):

  • output/openmvs/scene.mvs - Initial MVS scene
  • output/openmvs/scene_dense.mvs - Dense point cloud
  • output/openmvs/scene_mesh.mvs - Reconstructed mesh
  • output/openmvs/scene_mesh_refined.mvs - Refined mesh
  • output/openmvs/scene_mesh_textured.obj - Textured mesh (final output)

1. COLMAP Feature Extraction: Extract features from images
2. COLMAP Feature Matching: Match features between images
3. COLMAP Sparse Reconstruction: Create sparse point cloud and estimate camera poses
4. COLMAP Dense Reconstruction: Create dense point cloud
5. OpenMVS Interface: Convert COLMAP output to OpenMVS format
6. OpenMVS Dense Point Cloud: Densify the point cloud
7. OpenMVS Mesh Reconstruction: Create a mesh from the point cloud
8. OpenMVS Mesh Refinement: Refine the mesh
9. OpenMVS Mesh Texturing: Add texture to the mesh

The final textured mesh can be viewed with any 3D viewer that supports OBJ files, such as:

• MeshLab
• Blender

You can customize the reconstruction parameters by editing the scripts/reconstruct.sh file. The most common parameters to adjust are:

• Resolution levels
• Mesh refinement parameters
• Texture quality

• Docker
• Docker Compose

1. 构建Docker镜像：

docker-compose build

2. 启动容器：

docker-compose up -d

3. 进入容器：

docker-compose exec mvs bash

在容器内运行以下命令来验证安装：

# 验证COLMAP安装
colmap --version

# 验证OpenMVS安装
DensifyPointCloud --version

1. 将您的图像数据放在项目目录中
2. 进入容器后，您可以使用COLMAP进行特征提取和SfM重建
3. 使用OpenMVS进行密集重建和网格生成

• images/: 存放输入图像
• output/: 存放重建结果
• scripts/: 存放处理脚本