DBScan-PCL-Optimized

This project is taken from: Navarro-Hinojosa, Octavio, y Moisés Alencastre-Miranda. "DBSCAN modificado con Octrees para agrupar nubes de puntos en tiempo real." Research in Computing Science, Vol. 114: Advances in Image Processing and Computer Vision, pp. 173–186, 2016. Github: https://github.com/Hagen23/DBScan_Octrees

It was modified with:

• It was added a CMakeList.txt for cmake compilation with PCL 1.8.1 (support 1.9.1)
• It was added an argument param options
• It was added a pcl visualizer
• It was deleted the Glut visualizer
• It was added a cluster saving method
• It was added a cluster coloring method
• It was replaced the input file from CSV to PCD
• It was added a cluster coloring method for original color of the point cloud

Input file structure support

• .pcd
• .ply
• .txt
• .xyz

Output file structure (default = .pcd)

• cloud_cluster_#.txt:

    x y z r g b
  

Example

---

Compilation

• Set "YOUR OWN" PCL Build DIR in CMakeList.txt e.g: /opt/pcl-1.8.1/build and save it.
• Create a "build" folder

in the main folder:

- cd build  
- cmake ../
- make

Test

cd /build/bin
./dbscan <pcd file> <octree resolution> <eps> <min Aux Pts> <min Pts> <output dir> <output extension (optional)>

pcd file = path to point_cloud.pcd
octree resolution = 124
eps = 40
min Pts = 4
max Pts = 5
output dir = path to save
output extension (optional) = pcd (default) --> you can set ply, txt or xyz

Example:
./dbscan /home/xXx/Downloads/point_cloud.pcd 124 40 5 5 /home/xXx/Downloads/clusters     

¡You can modify the parameters to obtain better results!
I recommend modifying only the eps value, with 40 - 60 you can get better clusters.

Note

If you do not want to see the output clusters on PCL Viewer, set:

bool showClusters = False; //Main function