Cloud-subdivide: Subdivide point clouds into sub point clouds

This is a tool to subdivide a .pcd/.ply point cloud into a set of smaller sub point clouds with user defined dimensions. The tool starts at point P₀ = (x_min, y_min, z_min) of the point cloud and loops through the rest of the cloud from there. The resulting sub clouds are then saved with a .csv log file with some information on the sub clouds e.g. dimensions and number of points.

Visualization output example

Prerequisites

• Build environment [tested with GCC 7 and CMake 3.10]
• PointCloudLibrary [tested with PCL 1.8(.1)]
• Boost filesystem and Boost program_options

Build

$ cd path/to/build/dir
$ cmake path/to/src/dir
$ make

Example usage

$ cd path/to/build/dir
$ ./cloud-subdivide --input /path/to/your/point/cloud.ply --res 0.01 --x-dim 4.0 --y-dim 4.0 --z-dim 4.0

The output files will be dumped into /output_files in the root directory of this tool.

Inputs and options

Use input argument --help to show a list of input arguments.

$ ./cloud-subdivide --help

input argument	explanation
--help	List of input arguments. Default values will be used if there is no input for an argument.
--input arg	Set path to input file (.ply or .pcd). Use quotation marks (--input "/path to file") if file path has spaces.
--output arg	Set output file format (--output ply or --output pcd) for sub clouds [default: ply].
--res arg	Set resolution for octree box search [default: 0.01].
--x-dim arg	Set sub cloud dimension in x direction [default: 1.0].
--y-dim arg	Set sub cloud dimension in y direction [default: 1.0].
--z-dim arg	Set sub cloud dimension in z direction [default: 1.0].

Additional explanation

The user defined dimensions are the maximum dimensions for the sub point clouds. If no more points in the current search iteration and direction (z_{current max}, y_{current max}, x_{current max} based on box grid on input point cloud defined by x-dim, y-dim, z-dim) are left, the resulting sub cloud can be smaller than the defined value(s) for the respective dimension(s).

The point cloud is processed from P₀ → z_max → y_max → x_max.