downsample_filter#
Purpose#
The downsample_filter is a node that reduces the number of points.
Inner-workings / Algorithms#
Approximate Downsample Filter#
pcl::VoxelGridNearestCentroid is used. The algorithm is described in autoware_pcl_extensions
Random Downsample Filter#
pcl::RandomSample is used, which points are sampled with uniform probability.
Voxel Grid Downsample Filter#
pcl::VoxelGrid is used, which points in each voxel are approximated with their centroid.
Pickup Based Voxel Grid Downsample Filter#
This algorithm samples a single actual point existing within the voxel, not the centroid. The computation cost is low compared to Centroid Based Voxel Grid Filter.
Inputs / Outputs#
These implementations inherit autoware::pointcloud_preprocessor::Filter class, please refer README.
Parameters#
Note Parameters#
These implementations inherit autoware::pointcloud_preprocessor::Filter class, please refer README.
Core Parameters#
Approximate Downsample Filter#
| Name | Type | Default Value | Description |
|---|---|---|---|
voxel_size_x |
double | 0.3 | voxel size x [m] |
voxel_size_y |
double | 0.3 | voxel size y [m] |
voxel_size_z |
double | 0.1 | voxel size z [m] |
Random Downsample Filter#
| Name | Type | Description | Default | Range |
|---|---|---|---|---|
| sample_num | integer | number of indices to be sampled | 1500 | ≥0 |
Voxel Grid Downsample Filter#
| Name | Type | Description | Default | Range |
|---|---|---|---|---|
| voxel_size_x | float | the voxel size along x-axis [m] | 0.3 | ≥0 |
| voxel_size_y | float | the voxel size along y-axis [m] | 0.3 | ≥0 |
| voxel_size_z | float | the voxel size along z-axis [m] | 0.1 | ≥0 |
Pickup Based Voxel Grid Downsample Filter#
| Name | Type | Default Value | Description |
|---|---|---|---|
voxel_size_x |
double | 1.0 | voxel size x [m] |
voxel_size_y |
double | 1.0 | voxel size y [m] |
voxel_size_z |
double | 1.0 | voxel size z [m] |
Assumptions / Known limits#
This implementation uses the robin_hood.h hashing library by martinus, available under the MIT License at martinus/robin-hood-hashing on GitHub. Special thanks to martinus for this contribution.