[Documentation] [TitleIndex] [WordIndex

Package Summary

Filters for the sensor_msgs/PointCloud2 based on the filters and sensor_filters chains

Overview

Wrappers for some of the pcl filters ROS messages. The implementation and usage is based on the filter and sensor_filter packages, so it is different from the wrappers of the PCL filters provided by the package pcl_ros.

All the parameters are settable from the config file, but also online through the dynamic_reconfigure server.

Usage example

See launch and config folders

Filters list

PassThroughFilterPointCloud2

Wrapper for the pcl::PassThrough filter

Params

- active (bool, default: true) Activate the filter or not.

- input_frame (str, default: "") The input TF frame the data should be transformed into before processing

- output_frame (str, default: "") The output TF frame the data should be transformed into after processing

- keep_organized (bool, default: true) Keep the point cloud organized (`pcl::FilterIndices<PointT>::setKeepOrganized (bool keep_organized)`

- negative (bool, default: false) Set to true to return the data outside the min max limits

- filter_field_name (str, default: z) The field to be used for filtering data

- filter_limit_min (double, default: 0) The minimum allowed field value a point will be considered

- filter_limit_min (double, default: 1) The maximum allowed field value a point will be considered

CropBoxFilterPointCloud2

Wrapper for the pcl::CropBox filter.

Warning pcl::CrobBox parameter keep_organized is broken on ROS melodic (on noetic it is ok).

Params

- active (bool, default: true) Activate the filter or not.

- input_frame (str, default: "") The input TF frame the data should be transformed into before processing

- output_frame (str, default: "") The output TF frame the data should be transformed into after processing

- keep_organized (bool, default: true) Keep the point cloud organized (`pcl::FilterIndices<PointT>::setKeepOrganized (bool keep_organized)`

- negative (bool, default: false) Set to true to return the data outside the min max limits

- min_x (double, default: -1.0) The minimum allowed x value a point will be considered from. Range: -1000.0 to 1000.0

- max_x (double, default: -1.0) The maximum allowed x value a point will be considered from. Range: -1000.0 to 1000.0

- min_y (double, default: -1.0) The minimum allowed y value a point will be considered from. Range: -1000.0 to 1000.0

- max_y (double, default: -1.0) The maximum allowed y value a point will be considered from. Range: -1000.0 to 1000.0

- min_z (double, default: -1.0) The minimum allowed z value a point will be considered from. Range: -1000.0 to 1000.0

- max_z (double, default: -1.0) The maximum allowed z value a point will be considered from. Range: -1000.0 to 1000.0

VoxelGridFilterPointCloud2

Wrapper for the pcl::VoxelGrid filter.

Params

- active (bool, default: true) Activate the filter or not.

- input_frame (str, default: "") The input TF frame the data should be transformed into before processing

- output_frame (str, default: "") The output TF frame the data should be transformed into after processing

- negative (bool, default: false) Set to true to return the data outside the min max limits

- leaf_size_x (double, default: 0.01) The size of a leaf (on x) used for downsampling. Range: 0.0 to 1.0

- leaf_size_y (double, default: 0.01) The size of a leaf (on y) used for downsampling. Range: 0.0 to 1.0

- leaf_size_z (double, default: 0.01) The size of a leaf (on z) used for downsampling. Range: 0.0 to 1.0

- min_points_per_voxel (int, default:0) Set the minimum number of points required for a voxel to be used

- downsample_all_data (int, default:0) Set to true if all fields need to be downsampled, or false if just XYZ

- filter_field_name (str, default: z) The field to be used for filtering data, acting like a passthrough

- filter_limit_min (double, default: 0) The minimum allowed field value a point will be considered

- filter_limit_min (double, default: 1) The maximum allowed field value a point will be considered

SacSegmentationExtractFilterPointCloud2

Wrapper to extract a geometric model with pcl::SACSegmentation and pcl::ExtractIndices

Params

- active (bool, default: true) Activate the filter or not

- input_frame (str, default: "") The input TF frame the data should be transformed into before processing

- output_frame (str, default: "") The output TF frame the data should be transformed into after processing

- negative (bool, default: false) Set whether to filter out (remove) the model (true) or all the rest (false)

- model_type (int, default: 16) Geometric model to look for. Default to SACMODEL_NORMAL_PARALLEL_PLANE. Check pcl official doc

- method_type (int, default: 0) Segmentation model to use for. Default to SAC_RANSAC . Check pcl official doc

- axis_x (double, default: 0.0) The x component of the normal to the model to be removed. Range: 0.0 to 1.0

- axis_y (double, default: 0.0) The y component of the normal to the model to be removed. Range: 0.0 to 1.0

- axis_z (double, default: 1.0) The z component of the normal to the model to be removed. Range: 0.0 to 1.0

- eps_angle (double, default: 0.15) Tolerance angle (rad) to the model to be considered normal to the axis. Range: -3.15 to 3.15

- distance_threshold (double, default: 0.01) Range: 0 to 10

- optimize_coefficents (bool, default: 0.01) Optimize the coefficents or not.

- max_iterations (bool, default: 50)

- probability (bool, default: 0.99)

- min_radius (bool, default: -1)

- max_radius (bool, default: 1000)


2024-11-16 14:50