point_cloud_2_filters

Wrappers for some of the pcl filters for sensor_msgs/PointCloud2 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. Note that changing params with the dynamic_reconfigure server may take some seconds to have effect.

No ROS2 version (yet).

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
• pub_cloud(bool, default: false) Publish the cloud immediately after this filter. Note that this is a duplicate if the filter is the last in the chain. Useful for debug purposes and it will publish even if active is false.
• 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_max(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
• pub_cloud(bool, default: false) Publish the cloud immediately after this filter. Note that this is a duplicate if the filter is the last in the chain. Useful for debug purposes and it will publish even if active is false.
• 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
• pub_cloud(bool, default: false) Publish the cloud immediately after this filter. Note that this is a duplicate if the filter is the last in the chain. Useful for debug purposes and it will publish even if active is false.
• 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: “”) The field to be used for filtering data, acting like a passthrough. Empty for not using
• filter_limit_min(double, default: -FLT_MAX) The minimum allowed field value a point will be considered
• filter_limit_max(double, default: FLT_MAX) 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
• pub_cloud(bool, default: false) Publish the cloud immediately after this filter. Note that this is a duplicate if the filter is the last in the chain. Useful for debug purposes and it will publish even if active is false.
• 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. Please use integers according to the linked enum
• method_type (int, default: 0) Segmentation model to use for. Default to SAC_RANSAC . Check pcl official doc. Please use integers according to the linked enum
• 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)

Changelog for package point_cloud2_filters

1.0.3 (2024-08-06)

• Updated README.md
• option to pub cloud immediately after the filter execution
• Contributors: Davide Torielli, torydebra

1.0.2 (2023-08-04)

• install example
• Contributors: Davide Torielli

1.0.1 (2023-07-24)

• Create LICENSE
• Contributors: torydebra

1.0.0 (2023-07-24)

• Merge pull request #5 from ADVRHumanoids/devel merging into master
• Sac Segmentation filter with additional base class
• added dynamic reconfigure && derivation from FilterIndices
• Merge branch 'devel' of https://github.com/torydebra/pass_through_filter into devel
• better readme
• wrong interpretation of ROS_INFO_NAMED corrected
• other filters and reamde better
• derived class
• cropbox filter
• keep organized option
• solved bug
• corrected filename
• fixing #2
• Merge pull request #4 from peci1/patch-1 Improve CMakeLists.txt
• Modified License
• Improve CMakeLists.txt
• first version
• first commit
• Contributors: Davide Torielli, Martin Pecka