FeatureRegistration

../../_images/feature_registration.png

Align pointcloud using 3d feature. Currently only FPFH is supported.

Subscribing Topic

  • ~input (sensor_msgs/PointCloud2)

    Input pointcloud. The type of point is pcl::PointNormal.

  • ~input/feature (sensor_msgs/PointCloud2)

    Input feature. The type of point is pcl::FPFHSignature33.

  • ~input/reference/cloud (sensor_msgs/PointCloud2)

    Reference pointcloud. The type of point is pcl::PointNormal.

  • ~input/reference/feature (sensor_msgs/PointCloud2)

    Reference feature. The type of point is pcl::FPFHSignature33.

Publishing Topic

  • ~output (geometry_msgs/PoseStamped)

    Transformation to align reference cloud to input cloud.

  • ~output/cloud (sensor_msgs/PointCloud2)

    Reference pointCloud which is aligned to input cloud.

Parameters

  • ~max_iterations (Integer, default: 1000)

    Maximum number of iterations.

  • ~correspondence_randomness (Integer, default: 2)

    Number of neighbors to use when selecting a random feature correspondence.

    A higher value will add more randomness to the feature matching.

  • ~similarity_threshold (Double, default: 0.9)

    Similarity threshold in [0,1] between edge lengths of the underlying polygonal correspondence rejector object, where 1 is a perfect match.

  • ~max_correspondence_distance (Double, default: 0.0075)

    Maximum distance threshold between two correspondent points in source <-> target.

  • ~inlier_fraction (Double, default: 0.25)

    Required inlier fraction of the input in [0, 1]

  • ~transformation_epsilon (Double, default: 0.1)

    Maximum allowable difference between two consecutive transformations in order for an optimization to be considered as having converged to the final solution.

These parameters can be changed by dynamic_reconfigure.

Sample

roslaunch jsk_pcl_ros sample_feature_registration.launch