PrimitiveShapeClassifier

https://user-images.githubusercontent.com/1901008/27854434-ca088eee-61a1-11e7-9a98-0864013a8f63.pngprimitive_shape_classifier

Classify shape for each cluster point indices on planes.

Subscribing Topics

  • ~input (sensor_msgs/PointCloud2)

    Input XYZRGB point cloud

  • ~input/normal (sensor_msgs/PointCloud2)

    Input normal point cloud

  • ~input/indices (jsk_recognition_msgs/ClusterPointIndices)

    Input cluster point indices

  • ~input/polygons (jsk_recognition_msgs/PolygonArray)

    Input supporting planes

Publishing Topics

  • ~output (jsk_recognition_msgs/ClassificationResult)

    Classification result

  • debug/boundary_indices (jsk_recognition_msgs/ClusterPointIndices)

    Cluster point indices of boundary points for each clustered objects

  • debug/projected_cloud (sensor_msgs/PointCloud2)

    Projected boundary points on supporting planes

Parameters

  • ~queue_size (Int, default: 100)

    Queue size for message synchronization

  • ~min_points_num (Int, default: 10)

    Minimum number of points for each cluster.

  • ~sac_max_iterations (Int, default: 500)

    Maximum iteration number for SAC segmentation

  • ~sac_distance_threshold (Double, default: 0.005)

    Distance threshold for SAC Segmentation

  • ~sac_radius_limit_min (Double, default: 0.025)

    Minimum radius for circle estimation

  • ~sac_radius_limit_max (Double, default: 0.13)

    Maximum radius for circle estimation

  • ~box_threshold (Double, default: 0.70)

    Threshold to classify objects as boxes

  • ~circle_threshold (Double, default: 0.30)

    Threshold to classify objects as circles