Apply morphological filtering and average filter to fill small holes in pointcloud which is represented as heightmap.

Subscribing Topic

  • ~input (sensor_msgs/Image)

    Input heightmap. Hole should be represented as -FLT_MAX or nan.

    Encoding should be 32FC2.

  • ~input/config (jsk_recognition_msgs/HeightmapConfig)

    Config topic.

    This topic name is automatically resolved by ~input topic name.

Publishing Topic

  • ~output (sensor_msgs/Image)

    Output heightmap.

  • ~output/config (jsk_recognition_msgs/HeightmapConfig)

    Config topic.

    This node just relays ~input/config.


  • ~max_queue_size (Integer, default: 10):

    Max queue size of subscription callback.

  • ~mask_size (Integer, default: 2):

    Size of kernel operator of average filtering.

  • ~max_variance (Double, default: 0.1):

    Allowable max variance in kernel operator

  • ~smooth_method (String, default: average_variance)

    You can choose method of smoothing from average_variance and average_distance.

  • ~use_bilateral (Bool, default: false)

    use bilateral filtering after smooth(interpolation) method

  • ~bilateral_filter_size (Integer, default: 5)

    Kernel size of bilateral filtering.

  • ~bilateral_sigma_color (Double, default: 0.04)

    filter sigma of color space.

  • ~bilateral_sigma_space (Double, default: 5)

    filter sigma of coordinate space.


roslaunch jsk_pcl_ros sample_heightmap_morphological_filtering.launch