DepthImageCreator

../../_images/depth_image_creator.jpg

What is this

Create organized pointcloud from non-organized pointcloud. Currently it supports pcl::PointXYZ and pcl::PointXYZRGB as the input.

Subscribing Topics

  • ~input (sensor_msgs/PointCloud2):

    The input pointcloud to be reconstructed as organized pointcloud.

  • ~info (sensor_msgs/CameraInfo):

    Put a simulated camera according to ~info and generate organized pointcloud.

Publishing Topics

  • ~output (sensor_msgs/Image):

    Publish organized pointcloud as depth image.

  • ~output_image (sensor_msgs/Image):

    Publish image colorized according to the input cloud.

  • ~output_cloud (sensor_msgs/PointCloud2)

    Output pointcloud.

    See ~organize_cloud parameter.

  • ~output_disp (sensor_msgs/DisparityImage)

    Publish organized pointcloud as disparity image.

Parameters

  • ~scale_depth (Double, default: 1.0)

    scale depth value.

  • ~use_fixed_transform (Boolean, default: False):

  • ~translation (Array of double, default: [0, 0, 0])

  • ~rotation (Array of double, default: [0, 0, 0, 1])

    If ~use_fixed_transform is set to True, transformation between ~input and ~info is not resolved via tf but fixed transformation is used according to ~rotation and translation.

  • ~use_asynchronous (Boolean, default: False)

    If this parameter is set to True, this node process incoming PointCloud regardless of the synchronization between ~input and ~info.

  • ~use_approximate (Boolean, default: False)

    Synchronize ~input and ~info approximately if this parameter is set to True. If this parameter is set to False, and the timestamps of the ~input and ~info are not strictly matched, the callback will not be processed. (See http://wiki.ros.org/message_filters/ApproximateTime)

  • ~info_throttle (Integer, default: 0)

    The number of ~info messages to skip to generate depth image.

  • ~max_queue_size (integer, default: 100):

    Queue length for synchronization of topics. According to the message_filters synchronization policy, in case of ~input(1Hz) and ~info(30Hz) comes, message_filters needs at least over 30 queue_size for searching a synchronized set of the two. (You can also check like this rosrun jsk_topic_tools is_synchronized --approximate-sync --queue-size 3 /kinect_head/depth_registered/camera_info /kinect_head/depth_registered/points)

  • ~max_pub_queue_size (integer, default: ~max_queue_size_):

    Queue length of topic publishers. Default is value set for max_queue_size_.

  • ~max_sub_queue_size (integer, default: ~max_queue_size_):

    Queue length of topic subscribers. Default is value set for max_queue_size_.

  • ~fill_value (float, default: nan):

    Initial value of depth image. The pixels where there is no corresponding point are filled by this value.

  • ~organize_cloud (Boolean, default: False)

    Whether to organize ~output_cloud or not.

  • ~tf_duration (float, default: 0.001):

    TF Lookup transform duration. This value is only used when use_fixed_transform is false.

Sample

roslaunch jsk_pcl_ros sample_depth_image_creator.launch