sonar_image_proc

Code to draw data from forward-looking imaging sonars.

If built for ROS, it will build a node/nodelet draw_sonar which subscribes to an marine_acoustic_msgs/ProjectedSonarImage and publishes a sensor_msgs/Image.

The core library contains no ROS dependencies, and can be linked into non-ROS applications.

ROS Interfaces (draw_sonar_node)

rosrun sonar_image_proc draw_sonar_node

Subscribers

Subscribes to the topic sonar_image of type marine_acoustic_msgs/ProjectedSonarImage.

Publishers

By default publishes three sensor_msgs/Image topics:

• drawn_sonar contains an image of the sonar drawn with a cartesian projection with the origin (range = 0) of the sonar centered on the bottom edge of the image, and azimuth=0 extending vertically upwards in the image. By default, the image height is set by the number range bins, and the image width is automatically determined based on the height and the min/max azimuth of the sonar image. The color map used to convert the sonar intensity to RGB is set in code.

• drawn_sonar_rect is the contents of the SonarImage data mapped directly from polar to cartesian coordinates. Since the intensity data in the SonarImage is stored azimuth-major, the data is mapped into the image space with range in the X direction and azimuth in the Y direction:

  • Image width is the number of range bins in the data, with the minimum range on the left side and maximum range on the right side.

  • Image height is the number of azimuth bins in the data, with the lowest azimuth (typically the most negative) at the top, and most positive at the bottom.

• drawn_sonar_osd adds guidelines and annotations to drawn_sonar. Overlay parameters can be configured in realtime.

If the param publish_timing is true, the node will track the elapsed time to draw each sonar image and publish that information to the topic sonar_image_proc_timing as a std_msgs/String containing a JSON dict.

If the param publish_old is true, the node will also draw the sonar using the old draw_sonar algorithm for comparison.

Params

If max_range is set to a non-zero value, images will be clipped/dropped to that max range (or the actual sonar range, whichever is smaller).

If publish_timing is true the node will publish performance information as a JSON string to the topic sonar_image_proc_timing. Defaults to true

If publish_old is true the node will also draw the sonar using the old algorithm to the topic old_drawn_sonar. Defaults to false

If publish_histogram is true the node will publish a "raw" histogram information as a UInt32MultiArray to the topic histogram. It contains a vector of unsigned ints giving the count for each intensity value -- so for 8 bit data the vector will be 256 elements in length, and for 16-bit data it will be 65536 elements in length.

bag2sonar

The program bag2sonar reads in a bagfile containing a ProjectedSonarImage topic, draws the sonar image and writes those images to new bagfile in a Image topic.

Usage:

$ rosrun sonar_image_proc bag2sonar
Usage:

   bag2sonar [options]  <input file(s)>

Draw sonar from a bagfile:
  -h [ --help ]                         Display this help message
  -l [ --logscale ]                     Do logscale
  --min-db arg (=0)                     Min db
  --max-db arg (=0)                     Max db
  --osd                                 If set, include the on-screen display
                                        in output
  -o [ --output-bag ] arg               Name of output bagfile
  -t [ --output-topic ] arg (=/drawn_sonar)
                                        Topic for images in output bagfile

Note that bag2sonar is not a conventional ROS node, it is intended to run as a standalone commandline program. It uses ros_storage to read the input bagfile sequentially, rather than subscribing to a topic.

histogram_drawer

python/histogram_drawer is a Python script which subscribes to the histogram topic and uses numpy+Matplotlib to bin the data (into a fixed set of 128 bin right now), and draw a plot to the topic drawn_histogram.

Python API

Long term, I'd like to be able to call this drawing function from Python, however we're not there yet.

There IS a totally separate python node that publishes a pointcloud for visualization in rviz:

rosrun sonar_image_proc sonar_pointcloud.py

API

Sonar drawing is implemented in the SonarDrawer class, which takes an instance of an AbstractSonarInterface and returns a cv::Mat. SonarDrawer computes and stores pre-calculated matrices to accelerate the drawing.

A convenience function drawSonar is also provided. It is a trivial wrapper which creates an instance of SonarDrawer then calls it. Calls to drawSonar do not retain the cached matrices and are less efficient.

Related Packages

• liboculus provides network IO and data parsing for the Oculus sonar (non-ROS).
• oculus_sonar_driver provides a ROS node for interfacing with the Oculus sonar.
• marine_acoustic_msgs defines the ROS ProjectedSonarImage message type published by oculus_sonar_driver.
• rqt_sonar_image_view is an Rqt plugin for displaying sonar imagery (uses sonar_image_proc)

License

Licensed under BSD 3-clause license.