The HI DeckLink ROS package exposes BlackMagic Design DeckLink video playback and capture cards to a ROS network. It is based on the previous version DeckLink ROS developed at the NearLab (Politecnico di Milano).
HI DeckLink ROS leverages libdecklink, a higher-level level interface to the BlackMagic Design SDK, to control the underlying card(s). This component is installed as a git submodule.
This project has been tested with:
- Ubuntu 16.04 and Ubuntu 20.04
- ROS Kinetic, ROS Lunar, and ROS Noetic
- DeckLink Duo and DeckLink Quad 2
Clone the repository into your ROS workspace:
git clone --recursive https://github.com/MPI-IS/hi_decklink_ros.git
Build the nodes:
catkin build hi_decklink_ros
HI DeckLink ROS can also be built inside a Docker container using the provided Dockerfile.
In order to do so, simply do
docker build . -t hi_declink_ros # builds image
docker run -it -d --name my_container hi_declink_ros # starts containerNote that building this image is part of our continuous integration.
The publisher node reads images from one input of the DeckLink card and publishes them to ROS topic.
After having the roscore running, open in a different terminal:
rosrun hi_decklink_ros publisher _decklink_device:="DeckLink [model] ([input])"
This will create a publisher node that listens for images on from one input of your DeckLink card
and publishes them on the topic /image_raw.
To see the published image:
rosrun image_view image_view image:="image_raw"
The node will additionally publish sensor_msgs::CameraInfo messages synchronised to each image message.
If the camera is uncalibrated these will be empty.
If a camera calibration file is available you can pass the path to the file in the camera_info_url parameter.
This will allow the image_proc family of nodes to automatically generate rectified images.
The node accepts the following parameters:
| Parameter | Description |
|---|---|
decklink_device |
The name of the capture (input) interface on the DeckLink card from which to read images and the channel used. |
camera_name |
A name to identify the camera. The name is used to check that the right camera calibration information is being used. If the camera name in the calibration file and the name passed here differ a warning will be shown in the ROS console. If the name is not set the DeckLink device name is used instead. |
camera_frame |
The tf frame that the camera should attached to. This helps to keep point clouds generated with stereo_image_proc in the correct reference frame. The default value is the camera name. |
camera_info_url |
The location in which to locate the camera info file. This should a be an absolute file path. |
A launch file for a stereo endoscope is provided for documentation purposes in the launch/ folder.
The publisher node has been tested on:
- a clinical Intuitive da Vinci Si HD robotic system
- a dVRK (da Vinci Research Kit)
- a standard HD-SDI video camera
The subscriber node reads images from a ROS topic and writes them to the specified DeckLink output. This node can be used to perform keying.
After having the roscore running, open in a different terminal:
rosrun hi_decklink_ros subscriber _decklink_device:="DeckLink [model] ([input])" _topic:="[ros/image/topic]"
This will create a subscriber node that monitors the ROS image topic given as input.
This node expects BGRA8 formatted images for simplicity and will produce an error
if an image with a different formatting is used. The alpha channel is required to support keying.
To quickly test this node, please check the section "Test subscriber".
The node accepts the following parameters:
| Parameter | Description |
|---|---|
decklink_device |
The name of the playback (output) interface on the DeckLink card onto which the images will be written and the channel used. |
topic |
The ROS topic from which the images will be read. |
image_format |
The name of the display mode to use. |
keying (bool) |
Whether or not to enable keying on the card. |
opacity (int) |
The opacity of the keyed images in the range 0 (transparent) to 255 (opaque). |
The subscriber node has been tested on:
- the stereo viewer of a clinical Intuitive da Vinci Si HD robotic system
- the stereo viewer of a dVRK (da Vinci Research Kit)
Keying on DeckLink devices is extremely fast (less than 1ms extra latency on average). On modern cards it is possible to re-map each individual connector so that they can be used individually for input or for output. However, to use keying you must provide an input onto which the images will be keyed; as such, you must retain a pair (the left connector is for input and the right connector is for output). The output will be the input video with the keyed image overlaid on top with the specified opacity.
The pixel format is hard coded to YUV422.
We prepared a small demo to test the subscriber node both in writing and keying mode.
We placed an image (image.png in the folder sample) that you are welcome to use.
After having the roscore running:
rosrun hi_decklink_ros img2ros _path:="[/path/to/your/image.png]"
This will create a ROS topic image (image_ros). You can see it with:
rosrun image_view image_view image:="image_ros"
Then run the subscriber node, specifying this topic:
rosrun hi_decklink_ros subscriber _decklink_device:="DeckLink [model] ([input])" _topic:="image_ros"
By default, the node will write this image.
If you want this image to be keyed on the input video:
rosrun hi_decklink_ros subscriber _decklink_device:="DeckLink [model] ([input])" _topic:="image_ros" _keying:="True" _opacity:="150"
We designed the subscriber node, in a way that it possible to use the same node and change the two modes (write/keying) internally,
using a boolean topic function/output_write. To understand its functioning, we suggest to check it in subscriber.cpp.
We used these drivers to connect a workstation computer to the vision system of a clinical da Vinci Si HD surgical robot (Intuitive Inc.). In this way, we could overlay virtual content of the intraoperative images acquired by the endoscope.
Examples of launch files are in the folder launch.
For more information about how to connect the hardware to a da Vinci robot,
please refer to our paper.
Maria-Paola Forte, Haptic Intelligence Department - Max Planck Institute for Intelligent Systems
HI DeckLink ROS is based on the work done by Thibaud Chupin on DeckLink ROS.
Maria-Paola Forte, Haptic Intelligence Department - Max Planck Institute for Intelligent Systems
MIT license (see LICENSE.md).
DeckLink ROS is distributed under the MIT license (see LICENSE_declink_ros.md)
© 2020, Max Planck Society - Max Planck Institute for Intelligent Systems
The libdecklink tools (check the status of the DeckLink card, its model, the status of the channels etc.)
are not maintained anymore since this information can be obtained
through the softwares provided by BlackMagic Design.
We thank Jean-Claude Passy and the Software Workshop for their help to release the code.
If you use these drivers, please cite:
@article{Forte22-IJMRCAS-Design,
title={Design of Interactive {AR} Functions for Robotic Surgery and Evaluation in Dry-Lab Lymphadenectomy},
author={Forte, Maria-Paola and Gourishetti, Ravali and Javot, Bernard and Gomez, Ernest T. and Kuchenbecker, Katherine J.},
journal={The International Journal of Medical Robotics and Computer Assisted Surgery},
year={2022},
publisher={Wiley Online Library}
}