Imagery & Cameras

Image Formats

Cameras

Calibration in ROS (Monocular)

How to Calibrate a Monocular Camera

Useful Applications

guvcview

guvcview is a standardized camera interfacing application and is a great first step in making sure a uvc camera is functioning correctly.

ROS Interfaces

OpenCV

OpenCV is an all-in-one general purpose image processing library.

sudo apt install ros-kinetic-vision-opencv

Image Transport

Image Transport allows for smart and convenient handling of image streams and compressed images.

sudo apt install ros-kinetic-image-transport ros-kinetic-image-transport-plugins

Image Visualization

The RQT Image View Plugin allows for image streams to be viewed in RQT.

sudo apt install ros-kinetic-rqt-image-view

Camera Interfaces

TODO: Put this in a comparison table

• uvc_camera - Unsupported
• libuvc_camera - Unsupported
• usb_cam - Supports the widest range of cameras
• cv_camera
• video_stream_opencv - This is the best general purpose one to date
• raspicam_node - This is the only option for interfacing with the Raspberry Pi Camera

Exporting ROS Image Stream to Video

Setup

To export images to a video file, there are a few things that you will need to do first:

1. Change directory to the same location as your .bag files
2. Download the topic2video.

wget https://gist.githubusercontent.com/pryre/8d6f44e18d52efb616f64403de0838ef/raw/ed6b8851e5ebb4e6e7bfa7289f53a61733733211/topic2video.py

3. Figure out the framerate your video was running at (use rostopic hz /webcam/image_raw, but for your camera/image topic).

Note that you can see more information on the tool by running the command:

python topic2video.py --help

Exporting a Live Video

Run the following command and press 'CTRL+C' when you want to end the recording:

python topic2video.py /webcam/image_raw -r 30 

Where 30 is the framerate of your camera (used to encode the exported video at the correct rate).

Exporting from rosbag

Run the following command and wait for it to finish:

python topic2video.py /webcam/image_raw -r 30 -b recording.bag

Where 30 is the framerate of your camera (used to encode the exported video at the correct rate) and recording.bag is the filename of the bag file you want to use.

Exporting rosbag Images

Setup

To export images to a video file, there are a few things that you will need to do first:

1. Create a working directory folder:

mkdir /tmp/video_export

2. Download the export script to the working folder. Note that this will export .png compressed images and dump them into your the directory:

cd /tmp/video_export
wget https://raw.githubusercontent.com/qutas/info/master/Scripts/image_saver.py

3. Figure out the framerate your video was running at (use rostopic hz /webcam/image_raw/compressed, but for your camera/image topic)

Image Extraction

To do the actual export process, open three different terminals and run the following commands:

1. Terminal 1:

roscore

2. Terminal 2:

cd /tmp/video_export
python image_saver.py /camera_name/image_topic
# OR: python image_saver.py /camera_name/image_topic/compressed

3. Terminal 3 (replace BAG_FILE_NAME as with your bag file recording):

rosbag play BAG_FILE_NAME

Once the bag file recording is complete, the image files should now be saved in /tmp/video_export

Recombine the images into a video (optional)

The final step is to recombine all the images into a video file. To do this, we can use the tool ffmpeg.

The only trick here is that our image filenames are saved using their timestamps, so they may will need to be renamed such that ffmpeg can easily accept them:

cd /tmp/video_export
ls -v *.png | gawk 'BEGIN{ a=1 }{ printf "mv %s %06d.png\n", $0, a++ }' | bash

Finally, we can combine our images into a video (make sure to enter the framerate of your video in FRAMERATE):

ffmpeg -r FRAMERATE -i %06d.png -c:v libx264 -vf fps=30 -pix_fmt yuv420p -crf 18 out.mp4

Your exported video should now be compiled!