ArUco

Exploring potential applications of ArUco fiducial markers.

Installation

ArUco Installation

To get access to the ArUco fiducial markers in python/opencv:

pip install opencv-contrib-python

OGRE Installation

OGRE is a 3D library that can be used in conjunction with ArUco to create augmented reality applications. There is an OVIS module in OpenCV that uses OGRE, but it is not in the opencv-contrib-python installation by default. So we will need to build from source.

The docker folder has Dockerfile examples for building OpenCV with OGRE support.

I have successfully run the ovis demo by running

docker build -t cv_ovis -f docker/Dockerfile .
xhost +
docker run --rm -ti --net=host --env="DISPLAY" --volume="$HOME/.Xauthority:/root/.Xauthority:rw" -v $(pwd):/data cv_ovis /bin/bash
python3 ovis_demo.py

It is slow though, since it isn't running on GPU.

Marker Preparation

Generate marker images with generate_markers.py or use the MarkerPrinter which supports svg/pdf, but doesn't support as many dictionaries.

Scripts

Some examples steps in order.

Create the charuco board

python3 src/generate_charuco_board.py --width 16 --height 9 --output output/charuco_board_16x9_dict_4x4_100.png --dictionary DICT_4X4_100

Print the result for a later step.

Create the ArUco markers

python3 src/generate_markers.py --output output/marker_ --dictionary DICT_4X4_50

Print the results for later steps.

Calibrate with the charuco board

python3 src/charuco_calibrate.py --width 9 --height 16 --dictionary DICT_4X4_100 --feed data/charuco_calibration.mp4 --output output/charuco_calibration.npz

Use your own video with the printed dictionary if calibrating your own camera.

Detect the markers

python3 src/aruco_example.py --dictionary DICT_4X4_50 --input data/4x4_50_markers.png --camera-matrix data/charuco_calibration.npz --output output/4x4_50_markers.png

Show Sinbad in a still image without detecting markers

python3 src/aruco_ar_no_markers.py --input data/4x4_50_markers.png --camera-matrix data/charuco_calibration.npz

Show Sinbad and detect markers

python3 src/aruco_ar_demo.py --input data/4x4_50_markers.png --marker-length=1 --camera-matrix data/charuco_calibration.npz --dictionary DICT_4X4_50

Show Sinbad, detect markers, and save video

python3 src/aruco_ar_video_demo.py --input data/4x4_50_markers.png --marker-length=1 --camera-matrix data/charuco_calibration.npz --dictionary DICT_4X4_50 --output output/ar.mp4

shortcipher3/aruco_opencv