Checkerboard

In this tutorial we will show how n-fold-edge CLI can be used to detect corners of a checkerboard in an image.

First a image of checkerboard is needed. In this tutorial we will use the checkerboard from here. Page 8 which we print out at hold in front of a camera.

If n-fold-edge is not already installed, see installation.

With n-fold-edge installed the CLI can be run with:

python -m checkerboard

or if the installation have added n-fold-edge to the path it can be invoked simply with:

checkerboard

From here on out we will assume n-fold-edge is in the path, but if that is not the case for you replace checkerboard in the following with python -m checkerboard.

Then we can run the following:

checkerboard /path/to/image.png --kernel-size 21 --output tutorial-checkerboard.csv

Here we also supplied the a kernel-size 101 to use. A Larger kernel-size can give better results depending on the size of the checkerboard but will be slower. we also tell the program –output to save all found corners in a csv file. Another option is to use the –output-image to save a image with the found corners marked with red circles.

Below is a sample of the csv file with the checkerboard corner locations. Image x and image y are the image coordinates of a corner and object x and object y are the corners relative position to each other.

Image file	image x	image y	object x	object y
…
path/to/image.png	1017.99695	310.43902	0	1
path/to/image.png	1019.22826	742.89751	0	-1
path/to/image.png	1020.22152	966.98432	0	-2
path/to/image.png	1228.76734	525.54046	1	0
path/to/image.png	1225.81256	314.68380	1	1
path/to/image.png	1232.54707	740.86310	1	-1
path/to/image.png	1237.20839	963.41986	1	-2
path/to/image.png	1455.85207	742.89855	2	-1
path/to/image.png	801.36247	522.70910	-1	0
path/to/image.png	802.78263	299.01144	-1	1
path/to/image.png	799.59552	749.18699	-1	-1
path/to/image.png	797.43134	981.06289	-1	-2
path/to/image.png	564.19622	519.57029	-2	0
…