Search for a chessboard in an image

Question

So, I initially assumed that finding a chessboard in an image should be trivial because it is such an easily defined object. However, this is not so simple, and I was wondering if anyone knows how the cvFindChessboardCorners checkerboard seeker works in OpenCV. Ive tried a google search but havnt managed to find an algorithm. I suggest perhaps the following:

1) Binarize
2) Open and close to eliminate small clusters

A)
2) Find the angles of Harris
3) Create a matrix of distances between all points in the image
4)...?

IN)
2) Find the hough transform
3) all significant lines where they intersect are checked. If 4 or more lines intersect at a point, then these lines are part of a checkerboard. This includes infinity.
4)?

Does anyone know for sure?

Answer 1

This is quite ... complicated :) If you want to know exactly what the opencv source will be the place to search - in opencv 2.2, in the line modules/calib3d/src/calibinit.cpp 219. It also has a compiled DEBUG_CHESSBOARD switch to be able to see how it is works.