Tim created another superb solution. The highlights are an elegant center of square's determination using ndgrid. The crux of his method is a convolution for each square not using "same" with an interesting centroid kernel. The comparison between all square convolutions for all rotations utilizes a concise norm metric function. The method should work on non-binary images. The best six matches from the 180x180 upper triangle error array are handily converted into the output format. Thank You Tim for this elegant solution.
Find common elements in matrix rows
1272 Solvers
269 Solvers
Longest run of consecutive numbers
1811 Solvers
231 Solvers
578 Solvers
Find the treasures in MATLAB Central and discover how the community can help you!
Start Hunting!