You need to be a little systematic. Often when doing things like Pattern Recognition, it is easier to work with Binary Images, i.e. pixels either On or Off. There are various thresholding methods in IMAQ that will get you started in this. You may want to do one for the Outer Circle(s) that define your substrate -- once you get the circle, you know (more or less) the location of your disk. Now mask everything else away except the interior (which includes your red dots), and do another thresholding operation to get just the dots. Now all you need is the orientation -- I would guess that if you did a 2D Fourier Transform of the binary image, you could deduce the axes along which the dots are aligned (there are obviously several, but the most intense should be through rows and columns).
I've not done much "machine recognition", so I can't be more specific. However, I bet if you did searches on LabVIEW Object Recognition, you'd find useful hints and possibly even applicable code.
Bob Schor
