using IM in telescope tracking system

[anthony]

Very difficult problem. As they are close to single pixel points then it should be possible by generating a gradient radiating out from each pixel location.

Now if you mask the second image from the pixel gradients, you should be able to get a average value of how far the second 'points' are from the first points.

If you do this with a X gradient, then a Y gradient you should be able to get a rough measure of the distance the points moved, which can be calibrated to actual pixel distance (or directly to motor steps).

How to generate this is another problem, which may involve masking (to remove color intensity differences), and some type of convolution blur (to generate the gradient). Fred Weinhaus is probably a good one to help with that.