(Moving this thread to the proper CP board from CP 1.0)
One possible solution is to improve the contrast in your image so that it better resembles the mask. The only reason that alignment can work at all is that there needs to be a good amount of correlation between the regions inside vs. outside the plate. I suggest two things:
Improving the illumination of the image:
The background is lighter on the right than on the left. This inhomogeneity is probably causing some problems in alignment, or at least, making it harder than it needs to be. Try to achieve a uniform illumination across the entire image. - Improve color contrast:
This person made a good move by taking a color image that had really good contrast between the fungus and the background. This might also help in your case by increasing the contrast between the plate and the background. Perhaps you can use a darker background?
Lastly, even if you achieve alignment with a reasonable time per image, I’m hard-pressed to see how you would identify the colonies using a grayscale image alone. You should probably take color images of your plates to begin with.
Hope this helps!