In the image processing field, granularity is usually used in the context of morphology, where structuring elements (kernels) of different sizes are employed to figure out the sizes of objects in an image.
I(1:150,1:150) is not enough code to do anything with, as I think you know, since you've said before that you are a university instructor who develops demos for his students. And you should know that I is a horrible name for an image variable - it looks too much like l and 1.
