Edge detection

If an edge is going to be really there and not some accident of brightness, then it ought to have some extension in space and also in colour, that is to say, if we think we have a sharp change in colour at some point, then neighbouring pixels ought to show similar changes. This means that it suffices to ensure that a putative edge in the red image ought to have spatial extension, likewise the green and blue images. So edge detection is slightly easier, because there are often simultaneous changes in more than one of the greyscale images at the same pixel. It would not be surprising if there were changes in the different greyscale images at nearby pixels, and smarter methods can look for this effect. Again, this sort of issue belongs in Image Processing courses, but it is worth doing a little meditation on what is possible before reading the books or attending the courses.