Rationale for functions going into Computer Vision vs. Image Processing toolboxes

I am curious about the thought process is for deciding if a builtin function belongs in the Computer Vision vs. Image Processing toolboxes. As you can guess, I only own one, the Image Processing toolbox, but there are several nice image processing functions (examples here; I have run into others too) which come with the Computer Vision toolbox. From reading the toolbox descriptions, should I assume that anything to do with features belongs in the Computer Vision toolbox?