Why images are still recognizable (or not) after they go through
Colorer: The brain relies heavily on edge detection to do object identification. In edge detection, color doesn't matter; changes in luminance do. (Hence, colorblind people can identify objects.) The Colorer results preserve a lot of those edges. Some of them even make those edges easier to detect by increasing the luminance on the opposing sides of the edges.
(In nature, most lightness differences between edges are subtle. e.g. A yellow next to a slightly darker yellow gives the visual cortex a clue that one facet of a banana is ending and another is starting. When Colorer picks two colors at random, there's a good chance that the contrast is going to be higher between them than Yellow and Slightly Darker Yellow.)
There's also some Colorer images that make it harder to tell what is being depicted in the image by either cutting it down to too few colors or by making it easier to see edges that you may not have noticed in the original. However, those edges may suggest to you new things that were not in the original at all.
This picture was captioned "McNalley and Pines Fires in California: Natural Hazards". I at first thought that it was pine trees in a front of a mountain with smoke rising from them. In actuality, the
original (second image on that page) is a map of the Pacific Ocean and California.
BTW, I learned about human visual processing by reading
Vision and Art: The Biology of Seeing by Margaret Livingstone. It's incredibly good and very layperson-accessible.