in reply to Stuart Richman

If I'm correct they use 1550 nm photons, so Infrared photons that would typically be emitted naturally by a blackbody at 1000 Kelvins or so (about 800°C but that's not a precise number anyway)
So unless you have an active volcano in the line of sight the background should be fairly quiet at those wavelength (well obviously you should not try to image the sun directly 😛).
in reply to Stuart Richman

aw wait, I misread your question : we are now able to count photons one by one, for example with photomulitpliers. A single photon enters the detectors, rip an electron off a material, that electron is accelerated in an electric field and will generate an avalanche of other electrons. In the end you have a huge gain, like for one photon/one electron at input, you get more than enough electrons at output to detect electronically.
Practically, you see a pulse on the screen each time a photon is detected, this is the current pulse generated in the detector. At visible wavelength in daylight, you'll have so much photons you will not see anything unless you're very very very fast.
But here thanks to thin frequency band and I guess other tricks, they manage to get a background count of some hundreds photons per seconds, which is very low and detectable individually by a fairly slow electronics.