I'd think it's pretty much any high-energy ionizing radiation that causes those streaks - probably very few of which are muons. There are "local" sources of ionizing radiation pretty much everywhere.
And if [0] is correct about the approximate muon flux - being that "about one per second passes through a volume the size of a person’s head.", the volume of a the CCD sensor that it would have to interact with is so much smaller (being some 10 microns thick) that I doubt it'll be "Take a few 30s exposures" sort of chance, so much as "Winning the lottery" level chance to actually have a muon pass through the sensor, and interact.
It doesn't have to hit the sensor, it needs to pass through it, so the thinness of it doesn't matter as much as the orientation; it's a matter of flux density.
You would expect ap to 2-3 muons per minute to pass through a typical sensor but you might not capture all of them.
It has to interact, otherwise it wouldn't be visible at all, as it's the interactions that the CCD detects.
And as muons don't interact often, the can pass through a lot of matter without anything noticing - that's the reason why they can pass through the atmosphere to still be detected on the ground - or even deep underground, as many imaging detectors are used, to avoid other radiation sources that could cause noise while still penetrating the rock you want to image. Compared to hundreds of metres of rock in a deep mine, a fridge and roof is nothing.
The muon’s charge excites electrons as it passes through an atom’s electromagnetic field. The camera is detecting this trail of ionisation as the muon passes between the atoms of the sensor (and these sensors are very good at detecting excited electrons). The muon does NOT need to decay or to strike the atoms of the sensor directly in order to be detected.
In open air at sea level, you would expect 1 muon to pass through any square cm of ground every minute, on average. With a sensor measuring 2-3 sq cm, oriented correctly, and exposing for a long enough time you would certainly expect to catch a few.
Unlike x-rays or gamma radiation, muons can pass through several km of dense matter and penetrate deep inside the earth before they decay. They can pass through solid lead. Direct particle collisions are rare but more likely when passing through large amounts of dense matter. The ionisation process can also reduce the speed and trajectory. Muon tomography works by comparing how much the muon count has been reduced compared to an expected background level.
I guess that by far the most likely local source would be potassium-40 which is a gamma emitter and relatively abundant in organic stuff. Due to the low penetration of alpha and beta radiation, the source would have to be inside the camera to even have a chance of hitting the sensor, limiting the rate of such events.
[0] seems to imply the vast majority of radiation the average person experiences is gases, like Radon and Thoron decay (itself just an isotope of Radon), which would likely be as prevalent inside the fridge (and so inside the camera frame itself) as anywhere else.
That's because there's gaseous radon-222 continuously released from soil and rock, as a decay product of natural uranium. But it's an alpha emitter so is stopped by anything, including a few cm of air, or a sheet of paper – or the glass filter in front of a digital camera sensor.
And if [0] is correct about the approximate muon flux - being that "about one per second passes through a volume the size of a person’s head.", the volume of a the CCD sensor that it would have to interact with is so much smaller (being some 10 microns thick) that I doubt it'll be "Take a few 30s exposures" sort of chance, so much as "Winning the lottery" level chance to actually have a muon pass through the sensor, and interact.
[0] https://home.cern/science/physics/cosmic-rays-particles-oute...