Ralf66 said:
I am curious as to roughly how many photons strike a camera sensor (e.g. a DSLR) in dim light, where the camera can form an image?
Offhand I'd say a very grainy image needs a SNR of around 5 at bare minimum. Since there are multiple sources of noise, it's hard to give you an exact number of photons needed for this. However, to simplify the situation, let's assume there's only shot noise, which is the randomness of the arrival of the photons themselves.
A signal to noise ratio is expressed as SNR = N / √N, where N is the number of photons detected by each pixel. This means that each pixel has its own SNR.
To get an SNR of at least 5, N needs to be at least 25. In other words, we need at least 25 photons if we want our pixel to have an SNR of at least 5. Let's say we have a 10 megapixel sensor. So, 25 photons per pixel times 10 million pixels is 250 million photons total for the sensor.
In reality we would need even more since shot noise is only one source of noise.
Also how many photons are hitting a photosite for it to be triggered?
CCD and CMOS photosites aren't 'triggered', but instead detect photons by accumulating electrons excited when photons strike the photosite. Each photon excites one electron, so they can potentially detect individual photons. However the real life situation is not that simple. In addition to shot noise, the electrons in each photosite have thermal energy and this causes 'thermal noise' where an electron is excited high enough by the thermal energy to be 'detected'. Plus the electronics themselves can cause noise too. So when the sensor is read after an exposure is taken, you have combination of electrons excited by photons mixed in with electron excited by other means. In bright light the number of photons hitting the sensor per unit of time is so high that the signal overwhelmingly dominates over the noise. But in low light conditions you have to take longer exposures and/or take other steps to get a good SNR and thus a good image.