If the subject is not moving then stacking several images can average out the randomness of photon arrival at the sensor and stacking can sum the brightness of the individual images. Image stacking results are similar to long exposure times.
Or, a black frame might be subtracted to eliminate some of the noise. Some cameras are able to do this in camera.
If the subject is moving and a fast shutter speed is needed, a median filter will trade resolution for noise reduction in a high ISO image much like a larger photo site would. In camera high ISO noise reduction will do something similar to the median filter. Then the image may be downsized to achieve the smaller number of pixels that one would have gotten with larger photo sites. So one may do the pixel binning after the image is captured. After the downsize, sharpening will be needed because the edges will have been blurred by the median filter.
https://www.cambridgeincolour.com/tutorials/image-noise.htm
https://www.cambridgeincolour.com/tutorials/image-averaging-noise.htm
https://www.cambridgeincolour.com/tutorials/image-noise-2.htm
https://en.wikipedia.org/wiki/Median_filter
https://homepages.inf.ed.ac.uk/rbf/HIPR2/median.htm
https://core.ac.uk/download/pdf/82222933.pdf 8 page PDF
Also some like AI noise reduction which guesses what the value of each pixel should have been.
https://www.nature.com/articles/s41598-024-66649-y
Guan, S., Liu, B., Chen, S. et al. Adaptive median filter salt and pepper noise suppression approach for common path coherent dispersion spectrometer. Sci Rep 14, 17445 (2024). https://doi.org/10.1038/s41598-024-66649-y
