Different noises use different filters, so it is difficult to say that a certain noise reduction filter can represent all noises.

First, let's talk about the types of noise. The classification of noise is related to what model the noise distribution conforms to. Common noises include Gaussian white noise, salt and pepper noise, Poisson distribution noise and exponential distribution noise.

Secondly, the filters used are spatial filters, such as mean filter, median filter, low-pass filter, Gaussian filter and so on. Frequency domain filters, such as wavelet transform, Fourier transform, cosine transform, etc. Morphological filtering mainly denoises through morphological operations such as expansion and corrosion.

Third, corresponding to the occasion. Generally, we usually see Gaussian white noise, such as mean filtering, median filtering and Gaussian filtering. Also, in low illumination, such as images taken at night, the noise generally belongs to Poisson distribution, and some 3d denoising algorithms can be used, such as BM3D algorithm with good effect. Like salt and pepper noise, median filtering can basically denoise.

That's about it. If you want to know more, you can find some books on image or some comprehensive papers on denoising.