Generally speaking, I don't think that there is one best filter. Filtering is always a tradeoff between enhancing the image's quality and losing fine information. To be able to select a suitable filter technique, the following aspects have to be considered:

• Degree and distribution of noise: There are many different reasons why an image needs filtering. The most typical is noise caused by the acquisition of the image (in terms of photography or satellite images) which causes a wiedely even distribution of the same amout of noise throughout the image. Of course, this noise needs a different filter than images which show some small distributed artefacts (e.g. due to signal processing or previous image analysis) and these need other filters which are able to preserve the non-affected areas of the images as much as possible.
• Step within the processing chain: It is decesive if a filter is applied at the beginning of a longer workflow or at the end. In some cases, the image has to be smoothed before any other step can be applied whereas in other cases preprocessing itself causes the neef for filtering. So again, giving general recommendations on filters is impossible without knowing anything about the context of the image's analysis.
• Type of image: Most importantly, different images have specialized filters which were designed for limited applications. Whereas photography often suffers from salt-and-pepper-noise or sunglint, other types have a more regular noise, such as periodic patterns.

If you search for comparisons of filter techniques, I would advise to search for review papers or book chapters within one special domain (photography, video, satellite images....).

See the page 'Image Filtering', https://www.mathworks.com/help/images/linear-filtering.html

Generally speaking, I don't think that there is one best filter. Filtering is always a tradeoff between enhancing the image's quality and losing fine information. To be able to select a suitable filter technique, the following aspects have to be considered:

• Degree and distribution of noise: There are many different reasons why an image needs filtering. The most typical is noise caused by the acquisition of the image (in terms of photography or satellite images) which causes a wiedely even distribution of the same amout of noise throughout the image. Of course, this noise needs a different filter than images which show some small distributed artefacts (e.g. due to signal processing or previous image analysis) and these need other filters which are able to preserve the non-affected areas of the images as much as possible.
• Step within the processing chain: It is decesive if a filter is applied at the beginning of a longer workflow or at the end. In some cases, the image has to be smoothed before any other step can be applied whereas in other cases preprocessing itself causes the neef for filtering. So again, giving general recommendations on filters is impossible without knowing anything about the context of the image's analysis.
• Type of image: Most importantly, different images have specialized filters which were designed for limited applications. Whereas photography often suffers from salt-and-pepper-noise or sunglint, other types have a more regular noise, such as periodic patterns.

If you search for comparisons of filter techniques, I would advise to search for review papers or book chapters within one special domain (photography, video, satellite images....).

These publications may help you

https://www.researchgate.net/publication/316009623_Color_Invariant_Representation_and_Applications

https://www.researchgate.net/publication/304188224_Block-based_illumination-invariant_representation_for_color_images