Illumination theory for image inspection

Maybe not all...
If the purpose of imaging inspection is to "replace visual inspection," then I believe that ultimately what is most important is "correlation with visual inspection. Going back to that point of view, the general idea of current imaging illuminators is "?" I have been thinking more and more about it. A few.
1) "Specific" defects
In many cases, I believe that "a certain irradiation angle is good" or "a certain color (wavelength) is good" is selected to target specific defects. However, what is actually required in inspections is not an illuminator that can detect specific defects, but an illuminator that can detect all defects. Are there any defects that are obscured by forcibly increasing the contrast of a particular defect?
Why not switch illuminators? But it is obvious that it would be a game of tussle, considering the infinite number of defect types.
2) Who is the "authority" on illuminators?
Who knows best how to make a good lighting system? The answer is simple: the people in charge of visual inspections on site. What is easy to see with the eye is also considered easy to see with a camera.
I don't think the answer can be found by asking lighting manufacturers to consider lighting for specific defects.
(3) Are they really "LEDs"?
Most visual inspections are conducted under fluorescent lighting. Is it possible to correlate an inspection conducted in such an environment with an inspection using a single wavelength, low-luminance LED? As you can see when you try it, there are many items that are clearly visible under fluorescent light, but become difficult to see as soon as LED is used. I think the idea that "it has to be LED" sometimes makes image inspections difficult.
(4) What we want to see in illuminators
The same object should look the same when viewed in the same way. If the image of the same object is different each time it is imaged, I don't think it is possible to inspect the object.
Even if it is only a little, defects should appear in the image with a higher contrast than noise.
The only way to achieve this is to have the defects appear in the image with a higher contrast than the noise.
We are still in the process of searching for the right lighting and have not reached a conclusion.
However, we have abandoned the idea that "LEDs are the only way to go.