To avoid anyone misunderstanding my issue ill be more detailed.
This camera setup would involve taking a photograph with a webcam which is inside a box with a small hole cut at the top with a small piece of glass.
I need 2 images taken. One when the finger is just about to press the glass. And the second is when the finger actually presses the glass.
I need the both images to be as similar as possible when I take it outdoors,indoors, at night or in the morning.
Colour would also be a very important feature I need. Cause the finger is usually pink when its not pressed and white when its pressed. I need these 2 colours to be as obvious to the computer as possible. Any tips on how to improve the colors in an image would also be very much appreciated.
Not sure if I got this right, but I sense that rigging the box would be difficult, but the least of your problems, which would be:
1) Seriously low res and inconsistent results from the webcam,
2) Inconsistent luminence and colour (greatly due to the webcam factor) and also bcause light behaves differently, as you have identified, when going through a finger that is pressed and not pressed. Unless you adjust your lighting accordingly, and post-process it carefully, you are going to get variances in luminence, tone and colour.
3) the only way to achieve "both images to be as similar as possible when I take it outdoors, indoors, at night or in the morning." is to put in place a highly controlled artificial light set-up where that light is the primary and over-riding source, aka, even if you stand in the middle of an open field at 12 noon, the artificial light is much stronger than the sun.
Could I ask, exactly what is the intended output, aka, how do you need it used or presented in your project? Also, given your description of the box, I doubt if reflections would be the main concern depending on how and where you position it. I'm guessing that your project might be a bit ambitious, as usually these scanners are quite an example of precision engineering, and not cheap. Also, you would need a specialised soft-ware in place to 'illustrate' how the output of your scanner is used in finger-print recognition. I'm sure it's starting to get a bit complicated sounding becuase it is. I have a friend in the Netherlands who writes programmes specifically for electronic microscopic forensics and research/data gathering work, aka, ballistics matching, lining-up of wounds, comparative and contrast analysis etc, and he tells me that the equipment involved is fairly complex.
Have you considered Ortega's suggestion of a small name-card scanner?