What settings should I use on a webcam to get accurate light intensity measurements?

I hope I am not in the wrong place to ask this kind of question since it is more about image acquisition than photography.

I have to use a Microsoft Lifecam HD webcam to take snapshots of a laser beam and measure its relative intensity. This means that I do not need to compare it to any standard measurement, I just want a behavior that is as close as possible to linearity, that is, if I take two pictures of two different beams and one is twice as powerful than the other, I should be able to measure this 2x factor in the images. It is important to note that the frequency of the incoming light is always (approximately) the same, therefore I should not need to worry about different sensitivities at different frequencies since there are no different frequencies. I have no interest in chromatic fidelity, I only care about intensity. I think I can measure image intensity by simply summing the RGB values from each pixel.

I wonder how to set the camera so that I can obtain a close to linear relation between the image intensity I can measure from the saved image and the actual incoming light intensity. I have access to the following properties:

• Brightness


• Contrast


• White Balance


• Saturation (remember that I do not care about color, the light I am observing is nominally monochromatic at 632 nm, which is close to red)

Of course these settings have to stay the same between comparable images, what I am asking is whether there are specific values I should keep on all my snapshots to obtain linearity. For example: should I desaturate the image or does that change the intensity (measured as R+G+B) in a nonlinear way? Should I choose a different method to measure image intensity?

Answer

So long as they're fixed it shouldn't matter very much, except for gamma correction which should probably (there are always edge cases) be set to 1. Gain and brightness should probably be low to reduce the dark count. You must fix the exposure time. If that's not an option the experiment won't work.

Having done exactly this I'd urge you to calibrate (which would also be a good test). Get a 10% (OD1) and some 50% (OD0. 3) filters and check that 50% intensity gives you 50% counts. You may need to sum over the spot size. These filters aren't perfectly accurate (though probably more accurate than the webcam) so you still have some error. Varying the laser current won't work, at least at low or high powers (for that laser, and assuming a diode laser) as they're not very linear.

Also check that you avoid saturation. This is hard to do on some cameras, but you'll detect it as constant max value pixels in the red channel first. The max value may not be 255 by the time the software has messed around.