I was interested by your recent article about IR cameras [“The Infrared Choice,” Vision Systems Design, April 2011], being a designer in the field. However, in one part you talk about oversampling by the small pixels (second to last paragraph). Your description based on the Rayleigh criteria (diffraction-limited spot size)—although good for stars—is not useful for extended source imaging.
The MTF [modulation transfer function] curve is a better tool to use. Here is an excerpt from a recent analysis I did to show how it works.
Take the MTF curve for a diffraction-limited spot of an f/1 lens at 10-μm wavelength, and multiply it by the MTF of the 25-μm pitch array and you get the graph seen here [see figure].
As you can see, the lens MTF goes out to more than 8 cycles/mrad (black curve), whereas the 25-µm pixel pitch causes a cutoff at 4 cycles (pink curve). If the pitch were reduced to 12.5 µm, the array and the diffraction spot size would be well matched.
This is what most manufacturers in the industry are doing, reducing their pitch. However, it is a slippery slope because the signal goes down as the square of the pitch reduction, so the signal-to-noise suffers.
Bodkin Design & Engineering LLC, Wellesley, MA, USA
1. W.J. Smith, Modern Optical Engineering: The Design of Optical Systems, 3rd Edition, McGraw-Hill, p. 377 (2000).