I have a plate with several pinholes (8+9=17), underneath which there is a light detector (rectangular region in the cross-sectional view).
The point light is defined at the red point directly above the holes. When the only light source is the point light, I would like to see the image of the light source on the detector after the light travels through the holes. It should look similar to this:
but the cross-sectional picture of the detector only appears to be uniform in light intensity. 
Is there a way to more accurately represent the scenario?
I'm curious how you are approximating the light on the detector now. Are you using a ray trace add-in or are you using a rendering software? I use the OptisWorks add-in for this type of thing and the software will give you a very accurate prediction of the energy you will get on your detector assuming the model inputs are accurate. There are several lighting / ray trace add-ins available for SW like FRED, ASAP, Lightools, and OptisWorks in the SW partners listings. The software can seem expensive depending on the type of business or industry you are in and the capability of the software.