Flat accurate nonimaging point locator

Suppose there are one or a few point sources out there at infinity and our jib is to extract information about their location (θx and θy) relative to the sensor normal and their signal intensity I at the sensor. For each point, we need at least three measurements that embody those parameters and seek to compute the data we need. Imaging, the traditional way of doing this uses the presumption that θx and θy are converted to spatial coordinates in the image plane with a known (possibly θx and θy variant formula), so θx and θy are easily found. It farther assumes that the intensity at the point in the imaging plane is proportional to the intensity of the source. Here we examine the possibility of nonimaging systems for obtaining those data and argue that the nonimaging approach should be preferred in some cases.