Comparing imaging and non-imaging techniques for reducing background clutter and resolving distant point sources

To reach maximum sensitivity, any method used to search for orphan sources must be insensitive to local variations of the background. Using imaging and non-imaging techniques, we analyzed the same data acquired by a large-area, coded-mask-imager search instrument. Data from many passes past a 1 mCi source at 65 m from the instrument were used to construct a model of the instrument response. We then used the model to "hide" the source in data taken in a light urban environment. We compared the success of detecting the hidden sources using coded-mask imaging methods, pseudo-imaging based on a zero-area matched filter, and non-imaging using simple thresholding. The results clearly indicate the superiority of imaging with the coded-mask techniques returning the best results.