Evaluation of coded aperture radiation detectors using a Bayesian approach
Abstract
We investigate tradeoffs arising from the use of coded aperture gamma-ray spectrometry to detect and localize sources of harmful radiation in the presence of noisy background. Using an example application scenario of area monitoring and search, we empirically evaluate weakly supervised spectral, spatial, and hybrid spatio-spectral algorithms for scoring individual observations, and two alternative methods of fusing evidence obtained from multiple observations. Results of our experiments confirm the intuition that directional information provided by spectrometers masked with coded aperture enables gains in source localization accuracy, but at the expense of reduced probability of detection. Losses in detection performance can however be to a substantial extent reclaimed by using our new spatial and spatio-spectral scoring methods which rely on realistic assumptions regarding masking and its impact on measured photon distributions.
- Publication:
-
Nuclear Instruments and Methods in Physics Research A
- Pub Date:
- December 2016
- DOI:
- 10.1016/j.nima.2016.09.027
- arXiv:
- arXiv:1605.08499
- Bibcode:
- 2016NIMPA.839...29M
- Keywords:
-
- Coded aperture;
- Wide area nuclear detection;
- Distributed sensor networks;
- Statistics - Applications
- E-Print:
- Submission abstract for poster presented at nuclear science symposium (NSS) 2015