Comparison of the modified uniformly redundant array with the Singer array for near-field coded aperture imaging of multiple sources

Abstract

Coded aperture imaging (CAI) based on the modified uniformly redundant array (MURA) has been successfully applied in far-field imaging to locate possible weak radioactive hotspots, such as in astronomy and nuclear security. However, there are difficulties in its direct use in near-field nuclear imaging because of severe artifacts caused by the MURA method itself. Recently, a novel attempt based on the Singer array (SA) has been widely discussed to improve the image quality of CAI. In this study, we carefully compare and evaluate the image artifacts and contrast-to-noise ratio (CNR) of the MURA and SA methods under the same near-field conditions. First, the definitions of near-field CAI, two decoding algorithms based on deconvolution and iteration, and the principles of the SA are briefly illustrated. Second, a series of simulations for near-field CAI of multiple sources based on the MURA (19 × 19) and SA (17 × 21) methods are discussed in detail. Third, controlled experiments based on the MURA and SA methods for near-field CAI of multiple sources are set up. The simulation and experiment results show that compared with the regular but difficult-to-remove artifacts in the MURA images for near-field CAI, the artifacts in the SA images are irregular but can be eliminated through finite iterations. For single-source imaging, the CNR of the MURA image is better than that of the SA image, but for multiple sources, their CNRs are roughly equivalent. Therefore, the SA method is more suitable for near-field CAI applications of multi-source objects.