Comparison of Image Filters for Low Dose Neutron Imaging

Autor: W.J. Lewis, L. G. I. Bennett, J. Metzler, P.C. Hungler, George Bevan
Rok vydání: 2013
Předmět:
Zdroj: Physics Procedia. 43:169-178
ISSN: 1875-3892
Popis: Neutron imaging using low flux sources, such as accelerators or low flux nuclear reactors, produces images which contain significant amounts of noise. The noise indications are a result of high energy gamma radiation and some neutron scattering which hit the CCD detector despite heavy shielding. The amount of noise in an image is a factor of the exposure time required to produce images with adequate dynamic ranges. Minimization of noise and maximization of the dynamic range are inversely proportional and the exposure time is often extended to increase incident neutrons at the expense of noise. The resultant noise can be reduced using image filters; however, these filters usually increase the signal to noise ratio (SNR) at the expense of spatial resolution. Three filters were applied to low dose neutron images acquired at RMC; a median filter, a Z-projection filter and a hybrid PDE filter. The median filter and the hybrid PDE filter showed similar performance in 3D with regards to SNR and spatial resolution, however, the median filter created numerous artefacts in the resultant tomogram. The Z-projection filter using 5 projections had the best performance in 2D improving the SNR of the raw image from 10.2 ± 0.767 to 22.5 ± 1.52 and the spatial resolution from 331 ± 2.89 to 309 ± 0.846, respectively. The Z-projection filter was not evaluated in 3D due to facility induced constraints.
Databáze: OpenAIRE