GPU Accelerated Image Processing in CCD-Based Neutron Imaging

Abstract

Image processing is an important step in every imaging path in the scientific community. Especially in neutron imaging, image processing is very important to correct for image artefacts that arise from low light and high noise statistics. Due to the low global availability of neutron sources suitable for imaging, the development of these algorithms is not in the main scope of research work and once established, algorithms are not revisited for a long time and therefore not optimized for high throughput. This work shows the possible speed gain that arises from the usage of heterogeneous computing platforms in image processing along the example of an established adaptive noise reduction algorithm.

Comparison of Thermal Neutron and Hard X-ray Dark-Field Tomography

“…The detection system consisted of an Andor Neo CMOS camera coupled optically to a 100 LiF/ZnS scintillation screen and resulted in an effective pixel size of 33 (66 after binning for reconstruction). For every phase step, 3 frames with 5 exposure time each were acquired to remove complex noise structures, such as gamma spots [ 28 , 29 ]. A phase scan of every projection consisted of 8 phase steps, and in total 360 views were collected over a 180 sample rotation.…”

“…In the case of neutron data, the acquired images were corrected for dark current with averaged dark frames, and defective pixels were interpolated by a median with a kernel size of 3. Afterwards, artificial bright structures (gamma spots) characteristic for neutron scintillator-based CCD detectors were removed by a slightly adapted algorithm compared to the standard procedure [ 28 , 29 ] described in the following. For every frame, a median-filtered image with a kernel size of 9 pixels was calculated and subtracted from the original image.…”