view article
Figure 11
Creating a normalized background-noise image from experimental LabDCT projections. The figure shows (a) an experimental LabDCT projection. For part (b) most of the spots in the experimental projection are identified and removed by a segmentation method combining a Sobel filtering and binary thresholding. For (c) a filling-mask image is obtained by convoluting the image containing the segmented spots with a Gaussian filter (size = 2 pixels and SD = 1 pixel) followed by a dilation in a disk shape with a radius of 13 pixels. Part (d) involves filling the regions in (b) specified by the mask image (c), while (e) is the background-noise image obtained by rolling median over 60 sequential images of (d) followed by dividing by the maximum pixel value. Finally, (f) is a color scale of (e) showing the heterogeneous gradients of intensities.

IUCrJ
Volume 8| Part 5| September 2021| Pages 719-731
ISSN: 2052-2525
https://doi.org/10.1107/S2052252521005480