![]() hdf is not the same as the 16-bit integer representation, so the values need to be stretched to during this process. ![]() Slices should be extracted from each, and then merged later. Often, such as with the fragments scanned in that session, there is a separate. For example in the 2019 Diamond Light Source scans, the reconstruction output 32-bit float. ![]() For synchrotron scans, this step may be necessary. For benchtop sources, this is probably already done as part of the reconstruction process. The goal of this step is to get a 16-bit. Generate/extract slices, window, and crop x-y withing the ROI manager to see the bounding box specs. You can either mouse over the edges and get close estimates from viewing the mouse coordinates in the ImageJ main window, or you can use Analyze -> Tools -> ROI Manager, add the current ROI, and use More -> Specify. Record the x and y offsets as well as the width and height of the bounding box you have selected. When in doubt, it's OK to make the bounding box a little bigger to be sure it includes the object. It is good to crop tightly to create smaller datasets, however it is important to be sure to contain the entire object. Scrub through the slices, adjusting the bounding box to make sure it is always outside the bounds of the object of interest. Load the volume in Fiji/ImageJ, and select a rectangular bounding box. The next step is the same for either the original full set of slices (if that is feasible and on your machine) or the subset sampled using something like the above. output-dir /pscratch/seales_uksr/nxstorage_partial_copy_in_case_gemini_down/data/Herculaneum_Scrolls/PHercParis2/Frag47/CT/54keV/_sample_every_100_slices/ \ Python /usr/local/dri/ink-id/inkid/scripts/hdf_extract_slices.py \ hdfs (multiple desirable if object scanned in slabs) like this: Those can be extracted all at once from one or more. hdf files from Diamond, it is easiest to first extract some sample slices on LCC, transfer those to your machine and then view them there. If the slices are already generated and on your local disk in one volume, which is likely if they came from a benchtop machine, those can be viewed directly. It is necessary to load the volume into a visualizer such as ImageJ to determine the crop bounding box.
0 Comments
Leave a Reply. |