Skip to content

Version 2.3.0: SLIXCluster, Rework of SLIXLineplotParameterGenerator and several new additions

Compare
Choose a tag to compare
@Thyre Thyre released this 09 Sep 13:14
· 134 commits to master since this release

v2.3.0

Added

  • Added SLIXCluster, a tool which can be used to separate regions based on classification data through SLI measurements. The parameters aren't final yet and might change in the future.
  • Added SLIX.classification which includes the classification methods of generated parameter maps.
  • Added a new parameter to SLIXVisualizeParameter [...] vector named --distribution which allows to plot all vectors in a selected threshold region. This allows to see which regions have a high confidence in their reported orientation and which regions might not give any information. Please note that the performance for full measurements isn't that high in the current state.
  • Added SLIX.visualization.unit_vector_distribution which is used for the creating of the image described in the last bullet point.
  • Added DPI option to SLIXVisualizeParameter. The default was 1000 dpi but images with a lower dpi value might be interesting when presenting the image for a paper or presentation.
  • Added --value and --saturation to SLIXVisualizeParameter [...] fom which allows the user to set two images to weight the FOM.

Changed

  • Restructured some of the hidden methods to other Python files to make the basic package infrastructure easier to read.
  • Renamed methods in SLIX.visualization to prevent the repetition of visualize
  • The README.md will now use GitHub asset links instead of the old repository used during the first implementation of SLIX.
  • Changed the structure for the command line programs from _cmd.py to a package named _cmd containing all programs in separate files. Both solutions work but the new solution allows SLIX to scale more easily in the future.
  • When using the smoothing parameter in SLIXParameterGenerator, the resulting files will include the chosen algorithm as well as their respective parameters.
  • Rework of SLIXLineplotParameterGenerator. The tool will now generate way more detailed parameters when called with default parameters. You can use the --simple option to reduce the information to average values (just like before). The CSV structure was also altered to make parsing the results easier.
  • SLIXLineplotParameterGenerator now also supports the smoothing options from SLIXParameterGenerator
  • Multiple line profiles will be calculated in parallel when using SLIXLineplotParameterGenerator
  • When reading multiple files from a folder with SLIXParameterGenerator the filename will now not match the folder name but will instead use the file pattern in the read files instead.
  • Disabled the GPU usage for unit vectors in SLIXVisualizeParameter until a better solution for large input files has been found.
  • When using SLIXVisualizeParameter, a warning will now be shown when the background image dimensions do not match the image dimensions of the direction.
  • When using SLIXVisualizeParameter, the background image will only be rotated when it would match the orientation of the direction images

Fixed

  • Fixed a bug in the tests of the visualization images where the Matplotlib figure wasn't cleared as expected. The tests did check the right things though. It only resulted in a problem when adding another test for the new parameter.
  • Fixed a bug where a 3D stack with the shape (1, 1, x) would not be processed correctly with SLIXParameterGenerator
  • Fixed a bug in SLIXVisualizeParameter which could occur when the thinout is exactly the image size