Fork me on GitHub

Home

ImgBeamer

Try it live here, no setup or installation required!

Simple demo tool of the image formation1 process used in a Scanning Electron Microscope (SEM).

A quick start guide is available here.

The source code is available here.

Description of Stages / Boxes

  1. Sample Ground Truth (map navigation): a view the full image where the highlighted area represents the subregion area.
  2. Subregion or ROI (Region of Interest) View ("zoomed" view): a view of the subregion area as highlighted in the Sample Ground Truth.
  3. Spot Profile: visualization of the spot profile (shape and size) by changing the relative width/height, scale, and rotation.
  4. Spot Content: visualization of the spot content (Subregion “stenciled” with the spot profile) or area sampled by the spot.
  5. Spot Signal (rgba): the signal or pixel value to represent what has been sampled by the spot or beam, or the average pixel value from the Spot Content.
  6. Spot Layout: the layout of the sampling grid or array of spot sampling positions over the Subregion area.
  7. Sampled Subregion: the sampled or "stenciled" content of the Subregion as depicted by the Spot Layout.
  8. Resulting Subregion: the resulting image by filing each pixel in the grid by the signal of each corresponding spot as depicted in the Spot Layout.
  9. Resulting Image (full, "virtual SEM"): the resulting image full where the imaging process, as shown in the Resulting Subregion, is continued for the full extent of the Sample Ground Truth image.

Screenshot

screenshot

Notes

  • Minor pixel value differences in virtualSEM; improved, but could be due to sampling from downsized sampling of the larger full image
    • This is possibly due to canvas "alpha premultiplication", see here.
  • The application design is being documented here.
  • main is the stable branch, cas741 is an outdated development branch.

Developer Instructions and Notes

  • The main application is implemented entirely in javascript
    • You can run it either using the auto-deployed GitHub pages version ("live" link above).
    • Or you can run it on your own machine with a local web server pointed to the app/ folder
      • either XAMPP or even just a python HTTP server with python -m http.server --directory app/
    • jsdoc-generated documentation pages are available here
  • Tests for the image metrics are available under tests/:
    • og is code for "original" (unmodified) image, and fant refers to a down-sampled image using Fant's algorithm2.
    • js-tests/: for javascript implementations
      • open the index.html page (using a local web server as described above) and look in the webconsole.
      • Optionally, an online hosted version exists here.
      • mainly in the webconsole, you can use run_all(fant) where if fant is true, all the image comparison tests will be run using the "fant-sampled" image as the ground truth. Otherwise (false), it will use the "original" image as the ground truth instead.
    • py-tests/: for python implementations
      • Install Python v3.10.6 or better (has not been tested on other versions)
      • You'll likely need to run pip install sewar once to get the required image metrics module/library.
      • run imgquality.py
    • The other folders and files are from preliminary and manual testing that was done in the past.

References

  1. “Image Formation.” In Scanning Electron Microscopy and X-Ray Microanalysis, edited by Joseph I. Goldstein, Dale E. Newbury, Joseph R. Michael, Nicholas W.M. Ritchie, John Henry J. Scott, and David C. Joy, 93–110. New York, NY: Springer, 2018. https://doi.org/10.1007/978-1-4939-6676-9_6.
  2. Fant, Karl M. “A Nonaliasing, Real-Time Spatial Transform Technique.” IEEE Computer Graphics and Applications 6, no. 1 (January 1986): 71–80. https://doi.org/10.1109/MCG.1986.276613.

↑ Top