maxthemouse / multipleimageradiography Goto Github PK

One issue I run into when developing is making sure that there is no sensitive information left in the notebooks. For example, the path to the data can have project numbers, sample names and the like that should not be saved in git. Using a configuration file would make it easier to default to generic values while making it easier to save project specific values that can be saved with the data.

The current X-axis is arbitrary. Could use angle units.

Look at multi-threading or multi-processing to improve performance.

The code was updated to Qt-5. Would like to update to Python 3. This was actually done once and then reversed due to installation issues. Hopefully, these issues are not an issue now.

This code is only for MIR. The same approximation equations can be used for CT data. It would be useful to add the ability to process CT projections. I won't add CT processing but it may be useful to add an inference to UFO. It would be difficult.
https://github.com/ufo-kit
Other options include tomopy https://github.com/tomopy/tomopy and scikit-image https://scikit-image.org/.

While making the settings page it was noticed that there are two items that are not possible to put in a JSON file

1. output = io.Text_Output()
2. image_filter = scipy.ndimage.median_filter

These are objects. These objects are used to use specific functions where there are multiple options. These have to be put in at the appropriate time and are not particularly documented. These should be replaced or removed.

Currently there is an out of date docx file which is not in the repo. It needs to be updated. I also want to pick a better format such as markdown or even rtf.

It seems like a useful thing to move the stitching function into a python file for import. This would make it available for use in other scripts. I assume that it was done in the notebook as a way of development.

Current advice is to use numpy.polynomial instead of the older numpy.poly.
The fitting is supposed to be more numerically stable. Note that the coefficients are in the reverse order. The documentation has equivalents to change code.

The paper " A Gaussian extension for Diffraction Enhanced Imaging" has an algorithm for estimating the absorption, refraction and scattering portions of diffraction images. This is different than the Taylor series-based equations that I am using. I would like to add the newer equations as they may be better for some samples.

It should be possible to set-up doing all calculations for data in a folder. I.e. all samples. This should include multicalc, synthetic images, stitching and being able to add new calculations.
This probably could be done in notebooks but actually having a mechanism would be nice.

There are lots of options that are not implemented. Some of them have to do with manual file naming and ideas on what might be interesting to work on. These need to be cleaned up and either implemented or removed. This will also improve the GUI.

The performance is not ideal although not a severe issue, in my opinion. Other people may beg to differ. One idea would be convert to Julia. That changes the requirements but may not be overly onerous.
Polynomials.jl seems to have the fitting function needed although weights are defined differently. Reading tiff and bigtiff files will need to be supported. Looks like TiffImages.jl might be a good fit but need to test if 8/16 bit integer and 32 bit float are all supported.

I left out the Jupyter notebooks but there are a few that were created for testing and as a possible way of using the software. Would like to clean-up and and re-add the notebooks.

I have created some installers but I think most don't actually work. I have tried several tools and it is not clear what actually works. One issue is that the resulting distribution is missing files so it doesn't actually work if the computer doesn't have Python installed in the same place as my machine.

I think py2exe was working and cxfreeze was not. I suspect the code is a now a mixt of the two, i.e. setup.py and this needs to be fixed. It should be assumed that the building of a windows application is broken.

Recently collected data was in bigtiff format. Need to support reading and writing bigtiff format.