Example request: Pan-sharpening of Cornfield sample data

[karlHolmquist]

Hi, would it be possible to request an example for doing pan-sharpening of a measurement in post-processing?

Example input: measurements from the cornfield/session_002 data (.cu3 file, with black and white calibration, and pan-chromatic .tiff file)

Example output: New pan-sharpened measurement + either save to .tiff or visualization of single band using matplotlib.

Basic functionality:

• Given path to measurement
• Load measurement and required calibrations and pan-chromatic image
• Setup of processing (with clear settings of algorithm to use)
• Generation of pan-sharpened output
• Visualization and/or saving of generated output to file

[robertCubert]

Hi Karl,

thanks for you input! We are honoring your request and will provide an example in the near future!

[MaxC-Cubert]

@karlHolmquist , we created an example on how to pansharpen images. Let us know if you have any questions.

Best,
Max

[karlHolmquist]

Thanks Max,

A follow-up question, I noticed that the example exports the cube as a tiff file. Does the SDK allow for directly accessing the pansharpened cube as a numpy array or is it necessary to go through one of the exporter classes?

[MaxC-Cubert]

Directly accessing the pansharpened cube is currently not supported by default. The recommended procedure includes the export as shown in the example.
As a workaround, you could theoretically use cuvis.Viewer to generate a pansharpened cube, as this class supports pansharpening operations as well. To do so, you would have to write a userplugin (XML), which returns the whole cube, set this along the pansharpening settings in the cuvis.ViewerSettings, initialize a cuvis.Viewer with those settings and apply it to your measurement. That being said, I am not sure if someone tested this before.

I hope this helps.

Best,
Max

[birkholz-cubert]

A userplugin that directly returns the cube would look like this:

<?xml version="1.0"?>
<userplugin xmlns="http://cubert-gmbh.de/user/plugin/userplugin.xsd">
    <configuration name="PassThrough_Cube" citation="" plugin_author="Cubert GmbH"
        plugin_version="1.0.0" required_engine_version="3.2.0">

        <comment>
            This plugin directly returns the hyperspectral data cube.
        </comment>

        <!-- Output directly visualized cube -->
        <output_data id="Cube_View">
            <cube />
        </output_data>

    </configuration>
</userplugin>

to then get the pansharpened cube you would initialize the cuvis.Viewer as follows:

    if 'pan' in mesu.data.keys():
        mesu.cube  # regenerate cube if not available (this is the default for measurement saved in newer software versions)
        cube_plugin = Path(__file__).parent / 'cube_only.xml'
        sett = cuvis.ViewerSettings(userplugin=str(
            cube_plugin), pre_pan_sharpen_cube=True, pan_scale=1) # pre_pan_sharpen_cube, tells the viewer to first apply pansharpening and the calculating the userplugin output
        viewer = cuvis.Viewer(sett)
        cube = viewer.apply(mesu)
        cube.wavelength = mesu.cube.wavelength # if you would like to still have the wavelength information

    else:
        cube = mesu.cube

As for the python wrapper, there needs to be an additional bugfix applied that is currently located on the release/v3.3 branch and will be included in the next minor release.
But you can install this by hand.

If you encounter any issues let us know.