Notebooks for processing data from the NPEC greenhouse (Module 5)
Notebook list: planteye.ipynb snapscan_preprocessing.ipynb
This notebook processes hyperspectral ENVI images to segment green leaves, extract their spectral information, and save visual and quantitative results.
For each image in the specified folder, it loads the data, crops the image to just include the plant, and applies a segmentation algorithm that identifies green leaf pixels.
The code then calculates the average and standard deviation of the spectra for those pixels, and saves both the segmentation mask and an NDVI image as PNG files. Finally, it compiles the results for all images into an Excel spreadsheet, including the filename, average spectra, and standard deviation per wavelength, enabling further analysis or visualization.
This notebook gives a few examples of trait extraction from segmented RGB images. An elipse and boundingbox fit is used to get a height and width estimate in pixels. Checkerbord calibrations have shown pixels represent an equal number of mm's across the entire image. Therefore these pixel counts can use to compute the size in milimeters. The checkerbord images and code used to find this conversion factor is available on request.
Next to the shape analysis a simple color analysis is shown which can in somecases say something about plant development or health.