NDVI is an acronym for Normalised Difference Vegetation Index. It’s a very interesting measure of plant photosynthesis and health derived from measuring the difference in near-infrared and visual reflectance. I’ve recently been exploring its use for plant monitoring; here are a few calibration images.





The original image is on the left, with NDVI on the right

These images were shot using a full-spectrum modified Sony A7 camera with OM Zuiko 21mm/3.5 Lens and Schott BG3 filter at ISO 100, F11, 150 to 1120 sec exposure.

The NDVI images were generated using the open source software Fiji with the PhotoMonitoring Plugin (written by Ned Horning).

There’s a bit of calibration to do, the balance, spread of the NDVI and scaling of the false colour image is not quite right yet, but it’s already clearly visible that the photosynthesising vegetation is visible as strongly yellow in the image. The dark hues of the clouds also give accentuation to their fractal forms.

The excess signal on buildings should be correctable by white balancing more carefully. The source images are quite pink and I have stretched the IR range significantly to get good differentiation, this should be improved markedly by white-balancing to a light blue target.

I will also be trying out a red filter (Hoya 25A), there appears good reason to use red filters instead of bispectral blue (or even orange) to get better separation of NIR and VIS in the camera’s sensors.


Thanks to the many contributors at Public Lab who’ve pointed the path with practical NDVI experiments.

Edit 2015-7-17 The setup is working better and have tuned the white balancing as well as getting used to the processing options in the PhotoMonitoring plugin. I am surprised by being able to extract good IR data in quite challenging environments. The image of the dog below has been contrast and white point adjusted after NDVI conversion and shows that in good lighting and with warm subjects, quite a bit of IR data can be extracted.