I have bi-weekly UAV flights in soybean trials and I was wondering about removing or not pixels classified as "soil" from the images to calculate veg. indices. Initially, I am calculating for of the each image the Excess Green index (ExG = [2 * G - R - B] / [R + G + B]) and than assuming a threshold to characterize plant and "non-plant" pixels. Of course, this procedure I am doing while the canopy is still not fully covering soil between rows (i.e. early stages of the crop).
Then, using a zonal statistics in Qgis, I am retrieving the statistical metrics (avg, min, max,..) of NDVI (or any other VI) from a specific rectangle placed on each of the plots.
I would appreciate any thoughts about remove or not those pixels classified as "soil"?
Tks,
removing or masking the soil pixels can be beneficial for calculating VIs at the beginning of the crop cycle, when the vegetation is sparse and the soil is visible. This can improve the quality and reliability of the VIs and their interpretation for agricultural applications. However, the method of removing or masking the soil pixels should be chosen carefully, depending on the type of VI, the type of crop, the type of soil, and the availability of data.
- Badges
- Science topic
- Similar topics
- Vectorization