I did some searching but nothing came up around this. I'm trying to verify a workflow from images to an NDVI to verify I have a basic workflow correct.
So here's where I tried to manually calculate an index because I'm not sure why the NDVI looks incorrect. I zoom into what I KNOW is a plant, take the value of a pixel's R (0-255) and the value of NIR(0-255). But what I noticed is that the R value is pretty high.
For example a plant pixel is:
Let R = 123.0Let NIR = 158.0
( 158.0 - 123.0 ) / ( 158.0 + 123.0 ) = 0.124
I then look at the value using Value Tool in QGIS and sure enough, that's what SAGA calculated so I know it's doing a pixel value calculation exactly the same way I am to figure out the NDVI for that pixel.
That should be > 0.4 or even higher if it's a plant and it's healthy (I'm pretty sure the field is healthy).
So somewhere in my workflow, I think something isn't right and I'm not sure what?
I know the source images are standard RGB values (0-255) for each channel. I know the Blue Channel (NIR) is more pronounced which I'd expect if that is the NIR. But for the NDVI to be proper, I think the R channel should be a lot less than the NIR channel. I think something in my workflow is messed up.
أكثر...
- Process images example dataset through AgiSoft. I'm doing a very small region around the cross-roads
- Export a planar TIFF
- Open the planar TIFF in QGIS, Split the RGB bands (SAGA) and export raster layer using float (GDAL).
- Generate a NDVI using SAGA from the R and B bands. B should be the NIR data.
So here's where I tried to manually calculate an index because I'm not sure why the NDVI looks incorrect. I zoom into what I KNOW is a plant, take the value of a pixel's R (0-255) and the value of NIR(0-255). But what I noticed is that the R value is pretty high.
For example a plant pixel is:
Let R = 123.0Let NIR = 158.0
( 158.0 - 123.0 ) / ( 158.0 + 123.0 ) = 0.124
I then look at the value using Value Tool in QGIS and sure enough, that's what SAGA calculated so I know it's doing a pixel value calculation exactly the same way I am to figure out the NDVI for that pixel.
That should be > 0.4 or even higher if it's a plant and it's healthy (I'm pretty sure the field is healthy).
So somewhere in my workflow, I think something isn't right and I'm not sure what?
I know the source images are standard RGB values (0-255) for each channel. I know the Blue Channel (NIR) is more pronounced which I'd expect if that is the NIR. But for the NDVI to be proper, I think the R channel should be a lot less than the NIR channel. I think something in my workflow is messed up.
أكثر...