I’m trying to mosaic two Landsat inages and can’t apply harmonization.
Can’t load the input files and the result, source files are about 800 mb each and the result is about 1,5 Gb
simple mosaicing works well
this is the log:
2020-04-19 00:29:48 (INFO) Mosaic: Temporary files prefix is: D:/P5605_DB/DIST/mband/105_106_harm
2020-04-19 00:29:48 (INFO) Mosaic: Using LAB color space for harmonization
2020-04-19 00:31:02 (INFO) Mosaic: No feathering
2020-04-19 00:31:02 (INFO) Mosaic: Correction cost function: RMSE based cost function
2020-04-19 00:31:02 (INFO) Mosaic: computing correction model for band 0
2020-04-19 00:31:02 (INFO) Mosaic:
[ Band 0 ]
Gains : nan nan
2020-04-19 00:31:02 (INFO) Mosaic: computing correction model for band 1
2020-04-19 00:31:02 (INFO) Mosaic:
[ Band 1 ]
Gains : nan nan
2020-04-19 00:31:02 (INFO) Mosaic: computing correction model for band 2
2020-04-19 00:31:02 (INFO) Mosaic:
[ Band 2 ]
Gains : nan nan
2020-04-19 00:33:22 (INFO) Mosaic: Execution took 213.552 sec
Hello,
It looks like Mosaic cannot compute valid statistics in images overlaps.
How does it looks without radiometric harmonization? Do your images have the same coordinate reference system?
More crucial: Do your images overlap? The algorithm that performs the radiometric harmonization is based on the statistics computed on valid pixels (i.e. different that the nodata value) in the overlapping areas of the images. If there is not available pixels in these overlapping parts, it would work, and you have to to the radiometric harmonization with another method prior to the mosaic process.
Also, you can use the harmo.method bands (lab is for true colors RGB image only)
Hope that helps!
Rémi
I took another try, and clipped images to my work area, so they are now about 300 mb, i you could take a look at them, I’d be very grateful:
https://drive.google.com/open?id=1z2_InWob6RVX93iC3A0qlI2RgqNpG3_B
The images nave the same CRS, what format do the statistics have to ba calculatied in? I tried to calculate them to xml with the same name as raster, it didn’t help.
If there is not available pixels in these overlapping parts, it would work, and you have to to the radiometric harmonization with another method prior to the mosaic process.
Could you please specify, what tool are you referencing to
Hi Geogiy,
From the log you provide, I think you have used the “true colors” mode for harmonization. It seems that it only works for 3 bands images, and yours contain 6 bands.
With your images, I managed to make the mosaic with harmonization in “bands” mode.
Hope that helps!
Yannick