I am using OTB to make TOC calibration from WV3 images.
I get a result of a white image (values = 1). i tried to correct gain and bias text file but i have the same problem every time.
Some one had the same problem?
It is the first time I see this error. Sorry you encountered it.
Could you tell us what version of OTB you are using, how do you run the application (command line, QGIS, other), the parameters you set, and the logs you obtain? With this information, we will try to find out what’s going on.
For gain and bias, i use this structure for .txt format as described in documentation :
Here is an XML file showing parameters used :
OTBparamameters.xml (8.7 KB)
Here is the file containing log information :
log_OTB_calibration.txt (5.5 KB)
I use both monteverdi and Qgis3.14 and OTB version is 7.2.0
I am surprised to read in the log file that all your bands have the same wavelength (0.65µm). Maybe a problem with the metadata. Could you try to run a gdalinfo on one of the images?
Your gainbias.txt file seems well formatted.
I have the same problem but with images gaofen-1 WFV4 and running directly in monteverdi.
Dear @julienosman, @fils
Were you able to solve this issue? I have found the same error while trying to convert a 8 band WV-3 image to toc; My log file also shows all bands with the same wavelength (0.65µm).
I am running otbcli_OpticalCalibration inside R, and I am not being able to run with the “atmo.rsr” file, which I took from here (Input/Radiometry/DigitalGlobe/radiance_response_WV03.txt · master · Main Repositories / otb-data · GitLab) and use only untill 1100 nm.
Any thoughts you can share with me would be much appreciated.
After some research, it appears OTB is not compatible with WV3 products because metadata are not read (see here).
The release 8.0 of OTB coming this fall will largely simplify the way OTB deals with metadata, this should make it more simple to fix this issue.
@julienosman Thank you for replying. It is a great news that the release of OTB 8.0 will simplify things.
I found this link that you’ve posted and managed to interpolate the radiance response at 0.0025 as mentioned. In fact, I was able to perform the optical calibration, but I get some areas where the reflectance saturate. I think is because of the gainbias parameters now. I simply copied the gain and offset values from a digital globe file, but I suspect I am missing something. Is it because OTB doesnt read the WV3 metadata or I am looking to the wrong values? Sorry to bother, but I theres not much documentation on the subject. thank you!