Hello, I am using GDAL through OTB (the Orfeo Toolbox) to process TSX/TDX products with the specific driver: COSAR/COSAR Annotated Binary Matrix (TerraSAR-X).
With TSX products, the driver works fine. I have also tried to read CoSAR files from TDX products. Data have a huge difference with a floating point data format instead of integer values. This floating format is a specific format called "Half Precision Floating Point" for each of the complex values. Thus, TDX products are not well-interpreted. Further information are available in TD-GS-PS-3028_TanDEM-X-Experimental-Product-Description_1.2. The annotations seem to be the same than TSX products. I need your help to include this new format in GDAL and so in OTB. As far as I can understand, the specific driver for CoSAR binaries is implemented in https://github.com/OSGeo/gdal/blob/master/gdal/frmts/cosar/cosar_dataset.cpp. The data type is set to GDT_CInt16. I think that explains why TDX are not well-interpreted. In my opinion, in order to include TDX products a conversion has to be made before assigning the buffer. I have found an algorithm to convert "half float" to single float : http://blog.schmorp.de/data/binary16/fasthalffloatconversion.pdf Do you think, is it possible to implement this feature in GDAL? Regards, Gaëlle. -------------------------------------------------------------------------------------------------------------- Gaëlle Usseglio -Thales Services Responsable de prestation : Support HPC (Portages, Optimisations, Audits, Expertises, etc.)
_______________________________________________ gdal-dev mailing list gdal-dev@lists.osgeo.org https://lists.osgeo.org/mailman/listinfo/gdal-dev