Spot 5 orthorectification
Mantis Issue 97, reported by echristophe, assigned to jmichel, created: 2009-03-18
Seems to be some improvement needed with the orthorectification of SPOT5.
Take the otbOrthoRectifApplication Open the OTB-Data/LargeInput/SPOT5_SCENE01/IMAGERY.TIF Keep the parameter by default and orthorectify (without DEM)
Few hours later (profiling needed! use RPC model instead?), you can see the result: towards the bottom of the image, there is a black line running near the edge. (same on the right edge)
It should not be the case.
I assume it's a confusion with pixel coordinates from the center or the top left corner of the pixel.
1306339492 - christopMight be related to the nan from the sensor model outside the image.
1306403023 - julienmNaN problem shown in http://hg.orfeo-toolbox.org/OTB/rev/14dd7a3a9b2e
and fixed in http://hg.orfeo-toolbox.org/OTB/rev/648f6b0266e2
1306423940 - julienmAttached are the result of orthorectification of a part of the Teheran Spot5, showing the effect of the "NaN fix".
The artefact is less visible, but there is still a problem.
The artefact are there whether we use a DEM or not
1306424375 - julienmAnother note. nothing better on the Arcachon Spot 5 in the LargeInput : we still can't orthorectify it as some metadata are wrong or badly interpreted (NaN as corner coordinates)
1306432597 - julienmUsing the new output of http://hg.orfeo-toolbox.org/OTB/rev/0420230cc2ff, I attached two graph showing a big discontinuity in the sensor model when crossing the image border.
generation :
- take two points on the two side of the border (in the center, at the bottom)
- tran