IMAGINE Discussions

Discuss and share topics of interest using ERDAS IMAGINE the world’s leading geospatial data authoring system.
Showing results for 
Search instead for 
Do you mean 
Posts: 36
Registered: ‎05-02-2018

Orthorectification of several Satellite Images

Hello everyone,

I've got some Pleiades images covering 150x75 km approximately, we can name this area AOI. This site is divided on several smaller area (30x20 km approx.) and for each smaller area I've got a triplet (so, 3 images), we can name these area SAOI. I can build Orthophotography from each Pleiades images with Erdas Images using the Orthorectification without GCP tool. But, when I build 3 orthophotography of the 3 Pleiades images covering a same SAOI, the Orthorectified images aren't georeferenced well (a same object on the 3 images is approximately shifted from 10 m) and it's the same with two different SAOI, there is some shifts between the two SAOI (if we take a road, for example, on the 2 images, the road is shifted at the intersection of the two SAOI).


Is someone already face this issue and get a solution?

Thank's a lot in advance !

Technical Evangelist
Posts: 945
Registered: ‎10-01-2015

Re: Orthorectification of several Satellite Images

[ Edited ]



Before trying anything else, please check the following:


  1. When you start Orthorectify without GCPs, do you select the DIMAP header as the input raster source? Selecting just the .tif or .jp2 will miss out on important header information provided in the DIMAP XML header.
  2.  In the Model List did you select Pleiades RPC?
  3. Confirm in the Model Properties dialog that the rpc*.xml has been picked up automatically (or manually select it)
  4. Have you provided an accurate Elevation Source for orthocorrection?
  5. Does the elevation source have a vertical datum defined in its header? If it does not then the software will assume that the vertical datum is the same as the sensor model being used. This may not be the case and can introduce positional errors. For example, the RPC may be using WGS84, whereas the elevation source may be defined in terms of heights above mean sea level. If the software is not aware of the difference in vertical datum positional errors will occur. Check using ImageInfo that the vertical datum is defined (along with vertical units) so that the software can automatically adjust for any datum differences.

If all those are correct, but you are still getting spatial inaccuracies, then it may well be that the source geometric model information provided with the data is not spatially accurate (to the degree you desire). In this case you will need to measure Ground Control to tie the model down more accurately. Frequently this can be done with just a single GCP, but the number required will vary depending on the nature of the data and its inherent inaccuracies (and the local terrain and relative accuracy of your elevation source). For example, if you have two scenes of the same area taken from different viewing angles and your DEM doesn't adequately model the terrain displacement effects, the orthorectified images are going to have relative positional errors. 


If correcting multiple images in an area you may also wish to consider using IMAGINE Photogrammetry to perform a block bundle adjustment to get results which overlap with a high degree of accuracy.


For multi-date imagery which overlaps in a "SAOI" you may also wish to try IMAGINE Autosync (which is part of the IMAGINE Expansion Pack product) if the above steps aren't providing the required level of accuracy.


Hope this helps,




Ian Anderson
Chief Product Owner, Desktop Remote Sensing
Hexagon Geospatial