Using Oblique Satellite Imagery

[sarko]

I have successfully used Photoscan to work with a number of aerial images and I am trying to move to satellite imagery.  I have a set of oblique satellite images that I would like to use to generate a DEM.  I have one image that is collected at nadir, one at -35 degrees and one at +35 degrees oblique.  They were collected in the same swath, so are nearly simultaneous.  

I have tried to run them through photoscan, but they won't even align.  Am I just missing something fundamental and these images will not work?  


Thanks,
Scott

[gEEvEE]

Hi Scott,

PhotoScan works best with images that are taken from different positions. As the difference in camera pose for your satellite imagery will be very alike, this is most likely the cause of the alignment problem.
Which satellite images do you use? Some spaceborne imagery does not have the geometric properties of a frame camera (e.g. Corona has a well-known S-shaped deformation of the ground that is imaged). This image-based difference in geometry might also cause PhotoScan to fail.

Regards,
Geert

[GPC]

I think the reason the oblique imagery is failing is that it isn't captured with a conventional camera system with a frame sensor.

Oblique imagery in those 3-camera systems is typically acquired with a linear ccd sensor in long strips several kilometers long. The image data is then post processed using IMU and GPS data and then cropped to make an attractive, measurable image.

Depending on who acquired the photos, it could be that what you're loading is actually several thousand individual images (all 1pixel tall) merged into one photo.

[sarko]

Thanks cs.

I was wondering if that was the problem, but couldn't come up with a good reason why it wouldn't work.  Can you explain a bit more why a pushbroom type sensor will not work?  I was hoping the oblique nature of it would give enough shift for PhotoScan to align to.  

Thanks,
Scott

[GPC]

I think it fails for a couple reasons.

First, the image you see is really a combination of several thousand, 1 pixel tall, photos. Those very narrow photos don't contain enough data for the program to compute the distortion parameters and there's no overlap.

Second, during post processing, every single one of those 1pixel tall photos is scaled and translated using the GPS/IMU data to smooth out the plane's movement. RAW imagery out of an airborn linear CCD isn't very pleasing to look at.. too many undulations to be useful without a lot of post processing. That post processing changes every image and makes it worthless for photomodeler reconstructions.

If you took a set of photos that you take with a normal digital camera and open them all in photoshop and scaled just all a slightly different amount.. they wouldn't work in photo modeler anymore either. It's the same thing here.