I am very new in GIS and I am trying to ortho-rectificate my images. But after putting it on the map I see that results are not precise. We have images taken from the airplane, it's longitude, latitude, elevation, omega, phi, kappa. What is the best way to calculate rectification for our images? Is there any third party libraries for that? Or at least some tool to compare it with my results?

Thanks in advance! Serhiy.

UPDATE: what I am trying to do is to find how images are projected to the ground by using latitude, lontitude, elevation and angles (omega, phi kappa). Calculate 4 corners of that image and then georeference each image. After that I think I can make precise mosaic.

UPDATE 2: I spent almost whole day on GRASS but without any luck, it is not so simple and not intuitive.
I read that Direct Georeferencing is what I need. I found a lot of mathematical articles on internet and tried to write my own implementation, but it is not very precise. Is there any simple tool or library that can take coordinatates (x,y,z), angles(omega, phi, kappa), elevation, image height/lenght (in nadir position) or focal lense and calculate 4 coordinates that correspond to corners of the image on the ground?

  • GRASS is part of QGIS use the i.ortho.photo tool gis.stackexchange.com/questions/9334/…
    – Mapperz
    Dec 18 '13 at 16:34
  • Are you trying to eventually create a mosaic? Or do you want the images all separate?
    – Baltok
    Dec 18 '13 at 16:43
  • Thanks for your answers, Going to look to i.ortho.photo. First of all I want to rectificate image to know exactly how the image is projected to the ground and calculate 4 points (corners of the image). Then I am going to georeference the images using qgis georeferencer and make mosaic. Actually from very begining I was looking to OpenCV stitching, but at some point I realized that I am loosing all geo information after the stitching process. So currently looking for qgis and gdal. Looking to hear any suggestions. Dec 18 '13 at 19:18

If you want to do a "true" ortho-rectification, you need a DEM. the DEM could come from an ancillary source or from photogrammetry using a second photo that overlaps yours. If you don't have a precise DEM, geometric corrections with Ground Control Points becomes an alternative to the orthorectiication (the information about the plane position and viewing angles is then useless).

  • I don't think I need DEM on that stage. Currently I need something simpler, so I am supposing that ground is 100% flat:) Also I thought that all that geometric corrections are called ortho-rectification in general. Dec 18 '13 at 19:39
  • it depends on the accuracy that you expect. An image from a plane is like a central projection. If you can assume that you are on a flat surface, you need at least the average elevation (obviously, if you take a picture, the scale will change depending on the distance to the feature of interest). Now, if you are not using DEM, your error will change depending on the location on the image. In the center, it will be close to zero. then the apparent displacement increases with the sinus of the viewing angle. e.g: the top of a building of 10 m at an angle of 30° => error = 5 m
    – radouxju
    Dec 19 '13 at 6:53
  • "true" ortho-rectification achieves the best accuracy amongst the geometric correction methods. But you can warp your image with GCP's and no DEM if you don't need to be accurate and your relief is not too big.
    – radouxju
    Dec 19 '13 at 7:00

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.