Orientation quality & Aerotriangulation

The aerotriangulation process results in image orientation data which consists of camera position, rotation, and interior camera parameters (focal length, principal point, distortion parameters).

Together with the images, image orientations serve as input when creating a SURE project. Their accuracy reflects strongly on the quality of the results. Ensuring high-quality orientations leads to higher geometric accuracy and sharper edges.

Below you will find some best practices regarding the aerotriangulation process:

General:

  • Ensure a parallax below one pixel for all images (nadir, oblique, etc).

  • Do not use data retrieved from GPS/INS alone. Use Aerotriangulation, Structure from Motion, or Bundle Adjustment software instead to refine the navigation data.

Tie Points:

  • Make sure that tie points are matched in every image to ensure sub-pixel parallaxes.

  • Use a stereo viewer for assessing the quality of the orientations. The y-parallax / vertical shift perpendicular to epipolar direction should ideally not exceed more than one pixel in size.

  • Avoid using only the relative orientation of the camera rig (tie points only in nadir images, etc.). Tie points should also be matched in oblique images.

  • Ensure that images have enough tie points between each other. Also images of different viewing directions. Low standard deviation values are a good indicator of orientation precision only when they are based on a high number of tie points.

  • Use automatic tie/feature points for improving the relative accuracy of the orientations.

Ground Control Points (GCPs)

  • Use GCPs for improving the absolute accuracy of the orientations.

  • Use a higher number of GCPs when also estimating interior parameters of the cameras (mostly the case for close range- and UAV- mapping).

  • Use GCPs not considered during the adjustment as independent check points for quality assurance purposes.

  • Ensure that GCPs and check points are well distributed across the entire project area

  • Ensure that the accuracy of GCPs is at least 4x higher than the desired accuracy for the whole block.

  • If in doubt, contact your software vendor and ask how to achieve the recommended settings.

Orientation parameters

The quality of the generated results directly depends on the quality of the orientation parameters. For getting the best quality results from your images, it is important to have accurate orientation data.