Proceedings Article | 21 June 2011
Proc. SPIE. 8085, Videometrics, Range Imaging, and Applications XI
KEYWORDS: 3D image reconstruction, Cameras, Calibration, Remote sensing, 3D modeling, Image analysis, Photogrammetry, Computer vision technology, Machine vision, 3D image processing
Automatic image orientation of close-range image blocks is becoming a task of increasing importance in the practice of
photogrammetry. Although image orientation procedures based on interactive tie point measurements do not require any
preferential block structure, the use of structured sequences can help to accomplish this task in an automated way.
Automatic orientation of image sequences has been widely investigated in the Computer Vision community. Here the
method is generally named "Structure from Motion" (SfM), or "Structure and Motion". These refer to the simultaneous
estimation of the image orientation parameters and 3D object points of a scene from a set of image correspondences.
Such approaches, that generally disregard camera calibration data, do not ensure an accurate 3D reconstruction, which is
a requirement for photogrammetric projects. The major contribution of SfM is therefore viewed in the photogrammetric
community as a powerful tool to automatically provide a dense set of tie points as well as initial parameters for a final
rigorous bundle adjustment. The paper, after a brief overview of automatic procedures for close-range image sequence
orientation, will show some characteristic examples. Although powerful and reliable image orientation solutions are
nowadays available at research level, there are certain questions that are still open. Thus the paper will also report some
open issues, like the geometric characteristics of the sequences, scene's texture and shape, ground constraints (control
points and/or free-network adjustment), feature matching techniques, outlier rejection and bundle adjustment models.