Abstract
Object surface reconstruction from digital image data is a subproblem of Computer Vision. Presently, it has found a widespread interest in many technical and scientific areas, not only in photogrammetry. This paper reviews the recent developments of surface reconstruction from images in photogrammetry, where the correlation methods have evolved from step-by-step and patch-by-patch procedures to unified least-squares surface methods. By this way, digital image matching, DTM generation and orthophoto computation have been combined into one approach. On closer inspection the new theory shows up as rather complete, general and flexible, equally well suited for all kinds of multisensory and multispectral imagery. On the other hand, this photogrammetric contribution to Computer Stereo Vision may receive much benefit from the algorithmic research that has been developed to a very high level in Computer Vision. Still, these results have to be integrated into the basic approach to overcome the considerable computational burden, e.g. the estimation of the very large number of parameters such as convergence, surface discontinuities, occlusions, surface disturbances. Discussions on these topics and the list of references give an aid for future solutions.
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call