期刊名称：ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences
印刷版ISSN：2194-9042
电子版ISSN：2194-9050
出版年度：2007
卷号：XXXVI-3/W49A
页码：69-74
出版社：Copernicus Publications
摘要：This paper addresses extensions of the classical least-squares matching approaches of (Wrobel, 1987, Ebner and Heipke, 1988) partic- ularly in the direction of full three-dimensional (3D) reconstruction. We use as unknowns the movement in the direction of the normals for a triangulation of the surface. To regularize the ill-posed inverse reconstruction problem, we smooth the surface by enforcing a low curvature in terms of that the vertices of the triangulation are close to the average plane of their direct neighbors. We employ a hierarchy of resolutions for the triangulation linked to adequate levels of image pyramids, to expand the range of convergence, and robust estimation, to deal with occlusions and non-Lambertian re.ection. First results using highly precise and reliable, but sparse points from the automatic orientation of images sequences as input for the triangulation show the potential of the approach
