[Back]

M. Hornáček:
"Extracting Vanishing Points across Multiple Views";
Supervisor: W. Purgathofer, R.F. Tobler; Institut für Computergraphik und Algorithmen, 2010; final examination: 2010-09-28.

The realization that we see lines we know to be parallel in space as lines that appear to converge in a corresponding vanishing point has led to techniques employed by artists to render a credible impression of perspective. More recently, it has also led to techniques for recovering information embedded in images-or, indeed, in paintings that feature correct perspective-concerning the geometry of their underlying scene. In this thesis, we explore the extraction of vanishing points in the aim of facilitating the reconstruction of urban scenes. In departure from most vanishing point extraction methods, ours extracts a constellation of vanishing points across multiple views rather than in a single image alone. By making use of a strong orthogonality criterion per view, optimal segment intersection estimation and a novel tripod fitting technique, our approach allows for the extraction of results that correspond closely to the dominant three pairwise-orthogonal orientations of a typical urban scene. Accordingly, ours can fairly be described as a material refinement of the approach proposed in Sinha et al.

http://publik.tuwien.ac.at/files/PubDat_189073.pdf