Premium
Automatic Keyline Recognition and 3D Reconstruction For Quasi‐Planar Façades in Close‐range Images
Author(s) -
Li Chang,
Zhang Yongjun,
Zhang Zuxun
Publication year - 2016
Publication title -
the photogrammetric record
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.638
H-Index - 51
eISSN - 1477-9730
pISSN - 0031-868X
DOI - 10.1111/phor.12141
Subject(s) - computer science , artificial intelligence , plane (geometry) , line segment , intersection (aeronautics) , computer vision , regular polygon , planar , matching (statistics) , a priori and a posteriori , line (geometry) , bundle adjustment , photogrammetry , bundle , algorithm , block (permutation group theory) , mathematics , geometry , computer graphics (images) , geography , philosophy , statistics , cartography , epistemology , materials science , composite material
Critical keylines, such as concave and convex edges of a building façade, can be lost in photogrammetric recognition procedures. To solve this problem and to reconstruct quasi‐planar 3D façades automatically and precisely, a set of algorithms and techniques for the automatic recognition of lines and 3D reconstruction is proposed. This includes: (1) a procedure for line‐segment matching that satisfies the spatial requirements of a 3D scene based on “global independence” and “local dependence”; (2) a technique of generalised point bundle block adjustment combined with spatial line constraints (in the form of virtual observations) to control the propagation of error; and (3) the methods of perceptual organisation, plane fitting and plane–plane intersection are suggested to acquire the critical keylines corresponding to concave and convex building edges. Experimental results show that these new algorithms are feasible and applicable to recognition and 3D reconstruction. Recommendations for recognition methods are provided depending on whether or not a priori topological relationships are available between the planes under consideration.