Building facade modeling under line feature constraint based on close-range images

Abstract

To solve existing problems in modeling facade of building merely with point feature based on close-range images, a new method for modeling building facade under line feature constraint is proposed in this paper. Firstly, Camera parameters and sparse spatial point clouds data were restored using the SFM, and 3D dense point clouds were generated with MVS; Secondly, the line features were detected based on the gradient direction, those detected line features were fit considering directions and lengths, then line features were matched under multiple types of constraints and extracted from multi-image sequence. At last, final facade mesh of a building was triangulated with point cloud and line features. The experiment shows that this method can effectively reconstruct the geometric facade of buildings using the advantages of combining point and line features of the close - range image sequence, especially in restoring the contour information of the facade of buildings.