Calculating Vanishing Points in Dual Space

Vanishing points can be used to exploit the parallel and orthogonal lines in 3D scenes thus the cameras' orientation parameters for vision processing. This paper proposed a vanishing point detection and estimation method in the dual image space. First, edge line segments are extracted. Second, based on the point-line duality theory, lines are transformed into points in the dual space where the transformed points belong to the same vanishing point form collinear clusters. Third, vanishing points are estimated by grouping and fitting straight lines across those clusters. The novel points of our method are: 1) automatically grouping the edge line segments that are the support of a vanishing point; 2) calculating the vanishing points by fitting straight lines in the dual space. Experiment results validated the proposed method.