Line reconstruction using prior knowledge in single non-central view

Line projections in non-central systems contain more geometric information than central systems. The four degrees of freedom of the 3D line are mapped to the line-image and the 3D line can be theoretically recovered from 4 projecting rays (i.e. line-image points) from a single non-central view [3]. If the non-central system is properly calibrated we obtain a metric reconstruction of the 3D line. In practice, extraction of line-images is considerably more difficult and the resulting reconstruction is imprecise and sensitive to noise. In this paper we explore the reconstruction accuracy improvements when we impose geometrical constraints [1] exploiting prior knowledge. In particular, when the lines of the scene are arranged in two orthogonal directions and we know prior information about the direction of one of this directions (typically the vertical direction), the complexity of line fitting reduces, the accuracy of the metric reconstruction improves, and the extraction procedure is simplified.