Triangulation-Based Approaches to Three-Dimensional Scene Reconstruction - 3D Computer Vision: Efficient Methods and Applications - page 28

Fig. 1.4 Summary of projective techniques for scene reconstruction and self-calibration

Notably, all metric self-calibration methods yield the three-dimensional scene
structure only in arbitrary units. Consequently, the absolute scale of the reconstruc-
tion remains unknown as long as the camera positions are unknown or unless scene
knowledge is available (Wöhler et al., 2009 ), since an increase of the mutual dis-
tances between the scene points by a constant factor can be compensated by in-
creasing the mutual distances between the cameras and their distances to the scene
accordingly (cf. Wöhler et al., 2009 and Sect. 5.1 , where the issue of scene recon-
struction at absolute scale is examined further).

1.4.6.3 Self-calibration Based on Vanishing Points

The self-calibration methods described so far require a large number of point cor-
respondences between the acquired images. In certain scenes, especially those con-
taining human-made objects such as buildings, it may be more convenient to extract
lines instead of well-defined points from the acquired images.