Image Primitives and Correspondence

Abstract

In previous chapters we have seen how geometric primitives, such as points and lines in space, can be transformed so that one can compute the coordinates of their “image,” i.e. their projection onto the image plane. In practice, however, images are arrays of positive numbers that measure the amount of light incident on a sensor at a particular location (see Sections 3.1 and 3.2, and Appendix 3.A). So, how do we reconcile a geometric image formation model (Section 3.3) with the fact that what we measure with a camera is not points and lines, but light intensity? In other words, how do we go from measurements of light (photometry) to geometry? This is the subject of this chapter: we will show how geometric primitives can be extracted from photometric measurements and matched across different views, so that the rest of the book can concentrate on geometry.

Everything should be made as simple as possible, but not simpler.

— Albert Einstein