Computer vision: Signals, segments, and structures

In a computer vision system, we are given one or more digitized images of a scene, and our goal is to construct a description of the scene in terms of known types of objects that appear in it. Thus computer vision deals with digital images, but the techniques it uses are quite different from those used in digital image processing. In this article we describe some of these techniques, and indicate how they differ from methods based on signal-processing concepts.