A robot vision system consists of many components:

Image capture,
usually with a video camera, charge coupled device, laser or microwave radar, scanning sonar, etc., in which the image is converted from light or other radiation to an analog electrical signal;

Image digitization and signal processing,
in which the image is divided into a set of picture elements, or pixels, each of which is assigned a value representing the brightness, color, etc. of that part of the image;

Region, edge, and boundary detection,
in which the distinct visual elements of the image are separated;

Image scaling and alignment,
in which the digitized image is rotated, translated, and otherwise transformed to place it into some standard position and size; and

Image recognition,
in which the preprocessed image is submitted to a pattern recognition algorithm to allow the robot to categorize it, that is, recognize what the image represents.

Only the last of these problems will be considered in this paper.

Robert J. Brown 2003-05-14