chapter  8
The Computation of Image First principles: Motion as orientation in
ByMotion
Pages 56

The retinal image is alive with motion from a variety of sources. Whenever objects or living things around us move, their images move over our retinas. Even if the world is still, our eye, head, and body movements cause continual motion of the retinal images. In most circumstances, these effects will combine to produce a rich pattern of image motion. In Chapter 7 we saw how the changing view gained by a moving observer (“motion parallax”) provides information about the 3-D structure of objects and scenes. In Part III of this book we shall see how “optic flow”— the complex pattern of motion throughout the visual field as an animal or person moves aboutcan provide important information for the control of animals’ and people’s actions. To use Marr’s terms, introduced in Chapter 4 (pp. 80-81), all these considerations contribute to a computational theory of the information available in image motion, and the uses to which it can be put. But to understand how this information is obtained, we need to devise algorithms capable of extracting it from a time-varying retinal image and then to ask whether these algorithms are implemented in nervous systems.