The measurement of visual motion

Computer scientists designing machine vision systems, psychologists working in visual perception, visual neurophysiologists, and theoretical biologists will derive a deeper understanding of visual function - in particular the computations that the human visual system uses to analyze motion-from the important research reported in this book. The organization of movement in the changing image that reaches the eye provides our visual system with a valuable source of information for analyzing the structure of our surroundings. This book examines the measurement of this movement and the use of relative movement to locate the boundaries of physical objects in the environment. It investigates the nature of the computations that are necessary to perform this analysis by any vision system, biological or artificial. The author first defines the goals of these visual tasks, reveals the properties of the physical world that a vision system can rely upon to achieve such goals, and suggests general methods that can be used to carry out the tasks. From the general methods, she designs algorithms specifying a particular sequence of computations that a vision system can execute to perform these visual tasks. These algorithms are implemented on a computer system under a variety of circumstances. Combined with the traditional approaches of psychology and neurophysiology, this computational approach provides an exciting analysis of visual function, raising many new questions about the human vision system for further investigation. Ellen Catherine Hildreth received her doctorate from MIT. She is a Research Scientist in the MIT Artificial Intelligence Laboratory and associate director of the Center for Biological Information Processing at the Whitaker College of Health Sciences, Technology, and Management. The Measurement of Visual Motion is an ACM Distinguished Dissertation.