How does the visual system compute the global motion of an object from local views ofits contours? Although this important problem in computational vision (also called the apertureproblem) is key to understanding how biological systems work, there has been surprisingly littleneurobiologically plausible work done on it. This book describes a neurally based model, implementedas a connectionist network, of how the aperture problem is solved. It provides a structural accountof the model's performance on a number of tasks and demonstrates that the details of implementationinfluence the nature of the computation as well as predict perceptual effects that are unique to themodel. The basic approach described can be extended to a number of different sensorycomputations.Sereno first reviews current research and theories about motion detection. She thenconsiders the formal aspects of the aperture problem and describes a model of pattern motionperception that stands out in several respects. The model takes into account the structure of thevisual system and attempts to build on known neurophysiological structures that might be availablefor solving the aperture problem, comparing performances in tasks involving direction and speedacuity, transparency, and motion coherency to human performance. The model's emphasis on the detailsof implementation rather-than on the goals of computation show that the details of datarepresentation change the nature of the computation, producing predictions (including severalillusions) that are unique and that can be confirmed through psychophysical experiments.MargaretEuphrasia Sereno is Assistant Professor of Psychology at the University of Oregon.
This book describes a neurally based model, implemented as a connectionist network,of how the aperture problem is solved.
