Neural Networks for Vision and Image Processing: (A Bradford Book)

This interdisciplinary survey brings together recent models and experiments on how the brain sees and learns to recognize objects. It shows how to use these insights in technology and describes how neural networks provide a unifying computational framework for reaching these goals. Several chapters describe experiments in neurobiology and visual perception that clarify properties of biological vision and key conceptual issues that biological models need to address. Other chapters describe neural and computational models of biological vision that address such issues and clarify processes whereby biological vision derives its remarkable flexibility and power. Still other chapters use biologically derived models or heuristics to suggest neural network solutions to challenging technological problems in computer vision. Topics range from analyses of motion, depth, color and form to new concepts about learning, attention, pattern recognition, and hardware implementation.

Table of Contents:
Visual adaptation to a negative, brightness-reversed world - some preliminary observations, Stuart Anstis; prevailing lightness and hue and perceived texture segregation, Jacob Beck and William Goodwin; perception - a biological perspective, V.S. Ramachandran; the visual perception of 3-dimensional form, J. Farley Norman and James T. Todd; a new approach to shape from shading, Pierre Breton, et al; dynamic vision, Alex P. Pentland; figure-ground separation of connected scenic figures - boundaries, filling-in, and opponent processing, Stephen Grossberg and Lonce Wyse; toward a unified theory of spatiotemporal processing in the retina, Paolo Gaudiano; neurodynamics of real-time image velocity extraction, David A. Fay and Allen M. Waxman; why do parallel cortical systems exist for the perception of static form and moving form?, Stephen Grossberg; neural dynamics of visual motion perception - local detection and global grouping, Stephen Grossberg and Ennio Mingolla; neural circuits for visual attention in the primate brain, Robert Desimone; attentive supervised learning and recognition by an adaptive resonance system, Gail A. Carpenter, et al; synchronized oscillations for binding spatially distributed feature codes into coherent spatial patterns, Stephen Grossberg and David Somers; a quotient space hough transform for space variant visual attention, Alan S. Rojer and Eric L. Schwartz; optics and neural nets, David Casasent; neural networks for image analysis, Robert Hecht-Neilsen.

About the Author :
Gail A. Carpenter is Professor of Mathematics and Cognitive and Neural Systems and Director of the CNS Technology Lab at Boston University. Stephen Grossberg is Professor of Mathematics, Psychology, and Biomedical Engineering and Director of the Center for Adaptive Systems at Boston University.