Abstract

An important problem for both biological and machine vision is the construction of scene representations from 2-D image data that are useful for recognition. One problem is that there can be more than one world out there giving rise to the image data at hand. Additional constraints regarding the nature of the environment have to be used to narrow the range of solutions. Although effort has gone into understanding these constraints, relatively little has been done to understand how neurallike learning networks may be used to solve scene-from-image problems. A paradigm is proposed in which stochastic models of scene properties are used to provide samples of image and scene representations. Distributed associative networks are taught, by example, the statistical constraints relating the image to the representation of the scene. This technique is applied to problems in optic flow, shape-from-shading, and stereo.

© 1987 Optical Society of America

Color and spatial structure in natural scenes

G. J. Burton and Ian R. Moorhead
Appl. Opt. 26(1) 157-170 (1987)

Optical implementations of associative networks with versatile adaptive learning capabilities

Arthur D. Fisher, Wendy L. Lippincott, and John N. Lee
Appl. Opt. 26(23) 5039-5054 (1987)

Associative network applications to low-level machine vision

J. Michael Oyster, Fernando Vicuna, and Walter Broadwell
Appl. Opt. 26(10) 1919-1926 (1987)

From image pair to a computer generated hologram for a real-world scene

Sihao Ding, Siyang Cao, Yuan F. Zheng, and Robert L. Ewing
Appl. Opt. 55(27) 7583-7592 (2016)

Method for estimating scene parameters from color histograms

Carol L. Novak and Steven A. Shafer
J. Opt. Soc. Am. A 11(11) 3020-3036 (1994)