View Full Document

A Neural Network Architecture for Preattentive



View the full content.
View Full Document
View Full Document

12 views

Unformatted text preview:

Ittt TKANSACTIOKS O N BIO IFDICAL h N G I N F t K I N G VOL 36 N O I JANLJAKY I Y X Y 65 A Neural Network Architecture for Preattentive Vision STEPHEN GROSSBERG E N N I O MINGOLLA A N D DEJAN TODOROVIC Invited Paper Ahsfrucf Recent results towards de elopment of a neural network architecture for general purpose preattentive vision are summarized The architecture contains two parallel subsystems the boundarj cuntour system BCS arid the feature contour system FCS which interact together to generate a representation of form and color and depth Eiiiergent houndar segmentation w ithin the BCS and featural fillingin i t h i nthe FCS are herein eniphasized within a monocular setting Applications t o the analysis of houndaries textures and smooth surfaces are dewrihed as is a model for invariant brightness perception under variahle illumination condition5 The theory shows hoa suitably defined parallel and hierarchical interactions overcome computational uncertainties that iiecessarily exist at early processing stages Some of the psychophysical and neiiroph siological data upporting the theory s predictions are nientioned THEN E E n FOR A GENERAL PURPOSE PREATTENTIVE V I S I O NM A C H I N E M ANY AI algorithms for machine vision have been too specialized for applications to real world problems Such algorithms are often designed to deal with one type of information for example boundary disparity curvature shading or spatial frequency information Moreover such algorithms typically use different coniputational schemes to analyze each distinct type of information so that unification into a single general purpose vision algorithm is difficult at best For such AI algorithms other types of signals are often contaminants or noise elements rather than cooperative sources of ambigu it y red uc i ng info rnia t i on U n fo rt u na t e 1y most rea I ist i c scenes contain partial information of several different types in each part of a scene In contrast when we humans gaze upon a scene



Access the best Study Guides, Lecture Notes and Practice Exams

Loading Unlocking...
Login

Join to view A Neural Network Architecture for Preattentive and access 3M+ class-specific study document.

or
We will never post anything without your permission.
Don't have an account?
Sign Up

Join to view A Neural Network Architecture for Preattentive and access 3M+ class-specific study document.

or

By creating an account you agree to our Privacy Policy and Terms Of Use

Already a member?