Article View: la.seminars
Article #3COMPUTATIONAL VISION AND NEURAL NETWORKS
From: Gisele Pham
Date: Tue, 27 Oct 1992 22:22
Date: Tue, 27 Oct 1992 22:22
27 lines
1624 bytes
1624 bytes
DATE Thursday, October 29, 1992 TIME 4:30-6:00 P.M. PLACE 3400 Boelter Hall Cookies & Coffee Will Be Served Starting at 4:00 PM COMPUTATIONAL VISION AND NEURAL NETWORKS Josef Skrzypek Machine Perception Laboratory Department of Computer Science ABSTRACT One of the key problems in vision is segmentation of images into labeled patches that represent features, surfaces and objects in the scene. Analytical models for delineating and labeling surfaces as white, black, grey or color regardless of variation in intensity or hue of the illuminant remain a matter of controversy. We investigate these problems in the domain of computational neuroscience by testing the compatibility between physiological data and computational theories of lightness and color constancy. Since all foveal vision is based on invariant contrast information, reconstruction of original surface values is an important and common problem in many segmentation tasks. Time permitting we will present our neural net simulation results which address these problems as textural segmentation and boundary gap filling from illusory contours
Technical Details
Message-ID:
<1992Oct27.222241.16146@cs.ucla.edu>Path:
arc1-ca.pugleaf.net!sparky!uunet!convex!darwin.sura.net!sgiblab!sgigate!rutgers!ucla-cs!lanai.cs.ucla.edu!gphamShow Full Headers
Complete Article Headers
Path: arc1-ca.pugleaf.net!sparky!uunet!convex!darwin.sura.net!sgiblab!sgigate!rutgers!ucla-cs!lanai.cs.ucla.edu!gpham From: gpham@lanai.cs.ucla.edu (Gisele Pham) Newsgroups: la.seminars Subject: COMPUTATIONAL VISION AND NEURAL NETWORKS Keywords: CS201 Seminar Message-ID: <1992Oct27.222241.16146@cs.ucla.edu> Date: 27 Oct 92 22:22:41 GMT Sender: usenet@cs.ucla.edu (Mr Usenet) Distribution: la Organization: UCLA, Computer Science Department Lines: 27 Nntp-Posting-Host: lanai.cs.ucla.edu