Login| Sign Up| Help| Contact|

Patent Searching and Data


Title:
GRAPHIC SHAPE LEARNING RECOGNITION SYSTEM
Document Type and Number:
Japanese Patent JPH05197812
Kind Code:
A
Abstract:

PURPOSE: To provide a graphic shape learning recognition system which can automatically recognize various types of graphic shapes without programming each of these shapes.

CONSTITUTION: The curvature information is extracted and then inputted to a neural network 21 as the input information by means of a neural network technique and with notice paid to a fact that the feature of the shape of a graphic is accurately shown by the curvature information on the graphic like a case where the shape of a given graphic is learnt frequently and the graphic can be recognized and comprehended by an infant. Thus the network 21 learns the graphic and the relevant graphic is recognized by the result of learning.


Inventors:
NAGURA MASAKAZU
ABE SHINJI
KATAGIRI MASAJI
Application Number:
JP774892A
Publication Date:
August 06, 1993
Filing Date:
January 20, 1992
Export Citation:
Click for automatic bibliography generation   Help
Assignee:
NIPPON TELEGRAPH & TELEPHONE
International Classes:
G06F15/18; G06G7/60; G06N3/00; G06N99/00; G06T7/00; (IPC1-7): G06F15/18; G06F15/70; G06G7/60
Attorney, Agent or Firm:
Tadahiko Ito



 
Previous Patent: JPS5197811

Next Patent: FEATURE SELECTING METHOD