Recognizing apparatus

• US 5,119,438 A
• Filed: 03/09/1990
• Issued: 06/02/1992
• Est. Priority Date: 03/13/1989
• Status: Expired due to Term

• Current Assignee: Sharp Kabushiki Kaisha (Hon Hai Precision Industry Co., Ltd.)
• Original Assignee: Sharp Kabushiki Kaisha (Hon Hai Precision Industry Co., Ltd.)
• Inventors: Ueda, Toru, Ishizuka, Yasushi, Togawa, Fumio
• Primary Examiner(s): Moore, David K.
• Assistant Examiner(s): Jung, Yon
• Application Number:
  US07/491,039
• Time in Patent Office: 816 Days
• Field of Search: 382/14, 382/15, 382/30, 382/36, 382/37, 382/38, 364/513, 364/916, 364/916.4
• US Class Current: 382/158
• CPC Class Codes: G06F 18/285 Selection of pattern recognition techniques, e.g. of classifiers in a multi-classifier system
• CPC Class Codes: G06V 30/242 Division of the character sequences into groups prior to recognition; Selection of dictionaries
• CPC Class Codes: G06V 30/2504 Coarse or fine approaches, e.g. resolution of ambiguities or multiscale approaches

Abstract

A recognizing apparatus is provided for recognizing a class to which an inputted characteristic pattern belongs from among a plurality of classes to be discriminated using a neural network. The classes are classified into a plurality of categories. The apparatus includes a network selecting portion for selecting a category to which the inputted characteristic pattern belongs and for selecting a neural network for use in discriminating the class to which the inputted characteristic pattern belongs in the selected category. The apparatus further includes a network memory portion, a network setting portion and a details discriminating portion. The network memory portion stores structures of a plurality of neural networks which have finished learning for respective categories, weights of the neural networks set by the learning and a plurality of discriminating algorithms to be used when the classes are discriminated by the neural networks. The network setting portion sets the structure and weights of a neural network selected by the network selecting portion and a discriminating alogrithm appropriate to the selected category. The details discriminating portion recognizes the class to which the inputted characteristic pattern belongs by performing the details discriminating operation using the neural network set by the neural network setting portion.

Claims

1. A recognizing apparatus for recognizing a class to which an inputted characteristic pattern belongs from among a plurality of classes to be discriminated by using one of a plurality of neural networks with said classes being classified into a plurality of categories, said recognizing apparatus comprising:

network selecting means for selecting one of said plurality of categories to which the inputted characteristic pattern belongs by comparing the inputted characteristic pattern with a plurality of reference patterns and for selecting one of said plurality of neural networks for use in discriminating one of the classes to which the inputted characteristic pattern belongs in said one category;

network memory means for storing therein structures of said plurality of neural networks which have finished learning for respective categories, weights of said neural networks set by the learning and a plurality of discriminating algorithms to be used when the classes are discriminated by said neural networks;

network setting means for setting said structure and said weights of said one neural network selected by said network selecting means and one of said discriminating algorithms corresponding to said one category to create an optimized neural network; and

details discriminating means for recognizing said one class to which the inputted characteristic pattern belongs by performing a details discriminating operation by using said optimized neural network set by said neural network setting means.

2. The recognizing apparatus according to claim 1, further comprising reference pattern memory means for storing therein said plurality of reference patterns corresponding to said plurality of categories and similarity calculating means for calculating the degree of similarity between the inputted characteristic pattern and the reference patterns.

3. The recognizing apparatus according to claim 1, wherein said plurality of neural networks comprise a Kohonen type neural network and a Perceptron type neural network.

4. The recognizing apparatus according to claim 1, wherein said structure, said weights and said discriminating algorithms of said one neural network are learned by one of a plurality of learning algorithms.

5. The recognizing apparatus according to claim 4 wherein said plurality of learning algorithms comprise Backpropagation and LVQ2 algorithms.

6. A method for recognizing a class to which an inputted characteristic pattern belongs from among a plurality of classes to be discriminated by using one of a plurality of neural networks with said classes being classified into a plurality of categories, comprising the steps of:

(a) comparing the inputted characteristic pattern with a plurality of reference patterns;

(b) selecting one of said plurality of categories and one of said plurality of neural networks for use in discriminating one of the classes to which the inputted characteristic pattern belongs in said one category;

(c) storing structures of said plurality of neural networks which have finished learning for respective categories, weights of said neural networks set by the learning and a plurality of discriminating algorithms to be used when the classes are discriminated by said neural networks;

(d) setting said structure and said weights of said one neural network selected at said step (b) and one of said discriminating algorithms appropriate to said one category to create an optimized neural network; and

(e) recognizing said one class to which the inputted characteristic pattern belongs by performing a details discriminating operation by using said optimized neural network set at said step (d).