Hand-written, on-line character recognition apparatus and method

• US 5,010,579 A
• Filed: 08/24/1989
• Issued: 04/23/1991
• Est. Priority Date: 08/30/1988
• Status: Expired due to Term

• Current Assignee: Sony Corporation (Sony Group Corp.)
• Original Assignee: Sony Corporation (Sony Group Corp.)
• Inventors: Yoshida, Kimiyoshi, Chatani, Masayuki, Tamori, Hirofumi, Sakano, Akio
• Primary Examiner(s): Moore, David K.
• Assistant Examiner(s): Santos, Daniel
• Application Number:
  US07/398,086
• Time in Patent Office: 607 Days
• Field of Search: 382/11, 382/13, 382/37
• US Class Current: 382/187
• CPC Class Codes: G06V 10/75 Organisation of the matching processes, e.g. simultaneous or sequential comparisons of image or video features; Coarse-fine approaches, e.g. multi-scale approaches; using context analysis; Selection of dictionaries

Abstract

A hand-written, on-line character recognition apparatus, and the method employed by it, in which the structure of a dictionary for recognition is formed as a sub-routine type, whereby the dictionary can be made small in size and a time necessary for recognition can be reduced.

Claims

1. A hand-written, on-line character recognition apparatus comprising:

(a) means for electronically receiving hand-written stroke pattern information data;

(b) dictionary means organized in a tree and branch structure for electronically storing character data describing stroke characteristics of characters, the beginning and ending of each branch being taken as different nodes with one or more branches sharing the same nodes;

(c) evaluation scan control means for scanning the tree and branches of the dictionary means to search for character data corresponding to the hand-written stroke pattern information data, wherein the evaluation scan control means compares the hand-written stroke pattern information data with the character data stored in the successive branches of the dictionary means, continuing on to the next successive branch as long as the degree of inconsistency does not exceed a predetermined limit and returning to the previous node to perform the same operation with respect to the character data stored in an alternative branch when the degree of inconsistency does exceed the predetermined limit; and

(d) save means for storing the conditions of the evaluation scan control means, wherein at least the conditions of the evaluation scanning existing at the beginning of each branch are saved in the save means upon branching in order to allow recursive use of the branches of the character data stored in the dictionary means when the evaluation scan control means has returned to the alternative branch after the degree of inconsistency exceeded the predetermined limit.

3. A hand-written, on-line character recognition apparatus as cited in claim 1, wherein the evaluation scan control means proceeds to scan a successive branch after saving the best evaluation result when scanning each branch during locally parallel branching.

2. A hand-written, on-line character recognition apparatus comprising:

(a) means for electronically receiving hand-written pattern information data;

(b) dictionary means organized in a tree and branch structure for electronically storing character data describing characteristics of characters, wherein the tree structure of the dictionary means includes at least citation branching, parallel branching and tree branching, the parallel branching being classified into parallel tree branching and locally parallel branching;

(c) evaluation scan control means for scanning the tree and branches of the dictionary means to search for character data corresponding to the hand-written pattern information data; and

(d) save means for storing the conditions of the evaluation scan control means, wherein at least the conditions of the scanning existing at the beginning of each branch are saved in the save means upon branching in order to allow recursive use of branches of the character data stored in the dictionary means.

7. A hand-written, on-line character recognition apparatus as cited in claim 2, wherein the evaluation scan control means proceeds to scan a successive branch after saving the best evaluation result when scanning each branch during locally parallel branching.

4. A hand-written, on-line character recognition method comprising the steps of:

(a) electronically receiving hand-written stroke pattern information data;

(b) electronically storing character data describing stroke characteristics of characters in a tree and branch organizational structure, the beginning and ending of each branch being taken as different nodes with one or more branches sharing the same nodes;

(c) electronically evaluation scanning the tree and branches of the electronically stored character data to search for character data corresponding to the hand-written pattern information data, wherein the evaluation scanning includes comparing the hand-written stroke pattern information data with the successive branches of the stored character data nd continuing on to the next successive branch as long as the degree of inconsistency does not exceed a predetermined limit and returning to the previous node to perform the same operation with respect to the character data stored in an alternative branch when the degree of inconsistency does exceed the predetermined limit; and

(d) storing the conditions of the evaluation scanning, wherein at least the conditions of the evaluation scanning existing at the beginning of each branch are stored upon branching in order to allow recursive use of the branches of the stored character data when the evaluation scanning process returns to the alternative branch after the degree of inconsistency exceeded the predetermined limit.

6. A hand-written, on-line character recognition method as cited in claim 4, wherein the evaluation scanning step further includes the steps of proceeding to scan a successive branch after saving the best evaluation result when scanning each branch during locally parallel branching.

5. A hand-written, on-line character recognition method comprising the steps of:

(a) electronically receiving hand-written pattern information data;

(b) electronically storing character data describing characteristics of characters in a tree and branch organizational structure, including the steps of storing the data by at least citation branching, parallel branching the tree branching, the parallel branching being classified into parallel tree branching and locally parallel branching;

(c) electronically evaluation scanning the tree and branches of the electronically stored character data to search for character data corresponding to the hand-written pattern information data; and

(d) storing the conditions of the evaluation scanning, wherein at least the conditions of the evaluation scanning existing at the beginning of each branch are stored upon branching in order to allow recursive use of branches of the stored character data.

8. A hand-written, on-line character recognition method as cited in claim 5, wherein the evaluation scanning step further includes the steps of proceeding to scan a successive branch after saving the best evaluation result when scanning each branch during locally parallel branching.