Method and apparatus for on-line handwritten input character recognition and recording medium for executing the method

• US 6,144,764 A
• Filed: 01/20/1998
• Issued: 11/07/2000
• Est. Priority Date: 07/02/1997
• Status: Expired due to Fees

• Current Assignee: Mitsui High-Tec Incorporated
• Original Assignee: Mitsui High-Tec Incorporated
• Inventors: Yamakawa, Takeshi, Odahara, Toyoshi, Inagawa, Naohiro
• Primary Examiner(s): Au, Amelia
• Assistant Examiner(s): MILLER, MARTIN E
• Application Number:
  US09/009,533
• Time in Patent Office: 1,022 Days
• Field of Search: 382/187, 382/197, 382/198
• US Class Current: 382/187
• CPC Class Codes: G06V 30/347 Sampling; Contour coding; Stroke extraction
• CPC Class Codes: G06V 30/373 using a special pattern or subpattern alphabet

Abstract

The invention aims at recognition of a handwritten input character on-line with quite high accuracy. The method of on-line handwritten input character recognition of the invention is characterized in that ordinary strokes and transition strokes and/or start-end (s-e) strokes of a handwritten input character sampled on-line are compared with ordinary strokes and transition strokes and/or s-e strokes of dictionary'"'"'s characters previously registered in a dictionary and the character corresponding to the handwritten input character is recognized on the basis of the result of the comparison. Otherwise, when the dictionary'"'"'s character most similar to the input character corresponds to a preset character, the handwritten input character is identified by using characteristic features of the corresponding character. Thus, by comparing ordinary strokes and transition strokes and/or s-e strokes of a handwritten input character sampled on-line with ordinary strokes and transition strokes and/or s-e strokes of dictionary'"'"'s characters previously registered in a dictionary and recognizing the character corresponding to the handwritten input character on the basis of the result of the comparison, a handwritten input character which was difficult to recognize only by means of ordinary strokes can be recognized with high accuracy.

Claims

1. A method of on-line input handwritten character recognition for recognizing individual handwritten characters each formed by ordinary strokes sampled on-line, wherein the ordinary strokes of a handwritten character are strokes defined by continuous pen movement on a writing surface from a position whereat the pen is placed on the writing surface to a position whereat the pen is lifted from the writing surface, comprising the steps of:

converting each of said ordinary strokes of a handwritten character into component vectors having absolute angles representing directional information;

determining and converting transition strokes of the handwritten character, defined by pen movement, off a writing surface, from an end of an ordinary stroke to a beginning of a next ordinary stroke of the handwritten character, into component vectors having absolute angles representing directional information;

determining and converting start-end strokes of the handwritten character, defined by pen movement from a starting point of a first ordinary stroke to starting points and ending points of each of subsequent ordinary strokes of the handwritten character, into component vectors having absolute angles representing directional information;

dividing said absolute angles into a group of fuzzy vectors for the handwritten character indicative of reference directions;

converting said fuzzy vectors into membership grades of a plurality of directions representing similarity measures respectively corresponding to said reference directions by using a predetermined membership function;

calculating one of a similarity measure and a dissimilarity measure between the group of fuzzy vectors described with said converted membership grades and character data from dictionary data described with a fuzzy vector group respectively corresponding to component vectors of ordinary strokes and at least one of transition strokes and starting-end strokes previously registered in the dictionary; and

choosing a character from the dictionary data corresponding to said handwritten character based on one of the similarity measure being the greatest and the dissimilarity measure being the smallest.

2. The method according to claim 1, further comprising the steps of:

determining whether the chosen character is one of a predetermined set of characters; and

applying a proportionality rule to determine which of a group of characters, associated with said chosen character, corresponds to the input character if the chosen character is one said predetermined set.

3. The method according to one of claims 1 and 2, further comprising the steps of:

determining that said one of said similarity measure and said dissimilarity measure of said chosen character does not conform to a threshold indicating that there is no character data in the dictionary data corresponding to the handwritten character; and

inputting a visually determined correct character and registering the group of fuzzy vectors described with said converted membership grades of said handwritten character as dictionary data corresponding to said correct character.

4. A handwritten character recognition apparatus for recognizing individual handwritten characters each formed by ordinary strokes sampled on-line, wherein the ordinary strokes of a handwritten character are strokes defined by continuous pen movement on a writing surface from a position whereat the pen is placed on the writing surface to a position whereat the pen is lifted from the writing surface, comprising:

a handwritten character input device for inputting handwritten characters and sampling a handwritten character to produce sample data;

means for converting said sample data of each of said ordinary strokes of a handwritten character into component vectors having absolute angles representing directional information;

means for determining and converting transition strokes of the handwritten character, defined by pen movement, off a writing surface, from an end of an ordinary stroke to a beginning of a next ordinary stroke of the handwritten character, into component vectors having absolute angles representing directional information;

means for determining and converting start-end strokes of the handwritten character, defined by pen movement from a starting point of a first ordinary stroke to starting points and ending points of each of subsequent ordinary strokes of the handwritten character, into component vectors having absolute angles representing directional information;

means for dividing said absolute angles into a group of fuzzy vectors for the handwritten character indicative of reference directions;

means for converting said fuzzy vectors into membership grades of a plurality of directions representing similarity measures respectively corresponding to said reference directions by using a predetermined membership function;

means for calculating one of a similarity measure and a dissimilarity measure between the group of fuzzy vectors described with said converted membership grades and character data from dictionary data described with a fuzzy vector group respectively corresponding to component vectors of ordinary strokes and at least one of transition strokes and starting-end strokes previously registered in the dictionary; and

means for choosing a character from the dictionary data corresponding to said handwritten character based on one of the similarity measure being the greatest and the dissimilarity measure being the smallest.

5. A recording medium readable by a computer having stored thereon executable code for enabling the computer to recognize individual handwritten characters each formed by ordinary strokes sampled, wherein the ordinary strokes of a handwritten character are strokes defined by continuous pen movement on a writing surface from a position whereat the pen is placed on the writing surface to a position whereat the pen is lifted from the writing surface, the code comprising:

procedures for converting each of said ordinary strokes of a handwritten character into component vectors having absolute angles representing directional information;

procedures for determining and converting transition strokes of the handwritten character, defined by pen movement, off a writing surface, from an end of an ordinary stroke to a beginning of a next ordinary stroke of the handwritten character, into component vectors having absolute angles representing directional information;

procedures for determining and converting start-end strokes of the handwritten character, defined by pen movement from a starting point of a first ordinary stroke to starting points and ending points of each of subsequent ordinary strokes of the handwritten character, into component vectors having absolute angles representing directional information;

procedures for dividing said absolute angles into a group of fuzzy vectors for the handwritten character indicative of reference directions;

procedures for converting said fuzzy vectors into membership grades of a plurality of directions representing similarity measures respectively corresponding to said reference directions by using a predetermined membership function;

procedures for calculating one of a similarity measure and a dissimilarity measure between the group of fuzzy vectors described with said converted membership grades and character data from dictionary data described with a fuzzy vector group respectively corresponding to component vectors of ordinary strokes and at least one of transition strokes and starting-end strokes previously registered in the dictionary; and

procedures for choosing a character from the dictionary data corresponding to said handwritten character based on one of the similarity measure being the greatest and the dissimilarity measure being the smallest.

6. A method of on-line input handwritten character recognition for recognizing individual handwritten characters each represented by a character line data set comprising a series of coordinates sampled on-line, wherein the character line data set represents one of one ordinary stroke sampled and a plurality of ordinary strokes sampled, the ordinary stroke of a handwritten character being a stroke defined by continuous pen movement on a writing surface from a position whereat the pen is placed on the writing surface to a position whereat the pen is lifted from the writing surface, the method comprising the steps of:

dividing a character line data set of a handwritten character into dot coordinate data defining a predetermined number of component vectors between adjacent ones of dot coordinates;

determining and converting at least one of;

transition strokes of the handwritten character, defined by pen movement, off a writing surface, from an end of an ordinary stroke to a beginning of a next ordinary stroke of the handwritten character, into component vectors having absolute angles representing directional information when the character line data set represents a plurality of ordinary strokes sampled; and

start-end strokes of the handwritten character, defined by pen movement from a starting point of a first ordinary stroke to starting points and ending points of each of subsequent ordinary strokes of the handwritten character, into component vectors having absolute angles representing directional information when the character line data set represents a plurality of ordinary strokes sampled;

calculating absolute angles of said component vectors;

applying fuzzy logic to divide said absolute angles into fuzzy vectors indicative of reference directions;

converting said fuzzy vectors into membership grades in a plurality of directions representing similarity measures respectively corresponding to said reference directions by using a predetermined membership function;

calculating one of a similarity measure and a dissimilarity measure between the fuzzy vectors of the handwritten character data represented by said converted membership grades and character data from dictionary data representing fuzzy vectors respectively corresponding to component vectors of a character previously registered in the dictionary data; and

extracting a character from the dictionary data corresponding to said handwritten character based on one of the similarity measure being the greatest and the dissimilarity measure being the smallest.

7. The method according to claim 6, wherein said predetermined number is thirty two.

8. The method according to claim 6, further comprising the steps of:

determining whether the extracted character is one of a predetermined set of characters; and

applying a proportionality rule to determine which of a group of characters, associated with said extracted character, corresponds to the input character if the chosen character is one said predetermined set.

9. The method according to claim 6, further comprising the steps of:

determining that said one of said similarity measure and said dissimilarity measure of said extracted character does not conform to a threshold indicating that there is no character data in the dictionary data corresponding to the handwritten character; and

inputting a visually determined correct character and registering the group of fuzzy vectors described with said converted membership grades of said handwritten character as dictionary data corresponding to said correct character.

10. A handwritten character recognition apparatus for recognizing individual handwritten characters each represented by a character line data set comprising a series of coordinates sampled on-line, wherein the character line data set represents one of one ordinary stroke sampled and a plurality of ordinary strokes sampled, the ordinary stroke of a handwritten character being a stroke defined by continuous pen movement on a writing surface from a position whereat the pen is placed on the writing surface to a position whereat the pen is lifted from the writing surface, the apparatus comprising:

a handwritten character input device for inputting handwritten characters and sampling a handwritten character to produce a character line data set;

means for dividing a character line data set of the handwritten character into dot coordinate data defining a predetermined number of component vectors between adjacent ones of dot coordinates;

means for determining and converting at least one of;

transition strokes of the handwritten character, defined by pen movement, off a writing surface, from an end of an ordinary stroke to a beginning of a next ordinary stroke of the handwritten character, into component vectors having absolute angles representing directional information when the character line data set represents a plurality of ordinary strokes sampled; and

start-end strokes of the handwritten character, defined by pen movement from a starting point of a first ordinary stroke to starting points and ending points of each of subsequent ordinary strokes of the handwritten character, into component vectors having absolute angles representing directional information when the character line data set represents a plurality of ordinary strokes sampled;

means for calculating absolute angles of said component vectors;

means for applying fuzzy logic to divide said absolute angles into fuzzy vectors indicative of reference directions;

means for converting said fuzzy vectors into membership grades in a plurality of directions representing similarity measures respectively corresponding to said reference directions by using a predetermined membership function;

means for calculating one of a similarity measure and a dissimilarity measure between the fuzzy vectors of the handwritten character data represented by said converted membership grades and character data from dictionary data representing fuzzy vectors respectively corresponding to component vectors of a character previously registered in the dictionary data; and

means for extracting a character from the dictionary data corresponding to said handwritten character based on one of the similarity measure being the greatest and the dissimilarity measure being the smallest.

11. A recording medium readable by a computer having stored thereon executable code for enabling the computer to recognize individual handwritten characters each represented by a character line data set comprising a series of coordinates sampled on-line, wherein the character line data set represents one of one ordinary stroke sampled and a plurality of ordinary strokes sampled, the ordinary stroke of a handwritten character being a stroke defined by continuous pen movement on a writing surface from a position whereat the pen is placed on the writing surface to a position whereat the pen is lifted from the writing surface, the code comprising:

procedures for dividing a character line data set of a handwritten character into dot coordinate data defining a predetermined number of component vectors between adjacent ones of dot coordinates;

procedures for determining and converting at least one of;

transition strokes of the handwritten character, defined by pen movement, off a writing surface, from an end of an ordinary stroke to a beginning of a next ordinary stroke of the handwritten character, into component vectors having absolute angles representing directional information when the character line data set represents a plurality of ordinary strokes sampled; and

start-end strokes of the handwritten character, defined by pen movement from a starting point of a first ordinary stroke to starting points and ending points of each of subsequent ordinary strokes of the handwritten character, into component vectors having absolute angles representing directional information when the character line data set represents a plurality of ordinary strokes sampled;

procedures for calculating absolute angles of said component vectors;

procedures for applying fuzzy logic to divide said absolute angles into fuzzy vectors indicative of reference directions;

procedures for converting said fuzzy vectors into membership grades in a plurality of directions representing similarity measures respectively corresponding to said reference directions by using a predetermined membership function;

procedures for calculating one of a similarity measure and a dissimilarity measure between the fuzzy vectors of the handwritten character data represented by said converted membership grades and character data from dictionary data representing fuzzy vectors respectively corresponding to component vectors of a character previously registered in the dictionary data; and

procedures for extracting a character from the dictionary data corresponding to said handwritten character based on one of the similarity measure being the greatest and the dissimilarity measure being the smallest.