Character recognition method

• US 5,745,599 A
• Filed: 01/18/1995
• Issued: 04/28/1998
• Est. Priority Date: 01/19/1994
• Status: Expired due to Term

• Current Assignee: Nippon Telegraph and Telephone Corporation
• Original Assignee: Nippon Telegraph and Telephone Corporation
• Inventors: Hiraiwa, Akira, Uchiyama, Tadasu, Sonehara, Noboru
• Primary Examiner(s): Boudreau, Leo
• Assistant Examiner(s): KELLEY, CHRISTOPHER S
• Application Number:
  US08/374,107
• Time in Patent Office: 1,196 Days
• Field of Search: 382/187, 382/188, 382/189, 382/197, 382/198, 382/202, 382/209, 382/217
• US Class Current: 382/217
• 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 present invention provides a handwritten character recognition method in which neighboring points are connected by straight lines and the segments in-between points are then interpolated using broken lines. Each broken line is equally divided into k segments, and a number is assigned to each division point therein according to the time sequence of the respective point. A (k+1)N×N (N is the number of strokes) matrix P consisting of the elements of row (k+1)(n-1)+1 and column k(m-1)+j comprising angle Ω formed by the straight line connecting division point i of stroke n and division point j of stroke m and the straight line connecting division point i of stroke n and division point j+1 of stroke m, is then calculated. The distance di between handwritten character Cx and matrix Qi (i=1-M) formed in the same manner from M templates corresponding to characters possessing the same number of strokes as character Cx, is calculated as the sum of the squares of each element of matrix P-Qi. The character Cz corresponding to the template Qz with the smallest distance dz is then selected as the recognition result.

Claims

1. A method for recognizing a handwritten input character by comparing said handwritten input character with a plurality of templates corresponding to a plurality of various types of characters based on time series information of points characterizing each stroke of said handwritten input character formed from at least one stroke, said handwritten character recognition method comprising the steps of:

expressing time series information of said points characterizing each stroke as coordinate points in a coordinate system;

calculating, with regard to a plurality of coordinate points within said coordinate system corresponding to a plurality of points characterizing a stroke of said handwritten input character, the relative angles formed by two straight lines formed by means of connecting a coordinate point, used as a reference, characterizing one stroke to two neighboring coordinate points from either a plurality of other coordinate points characterizing the same stroke as this reference coordinate point or a plurality of coordinate points characterizing another stroke, with regard to two neighboring coordinates of all of either of a plurality of other coordinate points characterizing the same said stroke or a plurality of coordinate points characterizing another stroke;

respectively calculating an angle the same as said relative angle using each coordinate point as a reference, with regard to all other coordinate points characterizing the same said stroke as said reference coordinate point of said handwritten input character or all of a plurality of coordinate points characterizing another stroke;

expressing a shape of said handwritten input character by means of a matrix consisting of elements comprising a quantity relating to each of said relative angles calculated with regard to said handwritten input character;

calculating, with regard to a plurality of coordinate points within said coordinate system corresponding to a plurality of points characterizing a stroke of each template, the relative angles formed by two straight lines formed by means of connecting a coordinate point, used as a reference, characterizing one stroke to each of two neighboring coordinate points of either other coordinate points characterizing the same stroke as this reference coordinate point or a plurality of coordinate points characterizing another stroke, with regard to all pairs of neighboring coordinate points of either a plurality of other coordinate points characterizing the same said stroke or a plurality of coordinate points characterizing another stroke;

respectively calculating an angle the same as said relative angle using each coordinate point as a reference with regard to all other coordinate points characterizing the same said stroke as said reference coordinate point of said template or all of a plurality of coordinate points characterizing another stroke;

expressing a shape of each of said templates by means of a matrix consisting of elements comprising a quantity relating to each of said relative angles calculated with regard to said template;

calculating a distance between said handwritten input character and each of said plurality of templates using said matrix expressing a shape for said handwritten input character, and said matrix expressing a shape for each template; and

designating a character corresponding to a template selected based on said distance as a recognition result of said handwritten input character.

2. A method according to claim 1 wherein said time series information of points is a horizontal direction component and a vertical direction component of a velocity vector of a moving pen tip occurring at each time during a pen movement during handwritten input, and said coordinate system regards said horizontal direction component and said vertical direction component of said velocity vector of said moving pen tip as coordinate values of a horizontal axis and a vertical axis, respectively.

4. A method according to claim 1, wherein said time series information of points characterizing each stroke represents displacement of the pen tip moving along each stroke;

and said coordinate system is a displacement coordinate system.

5. A method according to claim 4, wherein at least one stroke order when said character is written in printed form;

connection position where a stroke and its preceding stroke are continuously written when said character is written in running style using each stroke order; and

stroke order information of a character formed from a minimum number of strokes when said character is written in running style using each stroke order, are included as structural elements of said template.

6. A method according to claim 4, wherein intervals between neighboring points in said displacement coordinate system are connected by straight lines, and interpolated by means of broken line interpolation;

said broken lines resembling each stroke are divided into k broken lines (k≧

1) based on predetermined conditions; and

an angle formed by means of a straight line connecting division point An, I (I=1, 2, . . . , k+1) of an nth stroke and division point Am, j (j=1, 2, . . . , k) of an mth stroke, and a straight line connecting said division point An, I and division point Am, h (h=1, 2, . . . , k;

wherein h≠

d) is designated as said relative angle.

7. A method according to claim 6, wherein intervals between neighboring points in said displacement coordinate system are connected by straight lines, and interpolated by means of broken line interpolation;

said broken lines resembling each stroke are equally divided into k broken lines (k≧

1); and

an angle formed by means of a straight line connecting division point An, I (I=1, 2, . . . , k+1) of an nth stroke and division point Am, j (j=1, 2, . . . , k) of an mth stroke, and a straight line connecting said division point An, I and division point Am, j+1 (j=1, 2, . . . , k) is designated as said relative angle.

8. A method according to claim 7, wherein a character, number of strokes of said character when written in printed form, and coordinates of points in said displacement coordinate system of each stroke of said character when written in printed form are registered as structural elements of a template of said character.

9. A method according to claim 8, further comprising the steps of:

expressing the relationship of stroke length forming said handwritten input character by means of a vector comprising distance between an end point of a stroke and starting point of a stroke subsequent to said stroke, and length of each stroke, as essential components;

expressing the relationship of stroke length forming said template by means of a vector comprising distance between an end point of a stroke and starting point of a stroke subsequent to said stroke of a character formed by continuously writing neighboring strokes at several connection positions using each stroke order contained in said template, in a manner such that a character of said template and said handwritten input character possess the same number of strokes, and length of each stroke of said formed character, as essential components;

calculating a distance between said vectors for each stroke order and combination of connection positions;

distributing points in said displacement coordinate system of said template according to said stroke order and connection points providing for a minimum distance; and

calculating said relative angle.

10. A method according to claim 2, wherein all points in said velocity coordinate system are connected by straight lines, and interpolated by means of broken line interpolation;

said broken lines resembling each stroke are divided into k broken lines (k≧

1) based on predetermined conditions; and

an angle formed by means of a straight line connecting division point An, I (I=1, 2, . . . , k+1) of an nth stroke and division point Am, j (j=1, 2, . . . , k) of an mth stroke, and a straight line connecting said division point An, I and division point Am, h (h=1, 2, . . . , k;

wherein h≠

j) is designated as said relative angle.

11. A method according to claim 10 wherein all points in said velocity coordinate system are connected by straight lines, and interpolated by means of broken line interpolation;

said broken lines resembling each stroke are equally divided into k broken lines (k≧

1); and

an angle formed by means of a straight line connecting division point An, I (I=1, 2, . . . , k+1) of an nth stroke and division point Am, j (j=1, 2, . . . , k) of an mth stroke, and a straight line connecting said division point An, I and division point Am, j+1 (j=1, 2, . . . , k) is designated as said relative angle.

12. A method according to claim 11, wherein a character, number of strokes of said character when written in printed form, and coordinates of points in said velocity coordinate system of each stroke of said character when written in printed form are registered as structural elements of a template of said character.

13. A method according to claim 11, wherein at least one of the following is included as structural elements of said template:

stroke order when said character is written in printed form;

connection position where a stroke and its preceding stroke are continuously written when said character is written in running style using each stroke order; and

stroke order information of a character formed from a minimum number of strokes when said character is written in running style using each stroke order.

14. A method according to claim 11, further comprising the steps of:

expressing the relationship of stroke length forming said handwritten input character by means of a vector comprising distance between an end point of a stroke and a starting point of a stroke subsequent to said stroke, and length of each stroke, as essential components;

expressing the relationship of stroke length forming said template by means of a vector comprising distance between an end point of a stroke and a starting point of a stroke subsequent to said stroke of a character formed by continuously writing neighboring strokes at several connection positions using each stroke order contained in said template, in a manner such that a character of said template and said handwritten input character possess the same number of strokes, and length of each stroke of said formed character, as essential components;

calculating a distance between said vectors for each stroke order and combination of connection positions;

distributing points of said velocity coordinate system of said template according to said stroke order and connection points providing for a minimum distance; and

calculating said relative angle.

3. A method for recognizing a handwritten input character by comparing said handwritten input character with a plurality of templates corresponding to a plurality of various types of characters based on time series information of points characterizing each stroke of said handwritten input character formed from at least one stroke, said handwritten character recognition method comprising the steps of:

expressing time series information of a pen movement, which is a horizontal direction component and a vertical direction component of an acceleration vector of said pen movement occurring at each time during pen movement when inputting a handwritten character, as coordinate points in an acceleration coordinate system in which the coordinate values of a horizontal axis and a vertical axis are a horizontal direction component and a vertical direction component of a pen movement acceleration vector;

calculating, with regard to a plurality of coordinate points within said coordinate system corresponding to a plurality of points characterizing a stroke of said handwritten input character, the relative angles formed by two straight lines formed by means of connecting a coordinate point, used as a reference, characterizing one stroke to two neighboring coordinate points from either a plurality of other coordinate points characterizing the same stroke as this reference coordinate point or a plurality of coordinate points characterizing another stroke, with regard to two neighboring coordinates of all of either of a plurality of other coordinate points characterizing the same said stroke or a plurality of coordinate points characterizing another stroke;

respectively calculating an angle the same as said relative angle using each coordinate point as a reference, with regard to all other coordinate points characterizing the same said stroke as said reference coordinate point of said handwritten input character or all of a plurality of coordinate points characterizing another stroke;

expressing a shape of said handwritten input character by means of a matrix consisting of elements comprising a quantity relating to each of said relative angles calculated with regard to said handwritten input character;

calculating, with regard to a plurality of coordinate points within said coordinate system corresponding to a plurality of points characterizing a stroke of each template, the relative angles formed by two straight lines formed by means of connecting a coordinate point, used as a reference, characterizing one stroke to each of two neighboring coordinate points of either other coordinate points characterizing the same stroke as this reference coordinate point or a plurality of coordinate points characterizing another stroke, with regard to all pairs of neighboring coordinate points of either a plurality of other coordinate points characterizing the same said stroke or a plurality of coordinate points characterizing another stroke;

respectively calculating an angle the same as said relative angle using each coordinate point as a reference with regard to all other coordinate points characterizing the same said stroke as said reference coordinate point of said template or all of a plurality of coordinate points characterizing another stroke;

expressing a shape of each of said templates by means of a matrix consisting of elements comprising a quantity relating to each of said relative angles calculated with regard to said template;

calculating a distance between said handwritten input character and each of said plurality of templates using said matrix expressing a shape for said handwritten input character, and said matrix expressing a shape for each template; and

designating a character corresponding to a template selected based on said distance as a recognition result of said handwritten input character.

15. A method according to claim 3, wherein all points in said acceleration coordinate system are connected by straight lines, and interpolated by means of broken line interpolation;

said broken lines resembling each stroke are divided into k broken lines (k≧

1) based on predetermined conditions; and

an angle formed by means of a straight line connecting division point An, I (I=1, 2, . . . , k+1) of an nth stroke and division point Am, j (j=1, 2, . . . , k) to an mth stroke, and a straight line connecting said division point An, I and division point Am, h (h=1, 2, . . . k;

wherein h≠

j) is designated as said relative angle.

16. A method according to claim 15, wherein all points in said acceleration coordinate system are connected by straight lines, and interpolated by means of broken line interpolation;

said broken lines resembling each stroke are equally divided into k broken lines (k≧

1); and

an angle formed by means of a straight line connecting division point An, I (I=1, 2, . . . , k+1) of an nth stroke and division point Am, j (j=1, 2, . . . , k) of an mth stroke, and a straight line connecting said division point An, I and division point Am, j+1 (j=1, 2, . . . , k) is designated as said relative angle.

17. A method according to claim 16, wherein a character, number of strokes of said character when written in printed form, and coordinates of points in said acceleration coordinate system of each stroke of said character when written in printed form are registered as structural elements of a template of said character.

18. A method according to claim 16, wherein at least one of the following are included as structural elements in said template:

stroke order when said character is written in printed form;

connection position where a stroke and its preceding stroke are continuously written when said character is written in running style using each stroke order; and

stroke order information of a character formed from a minimum number of strokes when said character is written in running style using each stroke order.

19. A method according to claim 18, further comprising the steps of:

expressing the relationship of stroke length forming said handwritten input character by means of a vector comprising distance between an end point of a stroke and starting point of a stroke subsequent to said stroke, and length of each stroke, as essential components;

expressing the relationship of stroke length forming said template by means of a vector comprising distance between an end point of a stroke and a starting point of a stroke subsequent to said stroke of a character formed by continuously writing neighboring strokes at several connection positions using each stroke order contained in said template, in a manner such that a character of said template and said handwritten input character possess the same number of strokes, and length of each stroke of said formed character, as essential components;

calculating a distance between said vectors for each stroke order and combination of connection positions;

distributing points of said acceleration coordinate system of said template according to said stroke order and connection points providing for a minimum distance; and

calculating said relative angle.