Character recognition method and apparatus

Claim 2

A character recognition apparatus comprising: a partial pattern dictionary for registering partial patterns of character images in a form of data indicating strokes, the data for each partial pattern are ordered in a predetermined rule which is independent of a stroke input sequence of a character image; a partial pattern composition dictionary for registering character images which are composed of a single or a plurality of partial patterns; input means for inputting strokes of a character image to be recognized; obtaining means for obtaining data indicating strokes inputted by said input means; sorting means for sorting the data obtained by said obtaining means in accordance with the predetermined rule; first extraction means for extracting a single or plurality of partial patterns by searching said partial pattern dictionary based upon the data sorted by said sorting means; second extraction means for extracting a pertinent character by searching said partial pattern composition dictionary with the partial pattern which is obtained by said first extraction means; and output means for outputting a result of character recognition based on the searched character which is obtained by said second extraction means.

Claim 3

The character recognition apparatus according to claim 1, wherein the inputting strokes in said input means is performed by a digitizer.

Claim 4

The character recognition apparatus according to claim 1, wherein said output means has a display which displays the result of the character recognition.

Claim 5

A character recognition apparatus comprising: feature stroke dictionary for registering a combination of direction vectors of each stroke, which are obtained by dividing a single stroke into a predetermined number of parts, as a feature stroke; partial pattern dictionary for registering partial patterns of a character image in a form of data indicating the feature strokes, the data for each partial pattern are ordered in a predetermined rule which is independent of a stroke input sequence of a character image; partial pattern composition dictionary for registering character images which are composed of a single or a plurality of the partial patterns; input means for inputting strokes of a character image to be recognized; direction vector obtaining means for obtaining direction vectors for each stroke by dividing a stroke into the predetermined number of parts; feature stroke obtaining means for obtaining data, each of which indicates pertinent feature stroke corresponding to the input stroke by searching said feature stroke dictionary with the combination of direction vectors which is obtained by said direction vector obtaining means; sorting means for sorting the data obtained by said feature stroke obtaining means based on the predetermined rule; first extraction means for extracting a single or a plurality of partial patterns by searching said partial pattern dictionary with the data sorted by said sorting means; second extraction means for extracting a pertinent character by searching a partial pattern composition dictionary, said partial pattern composition dictionary registers character images comprising a single or plurality of partial patterns, with the partial which is obtained by said first extraction step; and output step of outputting a character recognition result based on the searched character which is obtained by said second extraction step.

Claim 6

A character recognition apparatus comprising: feature stroke dictionary for registering a combination of direction vectors of each stroke, which are obtained by dividing a single stroke into a predetermined number of parts, as a feature stroke; partial pattern dictionary for registering partial patterns of a character image in a form of data indicating the feature strokes, the data for each partial pattern are ordered in a predetermined rule which is independent of a stroke input sequence of a character image; partial pattern composition dictionary for registering character images which are composed of a single or a plurality of the partial patterns; input means for inputting strokes of a character image to be recognized; direction vector obtaining means for obtaining direction vectors for each stroke by dividing a stroke into the predetermined number of parts; feature stroke obtaining means for obtaining data, each of which indicates pertinent feature stroke corresponding to the input stroke by searching said feature stroke dictionary with the combination of direction vectors which is obtained by said direction vector obtaining means; sorting means for sorting the data obtained by said feature stroke obtaining means based on the predetermined rule; first extraction means for extracting a single or a plurality of partial patterns by searching said partial pattern dictionary with the data sorted by said sorting means; second extraction means for extracting a pertinent character by searching a partial pattern composition dictionary, said partial pattern composition dictionary registers character images comprising a single or plurality of partial patterns, with the partial which is obtained by said first extraction step; and output step of outputting a character recognition result based on the searched character which is obtained by said second extraction step.

Claim 7

The character recognition apparatus according to claim 4, wherein said input means inputs strokes of a character image to be recognized as a plurality of coordinate data by sampling points of the stroke at predetermined interval, and said direction vector obtaining means divides the stroke into the predetermined number of portions based on the coordinate data.

Claim 8

The character recognition apparatus according to claim 4, wherein said direction vector obtaining means obtains three direction vectors for each stroke by dividing each stroke into three parts.

Claim 9

The character recognition apparatus according to claim 4, wherein said first extraction means searches the partial pattern dictionary with the arranged order while the number of feature strokes is changed from a maximum number of a minimum number in order to extract a partial pattern which has the most feature strokes obtained by said feature stroke obtaining means.

Claim 10

The character recognition apparatus according to claim 4, wherein said partial pattern composition dictionary classifies each character in the dictionary based on a partial pattern of a character image first input in a process of stroke input and other partial patterns sequentially input after that, and registers these classified partial patterns, and said second extraction means searches said partial pattern composition dictionary by means of the partial pattern which is first extracted by said first extraction means based on the input order of the strokes and searches a pertinent character by using partial patterns other than the first extracted partial pattern.

Claim 11

The character recognition apparatus according to claim 4, wherein said second extraction means searches said partial pattern composition dictionary based on the partial pattern which is obtained by said first extraction means and extracts the character having the most number of partial patterns.

Claim 12

The character recognition apparatus according to claim 4, wherein the inputting strokes in said input means is performed by a digitizer.

Claim 13

The character recognition apparatus according to claim 4, wherein said output means has a display which displays the result of the character recognition.

Claim 14

A character recognition method comprising: input step of inputting strokes of a character image to be recognized; obtaining step of obtaining direction data indicating direction of strokes inputted by said input step; arranging step of arranging the data obtained by said obtaining step in accordance with a predetermined rule which is independent of a stroke input sequence of a character image; first extraction step of extracting a single or plurality of partial patterns of the character image by searching a partial pattern dictionary, said partial pattern dictionary registers partial patterns of character images in a form of data indicating strokes, the data for each partial pattern are ordered based on the predetermined rule based upon the data arranged by said arranging step; second extraction step of extracting a pertinent character by searching a partial pattern composition dictionary, said partial pattern composition dictionary registers character images comprising a single or plurality of partial patterns, with the partial pattern which is obtained by said first extraction step; and output step of outputting a character recognition result based on the searched character which is obtained by said second extraction step.

Claim 15

A character recognition method comprising: input step of inputting strokes of a character image to be recognized; obtaining step of obtaining direction data indicating direction of strokes inputted by said input step; arranging step of arranging the data obtained by said obtaining step in accordance with a predetermined rule which is independent of a stroke input sequence of a character image; first extraction step of extracting a single or plurality of partial patterns of the character image by searching a partial pattern dictionary, said partial pattern dictionary registers partial patterns of character images in a form of data indicating strokes, the data for each partial pattern are ordered based on the predetermined rule based upon the data arranged by said arranging step; second extraction step of extracting a pertinent character by searching a partial pattern composition dictionary, said partial pattern composition dictionary registers character images comprising a single or plurality of partial patterns, with the partial pattern which is obtained by said first extraction step; and output step of outputting a character recognition result based on the searched character which is obtained by said second extraction step.

Claim 16

The character recognition apparatus according to claim 12, wherein the inputting strokes in said input means is performed by a digitizer.

Claim 17

The character recognition method according to claim 12, wherein said output step displays the result of the character recognition on a display.

Claim 18

A character recognition method comprising: input step of inputting a stroke of a character image to be recognized; direction vector obtaining step of obtaining direction vectors for each stroke by dividing a stroke into a predetermined number of parts; feature stroke obtaining step of obtaining data each of which indicates a pertinent feature stroke corresponding to the stroke inputted by said input step by searching a feature stroke dictionary which registers a combination of direction vectors of each stroke, which are obtained by dividing a single stroke into a predetermined number of parts, as a feature stroke with the combination of direction vectors which is obtained by said direction vector obtaining step; sorting step of sorting the data obtained by said feature stroke obtaining step in accordance with a predetermined rule which is independent of a stroke input sequence of a character image; first extraction step of extracting a single or plurality of partial patterns by searching a partial pattern dictionary, which registers partial patterns of a character image in a form of data which is ordered based on the predetermined rule, with feature strokes sorted by said sorting step; second extraction step of extracting a pertinent character by searching a partial pattern composition dictionary which registers a character image which is composed of a single or plurality of the partial patterns with the partial pattern which is obtained by said first extraction step; and output step of outputting a result of character recognition based on the searched character which is obtained by said second extraction step.

Claim 19

A character recognition method comprising: input step of inputting a stroke of a character image to be recognized; direction vector obtaining step of obtaining direction vectors for each stroke by dividing a stroke into a predetermined number of parts; feature stroke obtaining step of obtaining data each of which indicates a pertinent feature stroke corresponding to the stroke inputted by said input step by searching a feature stroke dictionary which registers a combination of direction vectors of each stroke, which are obtained by dividing a single stroke into a predetermined number of parts, as a feature stroke with the combination of direction vectors which is obtained by said direction vector obtaining step; sorting step of sorting the data obtained by said feature stroke obtaining step in accordance with a predetermined rule which is independent of a stroke input sequence of a character image; first extraction step of extracting a single or plurality of partial patterns by searching a partial pattern dictionary, which registers partial patterns of a character image in a form of data which is ordered based on the predetermined rule, with feature strokes sorted by said sorting step; second extraction step of extracting a pertinent character by searching a partial pattern composition dictionary which registers a character image which is composed of a single or plurality of the partial patterns with the partial pattern which is obtained by said first extraction step; and output step of outputting a result of character recognition based on the searched character which is obtained by said second extraction step.

Claim 20

The character recognition method according to claim 15, wherein said input step inputs strokes of a character image to be recognized as a plurality of coordinate data by sampling points of the stroke at predetermined interval, and said direction vector obtaining step divides the stroke into the predetermined number of portions based on the coordinate data.

Claim 21

The character recognition method according to claim 15, wherein said direction vector obtaining step obtains three direction vectors for each stroke by dividing each stroke into three parts.

Claim 22

The character recognition method according to claim 15, wherein said first extraction step searches the partial pattern dictionary with the arranged order while the number of feature strokes is changed from a maximum number to a minimum number in order to extract a partial pattern having the most feature strokes obtained by said feature stroke obtaining step.

Claim 23

The character recognition method according to claim 15, wherein said partial pattern composition dictionary classifies each character in the dictionary based on a partial pattern of a character image first input in a process of stroke input and other partial patterns sequentially input after that, and registers these classified partial patterns, and said second extraction step searches said partial pattern composition dictionary by means of the partial pattern which is first extracted by said first extraction step based on the input order of the strokes and searches a pertinent character by using partial patterns other than the first extracted partial pattern.

Claim 24

The character recognition method according to claim 15, wherein said second extraction step searches said partial pattern composition dictionary based on the order of the partial pattern which is obtained by said first extraction step and extracts the character including the most number of partial patterns.

Claim 25

The character recognition apparatus according to claim 15, wherein the inputting strokes in said input means is performed by a digitizer.

Claim 26

The character recognition method according to claim 15, wherein said output step displays the result of the character recognition on a display.

Claim 27

A character recognition apparatus comprising: a pattern dictionary for registering reference data constituted by a plurality of direction data of a pattern, wherein the plurality of direction data are ordered in a predetermined rule which is independent of a stroke input sequence of a character image; input means for inputting strokes; direction obtaining means for obtaining direction data of the strokes input by said input means; sorting means for sorting the direction data obtained by said direction obtaining means in accordance with said predetermined rule; matching means for matching the sorted direction data with the reference data registered in the pattern dictionary wherein the matching is performed based on a sorted order; output means for outputting a pattern corresponding to reference data which get high score in said matching means.

Claim 28

A character recognition apparatus comprising: a pattern dictionary for registering reference data constituted by a plurality of direction data of a pattern, wherein the plurality of direction data are ordered in a predetermined rule which is independent of a stroke input sequence of a character image; input means for inputting strokes; direction obtaining means for obtaining direction data of the strokes input by said input means; sorting means for sorting the direction data obtained by said direction obtaining means in accordance with said predetermined rule; matching means for matching the sorted direction data with the reference data registered in the pattern dictionary wherein the matching is performed based on a sorted order; output means for outputting a pattern corresponding to reference data which get high score in said matching means.

Claim 29

The character recognition apparatus according to claim 23, wherein said input means inputs strokes of a character image to be recognized as a plurality of coordinate data by sampling points of the stroke at predetermined intervals, and said direction obtaining means divides the stroke into the predetermined number of portions based on the coordinates data.

Claim 30

The character recognition apparatus according to claim 23, wherein said direction obtaining means obtains three direction vectors for each stroke by dividing each stroke into three parts.

Claim 31

The character recognition apparatus according to claim 23, wherein the inputting strokes in said input means is performed by a digitizer.

Claim 32

The character recognition method according to claim 23, wherein said output step displays the result of the character recognition on a display.

Claim 33

A character recognition method comprising: input step of inputting strokes; direction obtaining step of obtaining direction data of the strokes inputted in said input step; sorting step of sorting direction data obtained in said direction obtaining step based on a predetermined rule which is independent of a stroke input sequence of a character image; matching step of matching the stored direction data with reference data registered in a pattern dictionary, which registers the reference data composed of a plurality of direction data wherein the plurality of direction data are ordered based on the predetermined rule; and output step of outputting a pattern corresponding to reference data which get high score in said matching step.

Claim 34

A character recognition method comprising: input step of inputting strokes; direction obtaining step of obtaining direction data of the strokes inputted in said input step; sorting step of sorting direction data obtained in said direction obtaining step based on a predetermined rule which is independent of a stroke input sequence of a character image; matching step of matching the stored direction data with reference data registered in a pattern dictionary, which registers the reference data composed of a plurality of direction data wherein the plurality of direction data are ordered based on the predetermined rule; and output step of outputting a pattern corresponding to reference data which get high score in said matching step.

Claim 35

The character recognition method according to claim 28, wherein said input step includes strokes of a character image to be recognized as a plurality of coordinate data by sampling points of the stroke at predetermined intervals, and said direction obtaining step divides the stroke into the predetermined number of portions based on the coordinate data.

Claim 36

The character recognition method according to claim 28, wherein said direction obtaining step obtains three direction vectors for each stroke by dividing each stroke into three parts.

Claim 37

The character recognition apparatus according to claim 28, wherein the inputting strokes in said input means is performed by a digitizer.

Claim 38

The character recognition method according to claim 28, wherein said output step displays the result of the character recognition on a display.