Character recognition apparatus

Claim 2

A character recognition apparatus comprising document data input means for inputting document data representing at least one row of characters, and character separating means for extracting the row of characters from the document data input by said document data input means, and separating the characters; characterized in that said character recognition apparatus includes: histogram forming means (21) for forming a histogram pertaining to the row of characters which said character separating means (14) has extracted from the document data; reference line defining means (15, 15a) for defining at least one peak in the histogram formed by said histogram forming means, thereby to define a reference line from the peak thus detected; character sorting means (16) for sorting the characters into a plurality of categories in accordance with the positions which the characters take with respect to the reference line defined by said reference line defining means; and character recognizing means (17) for recognizing the characters sorted by said character sorting means, using dictionary patterns of a category corresponding to the sorted characters.

Claim 3

The apparatus according to claim 1, charac­terized in that said reference line defining means includes means (35) for detecting two peaks of said histogram and defining two reference lines from these peaks, and said character sorting means includes means (41, 43, 44) for sorting the characters in accordance with the sizes of the characters and the positions which the character take with respect to the two reference lines.

Claim 4

The apparatus according to claim 2, charac­terized in that said characters include tiny characters, standard-size characters, and nonstandard-size charac­ters; and said reference line defining means includes means (25) for forming a histogram of all characters except for the tiny characters, and means (26, 27) for defining said reference lines from the peaks of this histogram.

Claim 5

The apparatus according to claim 3, charac­terized in that said reference line defining means has means (24) for detecting tiny characters, and means (28) for detecting characters other than tiny characters, which said means has detected by error.

Claim 6

The apparatus according to claim 2, charac­terized in that said reference line defining means includes means (31, 32) for detecting the inclination of the row of characters, and means (33) for eliminating the inclination of the row of characters.

Claim 7

The apparatus according to claim 1, charac­terized in that said character sorting means (16) sorts said characters into a group of tiny characters, a group of standard-size characters, and a group of any other characters.

Claim 8

The apparatus according to claim 6, charac­terized in that said character sorting means (16) sorts said characters into superscripts, subscripts, and central characters, in accordance with the positions which these characters take with respect to said two reference lines.

Claim 9

The apparatus according to claim 6, charac­terized in that said character sorting means (16) sorts said characters into ascender characters, descender characters, and tall characters, in accordance with the lengths of those portions of the characters which project above and below said reference lines.

Claim 10

The apparatus according to claim 1, charac­terized in that said character separating means (14) outputs data items representing rectangles bordering the extremities of said characters, and said reference line defining means (15a) comprises center-line detecting means including data-converting means (51) for each of the rectangle into a function data item representing the position and height of the character corresponding to the rectangle, means (52) for forming a histogram from the function data items output by said data-converting means, and means (53) for detecting the peak of the histogram, thus defining a center line from the peak, and outputting data representing the center line, to said character sorting means.

Claim 11

The apparatus according to claim 9, charac­terized in that said character sorting means (16) sorts the characters into a plurality of groups in accordance with the difference in the height of the rectangles.

Claim 12

The apparatus according to claim 9, charac­terized in that said character sorting means (16) sorts the characters of each group into a plurality of categories in accordance with the signs of the differences in the height of the rectangles.

Claim 13

A character recognition apparatus comprising: document data input means for inputting document data representing at least one row of characters includ­ing tiny ones, standard-size ones, and ones of any other sizes; and character separating means for extracting the row of character from the document data output by said docu­ment data output meant, and separating the characters of the row: characterized in that said character recognition apparatus includes: histogram forming means (21) for forming a histo­gram pertaining to the row of characters which said character separating means has extracted from the document data; reference line defining means (35) for defining at least two peaks in the histogram formed by said histogram forming means, thereby to define two reference lines from the peaks thus detected; character sorting means (16) for sorting the characters into tiny ones, standard-size ones, and ones of any other sizes in accordance with the sizes of the characters and the positions which the characters take with respect to the two reference line defined by said reference line defining means; and character recognizing means (17) for recognizing the character sorted by said character sorting means, by comparing the patterns of these characters with dictionary patterns.

Claim 14

A character recognition apparatus comprising: document data input means for inputting document data representing at least one row of characters includ­ing tiny ones, standard-size ones, and ones of any other sizes; and character separating means for extracting the row of character from the document data output by said docu­ment data output meant, and separating the characters of the row: characterized in that said character recognition apparatus includes: histogram forming means (21) for forming a histo­gram pertaining to the row of characters which said character separating means has extracted from the document data; reference line defining means (35) for defining at least two peaks in the histogram formed by said histogram forming means, thereby to define two reference lines from the peaks thus detected; character sorting means (16) for sorting the characters into tiny ones, standard-size ones, and ones of any other sizes in accordance with the sizes of the characters and the positions which the characters take with respect to the two reference line defined by said reference line defining means; and character recognizing means (17) for recognizing the character sorted by said character sorting means, by comparing the patterns of these characters with dictionary patterns.

Claim 15

The apparatus according to claim 12, charac­terized in that said histogram forming means (21) forms a histogram pertaining to all characters of each row, except for the tiny ones.

Claim 16

The apparatus according to claim 12, charac­terized by further comprising a pattern dictionary (18) storing the dictionary patterns of characters sorted in various categories, said pattern dictionary (18) being so designed to output the patterns of characters of any category for comparison with any character of the same category.

Claim 17

A character recognition apparatus comprising: document data input means (11) for inputting docu­ment data representing at least one row of characters; character separating means (14) for extracting the row of character from the document data input by said document data output means, and separating the charac­ters from the row; and reference line defining means (15) for simulating straight lines passing specific points of the characters separated from the row, forming a histogram pertaining to the characters in a parameter space defined by the coordinates parameters defining the straight lines, detecting, as a reference line, the straight line defined by a coordinates parameter corresponding to the peak of the histogram, and detecting the inclination of this reference line.

Claim 18

A character recognition apparatus comprising: document data input means (11) for inputting docu­ment data representing at least one row of characters; character separating means (14) for extracting the row of character from the document data input by said document data output means, and separating the charac­ters from the row; and reference line defining means (15) for simulating straight lines passing specific points of the characters separated from the row, forming a histogram pertaining to the characters in a parameter space defined by the coordinates parameters defining the straight lines, detecting, as a reference line, the straight line defined by a coordinates parameter corresponding to the peak of the histogram, and detecting the inclination of this reference line.

Claim 19

The apparatus according to claim 15, charac­terized in that said reference line detecting means (15) detects the reference line by means of the Hough voting.

Claim 20

The apparatus according to claim 15, charac­terized in that said reference line detecting means (15) detects the reference line by using the lower extremity of each character as said specific point.