Character recognition with variable subdivisions of a character region

Claim 2

A character recognizing method, comprising the steps of: optically reading an input character to be recognized to form a binary character image including at least one character line comprised of black pixels; broadening said character line in accordance with a predetermined manner; assigning a plurality of first features, different from one another, to each of pixels defining a contour of said binary character image depending on a local pattern of arrangement of black and white pixels along said contour; counting said first features assigned to said character image; subdiving said character image into a plurality of mesh sub-regions based on the count of said assigned first features; extracting a second feature from each of said sub-regions to define a collection of extracted second features; and identifying said input character by comparing said collection of extracted second features with each of a plurality of stored collections of second features for known characters and selecting one of said known characters, having a highest degree of similarity.

Claim 3

The method of claim 1, wherein said broadening step includes a step of changing a pair of white pixels on both of left-hand and right-hand sides of each of the black pixels of said character line into black pixels.

Claim 4

The method of claim 2, wherein said broadening step also includes a step of changing a pair of white pixels on both of top and bottom of each of the black pixels of said character line into black pixels.

Claim 5

The method of claim 1, wherein said plurality of first features include a predetermined number of directionality codes each of which is associated with a particular arrangement pattern of black and white pixels.

Claim 6

The method of claim 4, wherein said particular arrangement pattern includes a center pixel, a pair of left-hand and right-hand pixels adjacent to said center pixel and a pair of top and bottom pixels adjacent to said center pixel, whereby said center pixel functions as a pixel of interest when said assigning step is carried out.

Claim 7

The method of claim 5, wherein said center pixel is a black pixel.

Claim 8

The method of claim 5, wherein said center pixel is a white pixel.

Claim 9

The method of claim 1, wherein said contour defining pixels are black pixels.

Claim 10

The method of claim 1, wherein said contour defining pixels are white pixels.

Claim 11

The method of claim 1, wherein said second feature is a histogram of said first features in each of said sub-regions.

Claim 12

The method of claim 1, wherein the degree of similarity between said input character and each of said known characters is determined by calculating a distance between said second features of said input character and said second features of each of said known characters.

Claim 13

A character recognition method, comprising the steps of: providing a binary character image of an unknown character to be identified; assigning a plurality of directionality codes to pixels defining a contour of said binary character image selectively according to a predetermined rule to define a code-assigned character image; counting a total number of code-assigned pixels; forming a pair of first and second projected histograms of the number of code-assigned pixels, said first projected histogram being formed by projecting the code-assigned pixels onto a first predetermined axis and said second projected histogram being formed by projecting the code-assigned pixels onto a second predetermined axis; subdiving said code-assigned character image into a plurality of sub-regions based on said total number of code-assigned pixels and said pair of first and second histograms such that each of said sub-regions has a substantially equal number of code-assigned pixels; determining a histrogram of said codes for each of said sub-regions to form a collection of histograms of said codes for said input character; and identifying said input character by comparing said collection of histograms of said codes with each of a plurality of stored collections of histograms of said codes for known characters and selecting one of said known characters, having a highest degree of similarity.

Claim 14

A character recognition method, comprising the steps of: providing a binary character image of an unknown character to be identified; assigning a plurality of directionality codes to pixels defining a contour of said binary character image selectively according to a predetermined rule to define a code-assigned character image; counting a total number of code-assigned pixels; forming a pair of first and second projected histograms of the number of code-assigned pixels, said first projected histogram being formed by projecting the code-assigned pixels onto a first predetermined axis and said second projected histogram being formed by projecting the code-assigned pixels onto a second predetermined axis; subdiving said code-assigned character image into a plurality of sub-regions based on said total number of code-assigned pixels and said pair of first and second histograms such that each of said sub-regions has a substantially equal number of code-assigned pixels; determining a histrogram of said codes for each of said sub-regions to form a collection of histograms of said codes for said input character; and identifying said input character by comparing said collection of histograms of said codes with each of a plurality of stored collections of histograms of said codes for known characters and selecting one of said known characters, having a highest degree of similarity.

Claim 15

The method of claim 12, wherein said step of providing includes a step of optically scanning said unknown character to form said binary character image.

Claim 16

The method of claim 12, wherein said assigning and counting steps are carried out simultaneously.

Claim 17

The method of claim 14, wherein said assigning and counting steps are carried out by scanning said binary character image.

Claim 18

The method of claim 14, wherein said assigning and counting steps are carried out simultaneously with said step of providing.

Claim 19

The method of claim 16, wherein said step of providing is carried out by optically scanning said unknown character.

Claim 20

The method of claim 12, wherein each of said directionality codes is associated with a particular arrangement pattern of black and white pixels.

Claim 21

The method of claim 18, wherein said particular arrangement pattern includes a center pixel, a pair of left-hand and right-hand pixels adjacent to said center pixel and a pair of top and bottom pixels adjacent to said center pixel, whereby said center pixel functions as a pixel of interest when said assigning step is carried out.

Claim 22

The method of claim 19, wherein said center pixel is a black pixel.

Claim 23

The method of claim 19, wherein said center pixel is a white pixel.

Claim 24

The method of claim 12, wherein said contour defining pixels are black pixels.

Claim 25

The method of claim 12, wherein said contour defining pixels are white pixels.

Claim 26

The method of claim 12, wherein the degree of similarity between said input character and each of said known characters is determined by calculating a distance between the collection of histograms of said codes of said input character and the collection of histograms of said codes of each of said known characters.

Claim 27

The method of claim 12, wherein said first and second predetermined directions extend normal to each other.

Claim 28

The method of claim 25, wherein said first direction is a horizontal direction and said second direction is a vertical direction.

Claim 29

A character recognition method comprising the steps of: providing a binary character image of an unknown character to be identified; assigning a plurality of directionality codes to pixels defining a contour of said binary character image selectively according to a predetermined rule to define a code-assigned character image; counting a total number of code-assigned pixels; forming a first projected histogram of the code-assigned pixels, said first projected histogram being formed by projecting the code-assigned pixels onto a first predetermined axis; subdiving said code-assigned character image into a plurality of first sections in said first predetermined axis based on said total number of code-assigned pixels and said first projected histogram such that each of said first sections has a substantially equal number of code-assigned pixels; subdiving each of said first sections into a plurality of second sections such that each of said second sections has a substantially equal number of code-assigned pixels, thereby subdividing said code-assigned character image into a plurality of sub-regions in the form of mesh; determining a histrogram of said codes for each of said sub-regions to form a collection of histograms of said codes for said input character; and identifying said input character by comparing said collection of histograms of said codes with each of a plurality of stored collections of histograms of said codes for known characters and selecting one of said known characters, having a highest degree of similarity.

Claim 30

A character recognition method comprising the steps of: providing a binary character image of an unknown character to be identified; assigning a plurality of directionality codes to pixels defining a contour of said binary character image selectively according to a predetermined rule to define a code-assigned character image; counting a total number of code-assigned pixels; forming a first projected histogram of the code-assigned pixels, said first projected histogram being formed by projecting the code-assigned pixels onto a first predetermined axis; subdiving said code-assigned character image into a plurality of first sections in said first predetermined axis based on said total number of code-assigned pixels and said first projected histogram such that each of said first sections has a substantially equal number of code-assigned pixels; subdiving each of said first sections into a plurality of second sections such that each of said second sections has a substantially equal number of code-assigned pixels, thereby subdividing said code-assigned character image into a plurality of sub-regions in the form of mesh; determining a histrogram of said codes for each of said sub-regions to form a collection of histograms of said codes for said input character; and identifying said input character by comparing said collection of histograms of said codes with each of a plurality of stored collections of histograms of said codes for known characters and selecting one of said known characters, having a highest degree of similarity.

Claim 31

The method of claim 27 further comprising the step of counting a number of code-assigned pixels in each of said first sections prior to the step of subdividing each of said first sections.

Claim 32

The method of claim 27, wherein said step of providing includes a step of optically scanning said unknown character to form said binary character image.

Claim 33

The method of claim 27, wherein said assigning step and total number counting step are carried out simultaneously.

Claim 34

The method of claim 30, wherein said assigning and total number counting steps are carried out by scanning said binary character image.

Claim 35

The method of claim 30, wherein said assigning and total number counting steps are carried out simultaneously with said step of providing.

Claim 36

The method of claim 32, wherein said step of providing is carried out by optically scanning said unknown character.

Claim 37

The method of claim 27, wherein each of said directionality codes is associated with a particular arrangement pattern of black and white pixels.

Claim 38

The method of claim 34, wherein said particular arrangement pattern includes a center pixel, a pair of left-hand and right-hand pixels adjacent to said center pixel and a pair of top and bottom pixels adjacent to said center pixel, whereby said center pixel functions as a pixel of interest when said assigning step is carried out.

Claim 39

The method of claim 35, wherein said center pixel is a black pixel.

Claim 40

The method of claim 35, wherein said center pixel is a white pixel.

Claim 41

The method of claim 27, wherein said contour defining pixels are black pixels.

Claim 42

The method of claim 27, wherein said contour defining pixels are white pixels.

Claim 43

The method of claim 27, wherein the degree of similarity between said input character and each of said known characters is determined by calculating a distance between the collection of histograms of said codes of said input character and the collection of histograms of said codes of each of said known characters.

Claim 44

The method of claim 27, wherein said first and second predetermined directions extend normal to each other.

Claim 45

The method of claim 41, wherein said first direction is a horizontal direction and said second direction is a vertical direction.

Claim 46

A character subdividing method, comprising the steps of: providing a binary character image of an input character; assigning a plurality of predetermined features to said binary character image selectively in accordance with a predetermined rule, thereby defining a feature-assigned character image; counting a total number of assigned features; forming a pair of first and second projected histograms of said assigned features, said first projected histogram being formed by projecting said assigned features onto a first predetermined axis and said second projected histogram being formed by projecting said assigned features onto a second predetermined axis; and subdividing said feature-assigned character image into a plurality of sub-regions based on said total number of said assigned features and said pair of first and second projected histograms such that said sub-regions have substantially equal numbers of said assigned features.

Claim 47

A character subdividing method, comprising the steps of: providing a binary character image of an input character; assigning a plurality of predetermined features to said binary character image selectively in accordance with a predetermined rule, thereby defining a feature-assigned character image; counting a total number of assigned features; forming a pair of first and second projected histograms of said assigned features, said first projected histogram being formed by projecting said assigned features onto a first predetermined axis and said second projected histogram being formed by projecting said assigned features onto a second predetermined axis; and subdividing said feature-assigned character image into a plurality of sub-regions based on said total number of said assigned features and said pair of first and second projected histograms such that said sub-regions have substantially equal numbers of said assigned features.

Claim 48

The method of claim 43, wherein said assigning and counting steps are carried out simultaneously.

Claim 49

The method of claim 44, wherein said step of providing is carried out simulataneously with said assigning and counting steps.

Claim 50

The method of claim 43, wherein said step of providing is carried out by optically scanning said input character.

Claim 51

A character subdividing method, comprising the steps of: providing a binary character image of an input character; assigning a plurality of predetermined features to said binary character image selectively in accordance with a predetermined rule, thereby defining a feature-assigned character image; counting a total number of assigned features; forming a first projected histogram of said assigned features, said first projected histogram being formed by projecting said assigned features onto a first predetermined axis and said second projected histogram being formed by projecting said assigned features onto a second predetermined axis; and subdividing said feature-assigned character image into a plurality of first sections in said first predetermined axis based on said total number and said first projected histogram such that said first sections have substantially equal numbers of said assigned features; and subdividing each of said first sections into a plurality of second sections such that each of said second sections has a substantially equal number of said assigned features, thereby subdividing said feature-assigned character image into a plurality of sub-regions in the form of a mesh.

Claim 52

A character subdividing method, comprising the steps of: providing a binary character image of an input character; assigning a plurality of predetermined features to said binary character image selectively in accordance with a predetermined rule, thereby defining a feature-assigned character image; counting a total number of assigned features; forming a first projected histogram of said assigned features, said first projected histogram being formed by projecting said assigned features onto a first predetermined axis and said second projected histogram being formed by projecting said assigned features onto a second predetermined axis; and subdividing said feature-assigned character image into a plurality of first sections in said first predetermined axis based on said total number and said first projected histogram such that said first sections have substantially equal numbers of said assigned features; and subdividing each of said first sections into a plurality of second sections such that each of said second sections has a substantially equal number of said assigned features, thereby subdividing said feature-assigned character image into a plurality of sub-regions in the form of a mesh.

Claim 53

The method of claim 47, wherein said assigning and counting steps are carried out simultaneously.

Claim 54

The method of claim 47, wherein said step of providing is carried out simultaneously with said assigning and counting steps.

Claim 55

The method of claim 47, wherein said step of providing is carried out by optically scanning said input character.

Claim 56

The method of claim 47 further comprising the step of counting a number of assigned features in each of said first sections prior to the step of subdividing each of said first sections.