US4977602A — Character normalization using an elliptical sampl…

Claim 1

Independent Claim 192 words

In an optical character recognition system in which a text image comprising a plurality of characters is separated into plural individual character images, each of said individual character images comprising an array of original pixels of dimensions x max and y max in respective orthogonal directions, each of said pixels being one of a plurality of pixel values, a character normalization device for normalizing each of said individual character images to a size X max =Ax max and Y max =By max in said respective orthogonal directions in accordance with orthogonal normalization ratios A and B respectively, A and B being any real numbers, comprising: means for defining a set of normalized image pixel locations, said normalized image pixel locations having a pitch relative to said original pixels in each of said orthogonal directions corresponding to said orthogonal normalization ratios A and B, respectively; means for computing a value for each one of said normalized pixel locations from the values of the original pixels which lie within an elliptical neighborhood surrounding said one normalized pixel location, whereby to compute a set of normalized pixel location values defining a normalized individual character image.

Claim 2

Dependent Claim 192 words

In an optical character recognition system in which a text image comprising a plurality of characters is separated into plural individual character images, each of said individual character images comprising an array of original pixels of dimensions x max and y max in respective orthogonal directions, each of said pixels being one of a plurality of pixel values, a character normalization device for normalizing each of said individual character images to a size X max =Ax max and Y max =By max in said respective orthogonal directions in accordance with orthogonal normalization ratios A and B respectively, A and B being any real numbers, comprising: means for defining a set of normalized image pixel locations, said normalized image pixel locations having a pitch relative to said original pixels in each of said orthogonal directions corresponding to said orthogonal normalization ratios A and B, respectively; means for computing a value for each one of said normalized pixel locations from the values of the original pixels which lie within an elliptical neighborhood surrounding said one normalized pixel location, whereby to compute a set of normalized pixel location values defining a normalized individual character image.

Claim 3

Dependent Claim 17 words

The device of claim 1 wherein said elliptical neighborhood is centered on said one normalized pixel location.

Claim 4

Dependent Claim 36 words

The device of claim 2 wherein said elliptical neighborhood is characterized by major and minor axes whose lengths are approximately equal to the lengths spanned in said original image by A and B original pixels, respectively.

Claim 5

Dependent Claim 62 words

The device of claim 3 wherein said plurality of pixel values is two such that said image is a binary image, and wherein said means for computing assigns said one normalized pixel location a first value if the value of any of the original pixels lying within the corresponding elliptical neighborhood is said first value and otherwise assigns it a second value.

Claim 6

Dependent Claim 74 words

The device of claim 3 wherein the location of said one normalized pixel location in said original image is (X,Y), the location of a given one of said original pixels is (x,y) and said means for computing determines that said given one of said original pixels lies within said elliptical neighborhood if: (X-x).sup.2 /(A/2).sup.2 +(Y-y).sup.2 /(B/2).sup.2 <R or =R, wherein R is a number which determines the amount of overlap between adjacent elliptical neighborhoods.

Claim 7

Dependent Claim 7 words

The device of claim 5 wherein R=1.

Claim 8

Dependent Claim 46 words

The device of claim 5 wherein, for each one of said elliptical neighborhoods, said given one of said original pixels is selected from a rectangle of pixels centered around said elliptical neighborhood, said rectangle characterized by sides of lengths A and B in said original image.

Claim 9

Dependent Claim 153 words

In an optical character recognition system in which a text image comprising a plurality of characters is separated into plural individual character images, each of said individual character images comprising an array of original pixels, each of said pixels characterized by one of a plurality of pixel values, a method for scaling each of said individual character images in accordance with orthogonal scaling ratios A and B comprising: defining a set of normalized image pixel locations, said normalized image pixel locations having a pitch relative to said original pixels in each of said orthogonal directions corresponding to said orthogonal normalization ratios A and B, respectively; computing a value for each one of said normalized pixel locations from the values of the original pixels which lie within an elliptical neighborhood surrounding said one normalized pixel location; and transmitting the normalized pixel location values computed by said computing step as a normalized individual character image.

Claim 10

Dependent Claim 153 words

In an optical character recognition system in which a text image comprising a plurality of characters is separated into plural individual character images, each of said individual character images comprising an array of original pixels, each of said pixels characterized by one of a plurality of pixel values, a method for scaling each of said individual character images in accordance with orthogonal scaling ratios A and B comprising: defining a set of normalized image pixel locations, said normalized image pixel locations having a pitch relative to said original pixels in each of said orthogonal directions corresponding to said orthogonal normalization ratios A and B, respectively; computing a value for each one of said normalized pixel locations from the values of the original pixels which lie within an elliptical neighborhood surrounding said one normalized pixel location; and transmitting the normalized pixel location values computed by said computing step as a normalized individual character image.

Claim 11

Dependent Claim 17 words

The method of claim 8 wherein said elliptical neighborhood is centered on said one normalized pixel location.

Claim 12

Dependent Claim 36 words

The method of claim 9 wherein said elliptical neighborhood is characterized by major and minor axes whose lengths are approximately equal to the lengths spanned in said original image by A and B original pixels, respectively.

Claim 13

Dependent Claim 62 words

The method of claim 10 wherein said plurality of pixel values is two such that said image is a binary image, and wherein said computing step comprises assigning said one normalized pixel location a first value if the value of any of the original pixels lying within the corresponding elliptical neighborhood is said first value and otherwise assigning it a second value.

Claim 14

Dependent Claim 73 words

The method of claim 10 wherein the location of said one normalized pixel location in said original image is (X,Y), the location of a given one of said original pixels is (x,y) and computing step comprises determining that said given one of said original pixels lies within said elliptical neighborhood if: (X-x).sup.2 /(A/2).sup.2 +(Y-y).sup.2 /(B/2).sup.2 <R or =R, wherein R is a number which determines the amount of overlap between adjacent elliptical neighborhoods.

Claim 15

Dependent Claim 7 words

The method of claim 12 wherein R=1.

Claim 16

Dependent Claim 39 words

The method of claim 12 further comprising selecting said given one of said original pixels from a rectangle of pixels centered around said elliptical neighborhood, said rectangle characterized by sides of lengths A and B in said original image.

Claim 17

Dependent Claim 153 words

In a method for character normalization in optical character recognition in which the image of an individual character comprising an array of original pixels is normalized in accordance with horizontal and vertical scaling factors A and B to construct a normalized character image comprising an array of normalized pixels, the improvement comprising the steps of: defining in said array of original pixels a plurality of elliptical neighborhoods each having a minor axis and a major axis, said minor and major axes having lengths corresponding to said horizontal and vertical scaling factors A and B respectively, said neighborhoods being located at periodic horizontal and vertical intervals in respective horizontal and vertical directions therein, said horizontal and vertical intervals corresponding to said horizontal and vertical scaling factors A and B, respectively; and computing the value of each normalized pixel from a set of said original pixels lying within a corresponding one of said elliptical neighborhoods.

Claim 18

Dependent Claim 153 words

In a method for character normalization in optical character recognition in which the image of an individual character comprising an array of original pixels is normalized in accordance with horizontal and vertical scaling factors A and B to construct a normalized character image comprising an array of normalized pixels, the improvement comprising the steps of: defining in said array of original pixels a plurality of elliptical neighborhoods each having a minor axis and a major axis, said minor and major axes having lengths corresponding to said horizontal and vertical scaling factors A and B respectively, said neighborhoods being located at periodic horizontal and vertical intervals in respective horizontal and vertical directions therein, said horizontal and vertical intervals corresponding to said horizontal and vertical scaling factors A and B, respectively; and computing the value of each normalized pixel from a set of said original pixels lying within a corresponding one of said elliptical neighborhoods.

Claim 19

Dependent Claim 66 words

The method of claim 15 wherein said original and normalized pixel arrays represent bi-tonal images in which the value of each of said pixels is either "on" or "off", and wherein said computing step sets the value of each normalized pixel to "on" if any of the original pixels in the corresponding one of said elliptical neighborhoods is "on", and otherwise sets said value to "off".

Annuity Stage	Window Opens	Payment Deadline	Surcharge Period Until	Estimated Fee	Status
3.5 Years (Year 4)	1993-12-10	1994-06-11	1994-12-11	$2000.00	Window Closed / Lapsed
7.5 Years (Year 8)	1997-12-10	1998-06-11	1998-12-11	$3760.00	Window Closed / Lapsed
11.5 Years (Year 12)	2001-12-10	2002-06-11	2002-12-11	$7700.00	Window Closed / Lapsed

Character normalization using an elliptical sampling window for optical character recognition

Abstract

Inventors

Bibliographic Details

Claims Analysis

References Cited (Backward Citations)

USPTO Annuity & Maintenance Windows

Assigned Categories & Tags

Assign Category Tag