Texture discrimination method
First Claim
1. A character recognition apparatus for discriminating an image into characters and a textured background, comprising:
- a computer;
a memory coupled to the computer;
a scanner coupled to the computer and optically scanning the printed image and producing digital data for each of a plurality of pixels to provide a virtual image, and the digital data including pixel densities in a character portion of the virtual image containing the characters and the textured background;
said memory including an image memory storing the digital data as a record of the pixel densities in a matrix corresponding to locations of the pixels in the character portion;
said memory including a program memory storing a program;
said computer running the program and locating, in the virtual image, a learning area A containing only the textured background and a learning area B containing both the textured background and at least a part of the characters;
said memory storing a plurality of different discriminant functions that each include the digital pixel density information of a pixel being processed and pixels in the immediate adjacent vicinity of the pixel being processed;
said memory storing an identification of the learning area A and the learning area B;
said computer running the program and retrieving from said memory at least a part of the stored digital data as digital pixel density information by using the identification;
said computer running the program and determining a specific discriminant function by producing output values of the plurality of different discriminant functions that each include the digital pixel density information of a pixel being processed and pixels in the immediate adjacent vicinity of the pixel being processed separately for each of the learning areas A and B, and selecting one of the plurality of different discriminant functions as the specific discriminant function based upon the output values of the one of the plurality of different discriminate functions for learning area A and learning area B best satisfying a criteria when compared to the output values of the others of the plurality of different discriminant functions;
said memory storing, under control of the computer, accumulated statistics of the output values of the specific discriminant function for a plurality of different pixels being processed for at least one of the learning areas A and B;
said computer running the program and discriminating with the specific discriminant function whether each pixel, of the virtual image, is part of the characters or part of the textured background by comparing an output value of the specific discriminant function for the each pixel as the pixel being processed with the stored accumulated statistics of the output values of the specific discriminant function for a plurality of different pixels of at least one of the learning areas A and B, so that said discriminating discriminates that part of the digital data representing the characters from the virtual image containing the characters and textured background with respect to at least the learning area B;
said memory storing the part of the digital data;
said computer running the program, and retrieving from said memory and converting the .part of the digital data of the virtual image into digital encoded information providing codes for corresponding discrete ones of the characters;
wherein said specific discriminant function produces dispersion values for learning areas A and B that are substantially non-overlapping and mutually exclusive; and
wherein the specific discriminant function has a sum of its dispersion values less than or equal to a predetermined value as the criteria and includes iterative steps using Lagrange'"'"'s method of indeterminate coefficients.
1 Assignment
0 Petitions
Accused Products
Abstract
To reliably discriminate the characters only from the headlines having any background textures, the present invention analyzes the projection profile of a grey image in a headline area in order to automatically set a learning area A in the area of background texture only of the periphery in the headline area, and set a learning area B in the area that includes background texture of a central portion and characters; determines a discriminant function having, as variables, characteristics that use a plurality of pixel densities in the vicinity of a pixel which is processed so that output values of the discriminant function at each of the positions in the learning area A forms a profile with an average Va, that the output values at each of the positions in the learning area B forms a profile with an average Vb, and that the sum of dispersion values of the two profiles becomes smaller than a predetermined value, in order to discriminate in which area is included the pixel which is processed in the headline area; and determines in which area of the background pattern or the characters the pixel is included depending upon whether the output value of the discriminant function is close to the value Va or to the value Vb for each of the pixels in the headline area, in order to discriminate the headline area into areas.
-
Citations
6 Claims
-
1. A character recognition apparatus for discriminating an image into characters and a textured background, comprising:
-
a computer; a memory coupled to the computer; a scanner coupled to the computer and optically scanning the printed image and producing digital data for each of a plurality of pixels to provide a virtual image, and the digital data including pixel densities in a character portion of the virtual image containing the characters and the textured background; said memory including an image memory storing the digital data as a record of the pixel densities in a matrix corresponding to locations of the pixels in the character portion; said memory including a program memory storing a program; said computer running the program and locating, in the virtual image, a learning area A containing only the textured background and a learning area B containing both the textured background and at least a part of the characters; said memory storing a plurality of different discriminant functions that each include the digital pixel density information of a pixel being processed and pixels in the immediate adjacent vicinity of the pixel being processed; said memory storing an identification of the learning area A and the learning area B; said computer running the program and retrieving from said memory at least a part of the stored digital data as digital pixel density information by using the identification; said computer running the program and determining a specific discriminant function by producing output values of the plurality of different discriminant functions that each include the digital pixel density information of a pixel being processed and pixels in the immediate adjacent vicinity of the pixel being processed separately for each of the learning areas A and B, and selecting one of the plurality of different discriminant functions as the specific discriminant function based upon the output values of the one of the plurality of different discriminate functions for learning area A and learning area B best satisfying a criteria when compared to the output values of the others of the plurality of different discriminant functions; said memory storing, under control of the computer, accumulated statistics of the output values of the specific discriminant function for a plurality of different pixels being processed for at least one of the learning areas A and B; said computer running the program and discriminating with the specific discriminant function whether each pixel, of the virtual image, is part of the characters or part of the textured background by comparing an output value of the specific discriminant function for the each pixel as the pixel being processed with the stored accumulated statistics of the output values of the specific discriminant function for a plurality of different pixels of at least one of the learning areas A and B, so that said discriminating discriminates that part of the digital data representing the characters from the virtual image containing the characters and textured background with respect to at least the learning area B; said memory storing the part of the digital data; said computer running the program, and retrieving from said memory and converting the .part of the digital data of the virtual image into digital encoded information providing codes for corresponding discrete ones of the characters; wherein said specific discriminant function produces dispersion values for learning areas A and B that are substantially non-overlapping and mutually exclusive; and wherein the specific discriminant function has a sum of its dispersion values less than or equal to a predetermined value as the criteria and includes iterative steps using Lagrange'"'"'s method of indeterminate coefficients. - View Dependent Claims (3)
-
-
2. A character recognition apparatus for discriminating an image into characters and a textured background, comprising:
-
a computer; a memory coupled to the computer; a scanner coupled to the computer and optically scanning the printed image and producing digital data for each of a plurality of pixels to provide a virtual image, and the digital data including pixel densities in a character portion of the virtual image containing the characters and the textured background; said memory including an image memory storing the digital data as a record of the pixel densities in a matrix corresponding to locations of the pixels in the character portion; said memory including a program memory storing a program; said computer running the program and locating, in the virtual image, a learning area A containing only the textured background and a learning area B containing both the textured background and at least a part of the characters; said memory storing a plurality of different discriminant functions that each include the digital pixel density information of a pixel being processed and pixels in the immediate adjacent vicinity of the pixel being processed; said memory storing an identification of the learning area A and the learning area B; said computer running the program and retrieving from said memory at least a part of the stored digital data as digital pixel density information by using the identification; said computer running the program and determining a specific discriminant function by producing output values of the plurality of different discriminant functions that each include the digital pixel density information of a pixel being processed and pixels in the immediate adjacent vicinity of the pixel being processed separately for each of the learning areas A and B, and selecting one of the plurality of different discriminant functions as the specific discriminant function based upon the output values of the one of the plurality of different discriminate functions for learning area A and learning area B best satisfying a criteria when compared to the output values of the others of the plurality of different discriminant functions; said memory storing, under control of the computer, accumulated statistics of the output values of the specific discriminant function for a plurality of different pixels being processed for at least one of the learning areas A and B; said computer running the program and discriminating with the specific discriminant function whether each pixel, of the virtual image, is part of the characters or part of the textured background by comparing an output value of the specific discriminant function for the each pixel as the pixel being processed with the stored accumulated statistics of the output values of the specific discriminant function for a plurality of different pixels of at least one of the learning areas A and B, so that said discriminating discriminates that part of the digital data representing the characters from the virtual image containing the characters and textured background with respect to at least the learning area B; said memory storing the part of the digital data; said computer running the program, and retrieving from said memory and converting the part of the digital data of the virtual image into digital encoded information providing codes for corresponding discrete ones of the characters; wherein said specific discriminant function determines separate values corresponding to the average pixel density of learning areas A and B; and wherein the specific discriminant function has a sum of its dispersion values less than or equal to a predetermined value as the criteria and includes iterative steps using Lagrange'"'"'s method of indeterminate coefficients. - View Dependent Claims (4)
-
-
5. A character recognition apparatus for discriminating an image into characters and a textured background, comprising:
-
a computer; a memory coupled to the computer; a scanner coupled to the computer and optically scanning the printed image and producing digital data for each of a plurality of pixels to provide a virtual image, and the digital data including pixel densities in a character portion of the virtual image containing the characters and the textured background; said memory including an image memory storing the digital data as a record of the pixel densities in a matrix corresponding to locations of the pixels in the character portion; said memory including a program memory storing a program; said computer running the program and locating, in the virtual image, a learning area A containing only the textured background and a learning area B containing both the textured background and at least a part of the characters; said memory storing a plurality of different discriminant functions that each include the digital pixel density information of a pixel being processed and pixels in the immediate adjacent vicinity of the pixel being processed; said memory storing an identification of the learning area A and the learning area B; said computer running the program and retrieving from said memory at least a part of the stored digital data as digital pixel density information by using the identification; said computer running the program and determining a specific discriminant function by producing output values of the plurality of different discriminant functions that each include the digital pixel density information of a pixel being processed and pixels in the immediate adjacent vicinity of the pixel being processed separately for each of the learning areas A and B, and selecting one of the plurality of different discriminant functions as the specific discriminant function based upon the output values of the one of the plurality of different discriminate functions for learning area A and learning area B best satisfying a criteria when compared to the output values of the others of the plurality of different discriminant functions; said memory storing, under control of the computer, accumulated statistics of the output values of the specific discriminant function for a plurality of different pixels being processed for at least one of the learning areas A add B; said computer running the program and discriminating with the specific discriminant function whether each pixel, of the virtual image, is part of the characters or part of the textured background by comparing an output value of the specific discriminant function for the each pixel as the pixel being processed with the stored accumulated statistics of the output values of the specific discriminant function for a plurality of different pixels of at least one of the learning areas A and B, so that said discriminating discriminates that part of the digital data representing the characters from the virtual image containing the characters and textured background with respect to at least the learning area B; said memory storing the part of the digital data; said computer running the program, and retrieving from said memory and converting the part of the digital data of the virtual image into digital encoded information providing codes for corresponding discrete ones of the characters; wherein said program for determining includes the criteria that the output values of the specific discriminant function at each of the locations of the learning area A forms a profile with an average value Va, that the output values at each of the locations of the learning area B form a profile with an average value Vb, and that the sum of dispersion values of the two profiles becomes smaller than a predetermined value; and said program for discriminating including discriminating in which of the textured background or the characters each pixel is included based upon whether the output value of the specific discriminant function is close to the value Va or to the value Vb for each of the pixels.
-
-
6. A character recognition apparatus for discriminating an image into characters and a textured background, comprising:
-
a computer; a memory coupled to the computer; a scanner coupled to the computer and optically scanning the printed image and producing digital data for each of a plurality of pixels to provide a virtual image, and the digital data including pixel densities in a character portion of the virtual image containing the characters and the textured background; said memory including an image memory storing the digital data as a record of the pixel densities in a matrix corresponding to locations of the pixels in the character portion; said memory including a program memory storing a program; said computer running the program and locating, in the virtual image, a learning area A containing only the textured background and a learning area B containing both the textured background and at least a part of the characters; said memory storing a plurality of different discriminant functions that each include the digital pixel density information of a pixel being processed and pixels in the immediate adjacent vicinity of the pixel being processed; said memory storing an identification of the learning area A and the learning area B; said computer running the program and retrieving from said memory at least a part of the stored digital data as digital pixel density information by using the identification; said computer running the program and determining a specific discriminant function by producing output values of the plurality of different discriminant functions that each include the digital pixel density information of a pixel being processed and pixels in the immediate adjacent vicinity of the pixel being processed separately for each of the learning areas A and B, and selecting one of the plurality of different discriminant functions as the specific discriminant function based upon the output values of the one of the plurality of different discriminate functions for learning area A and learning area B best satisfying a criteria when compared to the output values of the others of the plurality of different discriminant functions; said memory storing, under control of the computer, accumulated statistics of the output values of the specific discriminant function for a plurality of different pixels being processed for at least one of the learning areas A and B; said computer running the program and discriminating with the specific discriminant function whether each pixel, of the virtual image, is part of the characters or part of the textured background by comparing an output value of the specific discriminant function for the each pixel as the pixel being processed with the stored accumulated statistics of the output values of the specific discriminant function for a plurality of different pixels of at least one of the learning areas A and B, so that said discriminating discriminates that part of the digital data representing the characters from the virtual image containing the characters and textured background with respect to at least the learning area B; said memory storing the part of the digital data; said computer running the program, and retrieving from said memory and converting the part of the digital data of the virtual image into digital encoded information providing codes for corresponding discrete ones of the characters; wherein said program for determining a specific discriminant function includes determining a polynomial discriminant function f'"'"' having, as variables, character quantities that use a plurality of pixel densities in a predetermined vicinity area of a pixel being processed, such that output values of the polynomial discriminant function f'"'"' at each of the positions of the learning area A of the image form a profile with an average value Va, that the output values of the polynomial discriminant function f'"'"' at each of the positions of the learning area B form a profile with an average value Vb, and that the sum of dispersion values of these profiles becomes a minimum; said program for determining a discriminant function f'"'"' further includes; finding an average density in the subareas r1, . . . rk included in the vicinity of the pixel being processed as character quantities c1, . . . ck, finding, by Lagrange'"'"'s method of indeterminate coefficient, unknown coefficients a0, a1, a2 and a3 of a 0-th order basic discriminant function f'"'"'0ij (x, y) expressed by the following equation (3) such that the sum Sij of the following equation (1) becomes a minimum for each combination (ci, cj) of the character quantities and that the following equation (2) holds true, ##EQU3## where Sa and Sb denote areas of the learning areas A and B;
advances to reconstituting, below, when the minimum value Sij is smaller than S (a fixed value) and advances to the next step in other cases,selects k (an integer) 0-th order basic discriminant functions f'"'"'0ij (x, y) based on combinations ((io, io), (i1, j1), . . . ) of character quantities of small Sij values, and sends the outputs as input images to the following finding, finding, by Lagrange'"'"'s method of indeterminate coefficient, unknown coefficients am0, am1, am2 and am3 (wherein m=n+1) of an m-th order basic discriminant function (f'"'"'mij(x, y) expressed by the following equation (6) for each combination (f'"'"'ni(x, y) f'"'"'nj(x, y)) of the input images such that Smij of the following equation (4) becomes a minimum and that the following equation (5) holds true, ##EQU4## where i and j together are denoted by q (but (io,jo) when q=1, (i1 j1) when q=2, . . . , 1<
/=q<
/=k), the input image is denoted by f'"'"'nq(x, y) (where 0<
/=n), and Sa and Sb denote areas of the learning areas A and B,advances to the final reconstituting when the minimum value Smij is smaller than or equal to S or when the minimum value Smij is greater than or equal to the minimum value Snij, and advances to the following selecting in other cases, selecting m-th order basic discriminant functions f'"'"'mij (x, y) having k'"'"' small values Smij, and sending the outputs as input images f'"'"'mq(x,y)(where q-k'"'"'), and when the minimum value Smij or Sij is smaller than S, reconstituting a discriminant function f'"'"' that is to be found by reversely seeking a master basic discriminant function from the m-th order basic discriminant function f'"'"'mij (x,y) having the minimum value, and which, when the minimum value Smij is greater than the minimum value Snij, reconstituting a discriminant function f'"'"' that is to be found by reversely seeking a master basic discriminant function from the n-th order basic discriminant function f'"'"'nij(x, y) having the minimum value.
-
Specification