JP2001312696A - Document sorting device, character reader, genuineness judging device and method for these - Google Patents

Abstract

PROBLEM TO BE SOLVED: To speedily and precisely sort a document by using the fast and precise collation of a character string. SOLUTION: Picture data of a slip including a character string is fetched from a picture input device 1 and stored in a picture storing part 21. A slip position detecting means 31 detects the position of the slip in the stored picture data. In accordance with this detecting result, a coordinate transformation means 32 transforms the coordinate system of coordinates where the character string has to be positioned in the slip stored in a coordinate storing area 22a to the picture coordinate system being the coordinate system of the picture data. The deviation of the coordinates of the character in the picture data stored in the part 21 is corrected by the unit of the character string by a character string coordinate value correcting means 33 and by the unit of the character by a character coordinate value correcting means 34. A graphic collation means 35 extracts the feature quantity of each character in picture data whose coordinates are corrected and collates it with a collating featured value stored in a collating feature storing area 22b. In accordance with this collating result, an output device 9 outputs the code information on each character stored in a code information storing area 81a.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a document classifying apparatus utilizing character string collation, a character reading apparatus, an authenticity judging apparatus, and a method thereof.

[0002]

2. Description of the Related Art In recent years, computerization of information processing has been advanced. For example, an image of a form is read by an image scanner, and data obtained by converting a character string in the read image into code information is input to a database or the like. Is being done. In this case, in order to correctly input data, the computer needs to know the structure of the form, and it is necessary to determine whether or not the form is in a predetermined format and to classify the form for each form.

As a method of classifying such forms, a method of collating a character string pattern such as a form name described at a specific position on a read form with a previously stored pattern (such as a template) is known. is there. However, a character string on a form cannot be cut out accurately because, in general, misalignment occurs due to printing misalignment or distortion during reading, and local fluctuation in image density occurs during reading. It is difficult to determine the presence or absence of a character string even if the pattern of the character string cut out incorrectly as described above is compared with a pattern stored in advance, and the form cannot be classified.

Therefore, there is the following conventional character string collation method (hereinafter referred to as Conventional Example 1). In this method, a threshold value for separating a character portion from a background portion is obtained based on density information of an image, and the image is divided into two by this threshold value.
A connected component in which pixels having the same gradation are connected by a value is detected. Then, a graphic component in a character unit obtained by surrounding a connected component having a short physical distance with a rectangle is detected, and the detected graphic component in a character unit is collated with a character template registered in advance.

In the technique disclosed in Japanese Patent Application Laid-Open No. 7-239912 (hereinafter referred to as Conventional Example 2), the position of a template, which is a sample image, is determined within a region near each character in a read image. While changing, a cross-correlation value between the character and the template is calculated at each position, and character matching is performed according to the cross-correlation value.

[0006]

However, in the above-mentioned prior art example 1, when the density of the background portion changes in one character region, an appropriate threshold value is set for separating the character portion from the background portion. There was a problem that it was not possible. That is, no matter which threshold is set, a part of a character may be mistaken for a background part or a part of a background may be mistaken for a character part. Could not match template correctly.

Further, in the above-mentioned conventional example 2, every time the position of the template is changed, it must be compared with the cross-correlation value at the previous position. For this reason, there has been a problem that the required operation cost is extremely large, and it is difficult to perform character string collation at high speed.

The present invention has been made in order to solve the above-described problems of the conventional example, and utilizes a character string collation for correcting a print shift or a position shift due to a distortion generated at the time of image input with high speed and high accuracy. Accordingly, an object of the present invention is to provide a document classification device and method for quickly and accurately classifying documents such as forms.

Another object of the present invention is to provide a character reading apparatus and method for converting a character string in a document into code information at high speed and accurately by utilizing the above character string collation.

A further object of the present invention is to provide an authentication apparatus and method for quickly and accurately determining the authenticity of securities or the like by utilizing the above-described character string collation.

[0011]

In order to achieve the above object, a document classifying apparatus according to a first aspect of the present invention stores multi-gradation image data taken from a document in a predetermined format including a character string. Image storage means, coordinates of the position of each character of the character string in the document, feature values that the character string should have, and identification information for identifying the format of the document containing the character string. Dictionary storage means for adding and storing; a position detection means for detecting the position of the document in the image data stored in the image storage means; and a character string in the document according to a detection result of the position detection means. Coordinate conversion means for converting a coordinate system of coordinates where each character in the document should be located from a document coordinate system based on the document to an image coordinate system based on a state of capturing the image data, and stored in the image storage means. Image data Character string coordinate correction means for correcting a deviation between the coordinates of the entire character string included in the document and the coordinates at which the character string should be located in the document whose coordinate system has been converted by the coordinate conversion means in units of character strings; and The coordinates of each character included in the character string whose coordinates have been corrected in units of the character string by the column coordinate correction means and the coordinates where each character of the character string whose coordinate system has been converted by the coordinate conversion means should be located. A character coordinate correcting means for correcting the displacement in units of characters, and a characteristic amount of the character string in the image data taken from the document in which the coordinates of each character are corrected by the character coordinate correcting means, and the extracted characteristic amount Figure matching means for comparing the character string stored in the dictionary storage means with a feature amount to be possessed,
Output means for outputting the identification information stored in the dictionary storage means in accordance with the result of matching by the graphic matching means.

In order to achieve the above object, a character reading apparatus according to a second aspect of the present invention comprises: an image storage means for storing multi-gradation image data taken from a document in a predetermined format including a character string; A dictionary storage means for storing coordinates in which each character of the character string should be located in the document, a feature quantity to be included in the character string, and a character code corresponding to the character string, and the image storage means Position detecting means for detecting the position of the document in the image data stored in the image data, and a coordinate system of coordinates where each character in a character string should be located in the document according to the detection result of the position detecting means. Coordinate conversion means for converting from a document coordinate system based on the document to an image coordinate system based on the state of capture of the image data,
A deviation between the coordinates of the entire character string included in the image data stored in the image storage means and the coordinates where the character string should be located in the document whose coordinate system has been converted by the coordinate conversion means is corrected in character string units. Character string coordinate correcting means, and the coordinates of each character included in the character string whose coordinates have been corrected in units of the character string by the character string coordinate correcting means, and the character string whose coordinate system has been converted by the coordinate converting means. Character coordinate correcting means for correcting a deviation from the coordinates where each character should be located in character units, and a character string in image data taken from the document in which the coordinates of each character have been corrected by the character coordinate correcting means. A graphic matching means for extracting a feature quantity and matching the extracted feature quantity with a feature quantity to be possessed by the character string stored in the dictionary storage means; and And output means for outputting the character code stored in 憶 means, characterized in that it comprises a.

[0013] To achieve the above object, an authenticity determination device according to a third aspect of the present invention comprises image storage means for storing multi-tone image data taken from a document in a predetermined format including a character string, A dictionary storage unit that stores coordinates in which the characters of the character string should be located in the document, a feature amount that the character string should have, and an allowable range of the feature amount in association with each other; Position detecting means for detecting the position of the document in the image data being processed, and a coordinate system of coordinates where each character in a character string should be located in the document according to the detection result of the position detecting means. Coordinate conversion means for converting from a document coordinate system based on the image data to an image coordinate system based on the state of capture of the image data; and coordinates of the entire character string included in the image data stored in the image storage means and the coordinates Conversion means A character string coordinate correcting unit for correcting a deviation from a coordinate at which a character string should be located in the document whose coordinate system has been converted, in units of character strings; and Character coordinate correcting means for correcting a deviation between the coordinates of each character included in the corrected character string and the coordinates at which each character of the character string whose coordinate system has been converted by the coordinate converting means should be located in character units; Extracting a characteristic amount of a character string in image data taken from the document in which the coordinates of each character are corrected by the character coordinate correcting unit, and extracting the extracted characteristic amount from the character stored in the dictionary storage unit. Graphic matching means for matching with a feature quantity that the column should have;
Authenticity judging means for judging whether or not the difference in the feature amount collated by the graphic collating means is within a range of an allowable range stored in the dictionary storage means.

In the document classification device, character reading device and authenticity judging device according to the first to third aspects, document classification, character reading, and character reading are performed using character string collation.
Alternatively, the authenticity of a security or the like can be determined quickly and accurately.

In order to achieve the above object, a document classification method according to a fourth aspect of the present invention is a document classification method, comprising: a coordinate at which each character of a character string should be located in a document; A method for classifying a document in a computer system having a dictionary in which identification information for identifying a format of a document including a character string is stored in association with the identification information, wherein a multi-level document read from a document in a predetermined format including a character string is provided. Storing the key image data in a memory; detecting the position of the document in the image data stored in the memory; and detecting each position of the character string in the document according to the detected position of the document. Converting the coordinate system of the coordinates where the characters should be located from the document coordinate system based on the document to the image coordinate system based on the state of capture of the image data; and Correcting the deviation between the coordinates of the entire character string included in the image data and the coordinates where the character string should be located in the document converted into the image coordinate system in units of character strings, and correcting the coordinates in units of character strings. Correcting in each character a deviation between the coordinates of each character included in the character string and the coordinates where each character of the character string should be located in the document converted into the image coordinate system; and Extracting a characteristic amount of a character string in the image data taken in from the corrected document, and comparing the extracted characteristic amount with a characteristic amount to be included in the character string stored in the dictionary; Outputting the identification information stored in the dictionary according to the result.

In order to achieve the above object, a document reading method according to a fifth aspect of the present invention provides a document reading method, comprising the steps of: A method for reading characters in a computer system having a dictionary storing character codes corresponding to character strings in association with each other, wherein multi-tone image data taken from a document in a predetermined format including the character strings is stored in a memory. Causing the document to be located in the image data stored in the memory; and, according to the detected location of the document, the coordinates of the coordinates where each character in the character string should be located in the document. Converting the system from a document coordinate system based on the document to an image coordinate system based on the state of capturing the image data; Correcting the deviation between the coordinates of the entire character string to be converted and the coordinates where the character string should be located in the document converted into the image coordinate system in units of character strings; and Correcting the deviation between the coordinates of each character included in the document and the coordinates where each character of the character string should be located in the document converted into the image coordinate system on a character-by-character basis; and Extracting a feature amount of a character string in the image data captured from the document, comparing the extracted feature amount with a feature amount to be possessed by the character string stored in the dictionary, and according to the matching result, Outputting the character code stored in the dictionary.

In order to achieve the above object, an authenticity judgment method according to a sixth aspect of the present invention comprises: a coordinate where each character of a character string should be located in the document; A method for judging the authenticity of a document in a computer system having a dictionary in which the allowable range of the feature amount is stored in association with the dictionary, wherein multi-tone image data taken from a document in a predetermined format including a character string is stored in a memory. And detecting the position of the document in the image data stored in the memory, and according to the detected position of the document, a character string in the document stored in the memory in advance. Converting a coordinate system of coordinates where each character should be located from a document coordinate system based on the document to an image coordinate system based on a state of capturing the image data;
Correcting the deviation between the coordinates of the entire character string included in the image data stored in the memory and the coordinates where the character string should be located in the document converted into the image coordinate system in units of character strings; Correcting the difference between the coordinates of each character included in the character string whose coordinates have been corrected in units and the coordinates where each character of the character string should be located in the document converted into the image coordinate system in units of characters; Extracting a feature amount of a character string in image data taken from the document in which the coordinates of each character are corrected, and comparing the extracted feature amount with a feature amount that the character string should have in the dictionary stored in the dictionary And a step of determining whether or not the difference between the matched feature amounts falls within the allowable range stored in the dictionary.

[0018]

Embodiments of the present invention will be described below with reference to the accompanying drawings. In this embodiment, an example will be described in which a character string on a form is read and the form is applied to a form classification apparatus that classifies the form based on the read character string.

FIG. 1 is a functional block diagram showing a functional configuration of a form classification device according to this embodiment. As shown in the figure, the form classification device includes an image input device 1, a storage device 2, a data processing device 3, and an output device 4.

The image input device 1 is constituted by an image scanner or the like, reads an image of a form, and takes it in as a two-dimensional digital image having multiple gradations. When the image of the form is captured by the image input device 1, a tilt of about ± 5 ° is allowed. Further, in this embodiment, it is assumed that the image input device 1 only needs to be able to read at least the image of the area in the form where the form name is described.

The storage device 2 is constituted by a semiconductor memory or a magnetic disk or the like, and includes an image storage unit 21 and a dictionary storage unit 22. The image storage unit 21 stores the image input device 1
The digital image read from the form is stored. The digital image stored in the image storage unit 21 has an origin at the upper left corner of the image, and a coordinate system (hereinafter, referred to as an image coordinate system) including an x-axis and a y-axis taken in the horizontal and vertical directions from the origin, respectively. Refers to the position of each pixel.

The dictionary storage section 22 stores data for character string collation, and includes a coordinate storage area 22a, a collation characteristic storage area 22b, and a classification storage area 22c. The coordinate storage area 22a stores the coordinates of the collation character string on the form. Here, the coordinates of the collation character string are represented by a coordinate system in which the upper left corner of the form is the origin, the upper side is the u axis, and the left side is the v axis (hereinafter referred to as a form coordinate system).

In the collation feature storage area 22b, each character / graphic region of the collation character string is divided into lattice-like partial images, and the feature amount obtained by averaging the gradation values of each partial image is stored for each character. It is remembered. The classification storage area 22c stores identification information for identifying a form to be classified. The coordinates of the collation character string stored in the coordinate storage area 22a, the feature amounts stored in the collation characteristic storage area 22b, and the identification information stored in the classification storage area 22c are associated, for example, by a pointer. It is stored and remembered.

The data processing device 3 includes a memory for storing a program and a CPU (Central Pr
and a function realized by the CPU executing the program, such as a form position detecting means 31, a coordinate converting means 32, a character string coordinate value correcting means 33, a character coordinate value correcting means 34, and a figure collating means. 35
including.

The form position detecting means 31 is provided in the image storage unit 21.
, And the position of the form in the read image is detected. The coordinate conversion means 32 receives the information on the position of the form detected by the form position detection means 31, and according to the received information on the position of the form, the coordinates on the form of the collation character string stored in the coordinate storage area 22a. The system is converted from the form coordinate system to the image coordinate system.

The character string coordinate value correcting means 33 mainly corrects the positional deviation of the character string such as printing deviation, and the coordinate value of the collating character string whose coordinates have been converted by the coordinate converting means 32 and the image input means. The deviation of the coordinate value of the character string on the read form is corrected for each character string using the pixel projection profile.

The character coordinate value correcting means 34 corrects the positional deviation of each character mainly caused by the warpage of the form or the like, and further corrects the positional deviation of the coordinate value for each character by using a pixel projection profile. I do.

The figure collating means 35 cuts out the character string whose coordinate values have been corrected by the character string coordinate value correcting means 33 and the character coordinate value correcting means 34 into a partial image, extracts feature amounts, and compares the extracted feature amounts. Is compared with the light and dark feature stored in the feature storage area 22b. The graphic matching unit 35 outputs the matching result to the output device 4. The functions of the units 31 to 35 included in the data processing device 3 will be described later in more detail.

The output device 4 is composed of, for example, a display device, and outputs an identification code indicating the classification of the form stored in the classification storage area 22c based on the result of the comparison by the graphic matching means 35. When the form read by the image input apparatus 1 is determined to be not a form belonging to a class stored in advance based on the collation result of the graphic collating unit 35, the output device 4 outputs a statement to that effect.

Hereinafter, each function realized in the data processing device 3 will be described in detail.

The form position detecting means 31 is provided in the image storage unit 21.
Is read out, and the image is first scanned in order from the upper end to the lower end in the x-axis direction to detect a point on the scanning line at which the gradation greatly changes. Two or more points at which the gradation greatly changes by such scanning are detected, and the detected point group is subjected to regression analysis to obtain a straight line L1 on the upper side of the form.

Next, the form position detecting means 31 sequentially scans the image in the y-axis direction from the left end to the right end, and detects a point on the scanning line at which the gradation greatly changes. Two or more points where the gradation changes greatly by such scanning are detected, and the detected point group is subjected to regression analysis to obtain a straight line L2 on the left side of the form.
The form position detecting means 31 calculates a straight line L between the obtained upper side and the left side.
1. The intersection (x0, y0) of L2 is determined and detected as the position of the point at the upper left corner of the form.

The form position detecting means 31 further includes a line L
The size of the angle formed by 1 and the x-axis, the size of the angle formed by the y-axis of the straight line L2, or the average thereof is determined as the rotation angle θ0 of the form.

The coordinate conversion means 32 reads the coordinate value of each character of the collation character string stored in the coordinate storage area 22a of the dictionary storage unit 22 in the form coordinate system, and converts the read coordinate value in the form coordinate system into Equation 1 is calculated and converted into coordinate values in the image coordinate system.

[0035]

X = ucosθ0 + vsinθ0 + x0 y = −usinθ0 + vcosθ0 + y0 where u and v are coordinate values in the form coordinate system, and x and y are coordinate values in the image coordinate system.

When the character string stored in the coordinate storage unit 22a is composed of I characters, the character string coordinate value correcting means 33 converts the coordinate of the upper left corner of the circumscribed rectangle of the i-th character in the character string after the coordinate conversion. The coordinate values are (xs (i), ys (i)), and the height and width are H (i) and W (i), respectively. Also, the coordinates of the upper left corner of the character included in the image are represented by (xs2 (i), ys
2 (i)), the height and width are H2 (i) and W2, respectively.
(I).

Here, the offset sentences uniformly occurring in the entire character string are represented by Δx, Δy, ΔH, and the remaining offset sentences are dx
Assuming that (i), dy (i), dH (i), dW (i), xs2 (i), ys2 (i), H2 (i), W2
(I) can be represented by Equation 2, respectively.

[0038]

Xs2 (i) = xs (i) + Δx + dx (i) ys2 (i) = ys (i) + Δy + dy (i) H2 (i) = H (i) + ΔH + dH (i) W2 (i) = W ( i) + dW (i)

The character string coordinate value correcting means 33 outputs x
A rectangular area B (i) is obtained by enlarging a circumscribed rectangle of a character represented by four data of s (i), ys (i), H (i) and W (i) by L around the character. Character string coordinate value correction means 33
Aligns the y-coordinates so that ys (i), which is the y-coordinate value at the upper end of each character, matches, and x in this area B (i)
By performing pixel projection in the axial direction (to add a gradation value for each pixel; the same applies hereinafter), a horizontal pixel projection profile PY (i, y) of each of the I characters is generated.

Next, the character string coordinate value correcting means 33 adds the horizontal pixel projection profiles PY (i, y) of all I characters in the character string for each y coordinate, and sets the character string as a unit. Horizontal pixel projection profile PYA
(Y) is obtained.

The character string coordinate value correcting means 33 calculates the horizontal pixel projection profile PYA (y)
Using Equation 3 below, Δy = [− TY, + TY], Δ
Find a set of values of Δy and ΔH that minimizes the evaluation value VY in the range of H = [− TH, + TH].

VY = PYA (Δy) + PYA (Δy + H + ΔH) where TY is a value that is about half the average value of the heights of the characters in the character string, and TH is a minute value and is set in advance. ing.

Next, the character string coordinate value correcting means 33 projects pixels in the area B (i) also in the y-axis direction so that xs (i), which is the x coordinate value at the left end of each character, matches. , And generate a vertical pixel projection profile PX (i, x) for each of the I characters. Next, the character string coordinate value correcting unit 33 adds the vertical pixel projection profiles PX (i, x) of all the I characters in the character string for each x coordinate, and calculates the vertical direction using the character string as a unit. A pixel projection profile PXA (x) is obtained.

The character string coordinate value correcting means 33 calculates the vertical pixel projection profile PXA (x)
Using Expression 4 below, find Δx that maximizes the evaluation value VX in the range of Δx = [− TX, + TX].

VX = PXA ([Delta] x) -PXA ([Delta] x-1) where TX is a value that is about half the width of a character in a character string.

The character string coordinate value correcting means 33 uses the values of Δx, Δy, and ΔH obtained as described above, for example,
A character string shift on a form due to a print shift or the like is corrected.

The character coordinate value correcting means 34 calculates the I
The following processing is sequentially performed for each of the characters. In the following description, it is assumed that the character to be processed is the i-th character in the character string.

The character coordinate value correcting means 34 calculates dy (i) = [− TY2, + TY from the horizontal pixel projection profile PY (i, y) for each character using the following equation (5).
2], dH (i) = V in the range of [−TH2, + TH2].
Find a set of dy (i) and dH (i) that minimizes Y2.

[0047]

VY2 = PY (i, Δy + dy (i)) + PY
(I, Δy + H (i) + ΔH + dy (i) + dH
(I)) However, TY2 and TH2 are values given in advance, for example, TY = 5 and TH = 5.

The character coordinate value correcting means 34 calculates dx (i) = [− TX2, + TX from the vertical pixel projection profile PX (i, y) of each character using the following equation (6).
2], dW (i) = V in the range of [−TW2, + TW2].
Find the set of dx (i) and dW (i) that minimizes X2.

[0049]

VX2 = PX (i, Δx + dx (i)) + PX
(I, Δx + dx (i) + W (i) + dW (i)) where TX2 and TW2 are predetermined values, for example, TX = 5 and TW = 5.

The figure collating means 35 cuts out each character according to the circumscribed rectangular area of each character of the character string in the image whose coordinate value has been corrected by the character string coordinate value correcting means 33 and the character coordinate value correcting means 34. The graphic matching unit 35 divides each cut-out character into partial images and extracts a feature amount of each character.

The graphic collating means 35 further collates the extracted characteristic amount of each character with the collating characteristic amount of each character stored in the collating characteristic amount storage area 22b according to the following equation (7).
Each similarity s (i) is obtained.

S (i) = (VecX (i) · VecY
(I)) / | VecX (i) || VecY (i) |

The figure matching means 35 further calculates an average value sm of the obtained similarities s (i), and outputs this to the output device 4 as a result of the comparison.

Hereinafter, the operation of the form classification device according to this embodiment will be described with reference to the flowcharts of FIGS.

In order to facilitate understanding of the operation, a specific example will be described with reference to FIGS. Here, the forms to be classified include “out”, “gold”,
It is assumed that a character string composed of four characters of “den” and “vote” is described, and that coordinates of a form coordinate system in a standard form (a form for collation) are stored in the coordinate storage area 22a. I do. For the coordinates of each character, the coordinate value of the upper left corner of the circumscribed rectangle is (u
s (i), vs (i)), height H (i), width W
(I) (where i = 1, 2, 3, 4). In addition, it is assumed that the feature amounts of these characters are stored in the matching feature storage area 22b.

In the processing of the flowchart of FIG. 2, first, the image input device 1 takes in a form image as a two-dimensional digital image having a gradation. Image input device 1
Causes the captured digital image to be stored in the image storage unit 21 of the storage device 2 (step S1). The image stored in the image storage unit 21 is shown in FIG.

When the input of the image is completed, the form position detecting means 31 reads the image stored in the image storage section 21 and detects the upper side and the left side of the form in the read image. The form position detection means 31 calculates the coordinate value (x0, y0) of the upper left corner of the form from the detected top and left sides of the form.
And the inclination angle θ0 (step S2). The obtained coordinate value (x0, y0) and the inclination angle θ0 are supplied to the coordinate conversion means 32.

The coordinate conversion means 32 reads information on the coordinates of the circumscribed rectangle of each character of the character string stored in the coordinate storage area 22a of the dictionary storage unit 22 (step S3).
The coordinate conversion means 32 calculates the coordinate value (us (i), vs (i)) of the upper left corner of the circumscribed rectangle of each character in the read information.
Is calculated according to the above equation (1) using the coordinate values (xs
(I), ys (i)) (step S4). FIG. 5 shows these circumscribed rectangles by alternate long and short dash lines. The coordinates of the circumscribed rectangle converted to the image coordinate system are supplied to the character string correction unit 33.

Next, the character string coordinate value correcting means 33 performs a process of position adjustment in character string units (correction of coordinate shift in character string units) (step S5). FIG. 3 is a flowchart showing the details of the alignment process in units of character strings in step S5.

First, the character string coordinate value correcting means 33 expands the circumscribed rectangle of the collation character string coordinate-converted by the coordinate converting means 32 by L in the upper, lower, left and right directions, respectively.
Find (i). These partial areas B (i) are shown by broken lines in FIG.

The character string coordinate value correcting means 33 projects pixels in the x-axis direction for each of these partial areas B (i), and outputs a horizontal pixel projection profile PY for each character.
(I) is generated (step S501). here,
Partial area B of each character of "de", "gold", "den", "vote"
FIGS. 6A to 6D show (i), and FIGS. 6E to 6H show the corresponding horizontal pixel projection profiles.

Next, the character string coordinate value correcting means 33 superimposes the horizontal pixel projection profile PY (i, y) for each character obtained in step S501 so that the y coordinate value at the upper end of each character matches. In addition, a horizontal pixel projection profile PYA (y) using a character string as a unit is generated (step S502). FIG. 6I shows an example of the horizontal pixel projection profile PYA (y) using this character string as a unit.

Next, the character string coordinate value correcting means 33 initializes the variable VYmin with MAXINT, which is a value that is too large to take the evaluation value VY obtained by the above equation (step S503). The character string coordinate value correcting means 33
Further, Δy2 is initialized by a constant −TY (step S5).
04), ΔH is initialized by a constant −TH (step S5)
05).

The character string coordinate value correcting means 33 calculates the evaluation value VY by performing the operation of the above equation (step S).
506). Next, the character string coordinate value correcting unit 33 determines whether the evaluation value VY calculated in step S506 is smaller than the variable VYmin (step S507).

In step S507, the evaluation value VY is set to the variable VY.
When it is determined that it is not smaller than min, the character string coordinate value correcting means 33 proceeds to the processing of step S509 as it is. In step S507, the evaluation value VY is set to the variable VYmi.
When it is determined that the value is smaller than n, the character string coordinate value correcting unit 33 sets the value of the evaluation value VY to the variable VYmin,
The value of ΔH is substituted for ΔH2, and the value of ΔH2 is substituted for ΔH (step S508). Then, the character string coordinate value correcting unit 33 proceeds to the process of step S509.

In step S509, the character string coordinate value correcting means 33 increments the value of ΔH2 by “1”. Then, the character string coordinate value correcting means 33 determines whether or not the value of ΔH2 is larger than a preset constant TH (step S509).

In step S509, the value of ΔH2 is a constant TH
When it is determined that it is not larger than the above, the character string coordinate value correcting means 33 executes the processing of step S506 again.

In step S509, the value of ΔH2 is a constant TH
If it is determined that the value is larger than the value, the character string coordinate value correcting unit 33 increments the value of Δy2 by “1” (step S511). Then, the character string coordinate value correcting means 33 determines whether the value of Δy2 is larger than the constant TY (step S512).

In step S512, the value of Δy2 is a constant TY
When it is determined that it is not larger than the above, the character string coordinate value correcting means 33 executes the processing of step S505 again. If it is determined in step S512 that the value of Δy2 is larger than the constant TY, the character string coordinate value correcting unit 33
Restricts the value of the y coordinate in each partial area B (i) to a range from Δy to (Δy + H + ΔH), and performs pixel projection in the y-axis direction, and a vertical pixel projection profile PX for each character.
(I, x) is generated (step S513). here,
Partial area B of each character of "de", "gold", "den", "vote"
FIGS. 7A to 7D show (i), and FIGS. 7E to 7H show the corresponding vertical pixel projection profiles.

Next, the character string coordinate value correcting means 33 superimposes the horizontal pixel projection profile PY (i, y) for each character obtained in step S513 so that the x coordinate value at the left end of each character matches. In addition, a vertical pixel projection profile PXA (y) using a character string as a unit is generated (step S514). FIG. 7I shows an example of the vertical pixel projection profile PXA (x) using the character string as a unit.

Next, the character string coordinate value correcting means 33 initializes the variable VXmax with the value "0" (step S51).
5) Then, Δx2 is initialized with a constant −TX (step S516).

The character string coordinate value correcting means 33 calculates the evaluation value VX by performing the operation of the above equation (4) (step S).
517). Next, the character string coordinate value correcting unit 33 determines whether the evaluation value VX calculated in step S517 is larger than the variable VXmax (step S518).

At step S518, the evaluation value VX is
If it is determined that the value is not larger than max, the character string coordinate value correcting unit 33 proceeds to the process of step S520. In step S518, the evaluation value VX is set to the variable VXma.
When it is determined that the value is larger than the value of x, the character string coordinate value correcting unit 33 sets the value of the evaluation value VX to a variable VXmax,
Substitute the value of Δx2 into Δx (Step S51)
9). Then, the character string coordinate value correcting unit 33 proceeds to the process of step S520.

In step S520, the character string coordinate value correcting means 33 increments the value of Δx2 by “1”. Then, the character string coordinate value correcting means 33 determines whether the value of Δx2 is larger than the constant TX (step S521).

In step S521, the value of Δx2 is a constant TX
When it is determined that it is not larger than the above, the character string coordinate value correcting means 33 executes the processing of step S517 again. If it is determined in step S521 that the value of Δx2 is larger than the constant TX, the character string coordinate value correcting unit 33
Ends the processing of this flowchart, that is, the processing of step S5 in FIG.

The character string coordinate value correcting means 33 determines in step S
When the process of Step 5 (the process of the flowchart of FIG. 3) is completed, the character coordinate value correcting unit 34 performs a process of character unit alignment (correction of coordinate shift in character units) (Step S6).

FIG. 4 is a flow chart showing the details of the alignment process for each character in step S6. The process of this flowchart is executed for each character,
Since everything is performed by the same process,
The processing content for only one character will be described.

First, the character coordinate value correcting means 34 initializes the variable VY2min with MAXINT, which is a value that is too large for the evaluation value VY2 obtained by the above equation (5) (step S601). The character coordinate value correction means 34
Further, dy2 is initialized by a constant -TY2 (step S
602), dH2 is initialized with a constant -TH2 (step S603).

The character coordinate value correcting means 34 calculates the above equation (5).
To calculate the evaluation value VY2 (step S
604). Next, the character coordinate value correcting unit 34 determines whether the evaluation value VY2 calculated in step S604 is smaller than the variable VY2min (step S605).

In step S605, the evaluation value VY2 is
If it is determined that it is not smaller than Y2min, the character coordinate value correcting means 34 proceeds to the processing of step S607 as it is. In step S605, the evaluation value VY2 is set to the variable VY.
When it is determined that the value is smaller than 2 min, the character coordinate value correcting unit 34 sets the value of the evaluation value VY2 to the variable VY2min.
, The value of dy2 is dy (i), and the value of dH2 is dH
(I) are substituted (step S606). Then, the character coordinate value correcting unit 34 proceeds to the process of step S607.

In step S607, the character coordinate value correcting means 34 increments the value of dH2 by "1". Then, the character coordinate value correction unit 34 determines whether the value of dH2 is larger than a preset constant TH2 (step S608).

In step S608, the value of dH2 is a constant TH
If it is determined that it is not larger than 2, the character coordinate value correcting means 34 executes the processing of step S604 again.

In step S608, the value of dH2 is a constant TH
When it is determined that the value is larger than 2, the character coordinate value correcting unit 34 increments the value of dy2 by “1” (step S609). Then, the character coordinate value correcting means 3
No. 4 determines whether the value of dy2 is larger than the constant TY2 (step S610).

In step S610, the value of dy2 is a constant TY
When it is determined that it is not larger than 2, the character coordinate value correcting means 34 executes the processing of step S602 again.

In step S610, the value of dy2 is a constant TY
If it is determined that the evaluation value VX2 is larger than 2, the character coordinate value correction unit 34 sets the variable VX2
min is initialized (step S611). The character coordinate value correcting unit 34 further initializes dx2 with a constant -TX2 (step S612), and initializes dW2 with a constant -TW2 (step S613).

The character coordinate value correcting means 34 calculates the above equation (6).
To calculate the evaluation value VX2 (step S
614). Next, the character coordinate value correcting unit 34 determines whether or not the evaluation value VX2 calculated in step S614 is smaller than the variable VX2min (step S615).

In step S615, the evaluation value VX2 is
If it is determined that it is not smaller than X2min, the character coordinate value correcting means 34 proceeds to the processing of step S617 as it is. In step S615, the evaluation value VX2 is set to the variable YX.
When it is determined that the value is smaller than 2 min, the character coordinate value correcting unit 34 sets the value of the evaluation value VX2 to the variable VX2min.
, The value of dx2 to dx (i) and the value of dW2 to dW
(I) is assigned to each (step S616). Then, the character coordinate value correcting unit 34 proceeds to the process of step S617.

In step S617, the character coordinate value correcting means 34 increments the value of dW2 by "1". Then, the character coordinate value correcting means 34 determines whether or not the value of dW2 is larger than a preset constant TW2 (step S618).

In step S618, the value of dW2 is a constant TW
If it is determined that it is not larger than 2, the character coordinate value correcting means 34 executes the processing of step S604 again.

In step S618, the value of dW2 is a constant TW
If it is determined that the value is greater than 2, the character coordinate value correcting means 34 increments the value of dx2 by "1" (step S619). Then, the character coordinate value correcting means 3
No. 4 determines whether the value of dx2 is larger than the constant TX2 (step S620).

In step S620, the value of dx2 is a constant TX
When it is determined that it is not larger than 2, the character coordinate value correcting means 34 executes the processing of step S613 again.

In step S620, the value of dx2 is a constant TX
When it is determined that it is larger than 2, coordinates (Δx + dx (i), Δy + dy (i)) obtained by correcting the circumscribed rectangle of each character of the character string in the image stored in the image storage unit 21;
H (i) + ΔH + dH (i) and W (i) + dW (i) are calculated and supplied to the graphic matching means 35. Then, the character coordinate value correcting means 34 ends the processing of this flowchart, that is, the processing of step S6 in FIG.

The character coordinate value correcting means 34 determines in step S6
Is completed (the processing of the flowchart in FIG. 4),
Next, the graphic matching means 35 cuts out a partial image from each of the characters "out", "gold", "den", and "vote" according to the coordinates corrected in steps S5 and S6 (step S7).

The graphic matching means 35 extracts the characteristic amount of each character from the cut-out partial image and compares it with the matching characteristic amount stored in the matching characteristic amount storage area 22b of the dictionary storage unit 22. Then, the average value sm of the similarity of each character obtained as a result of the comparison is calculated (step S8). The calculated average value sm of the similarities is supplied to the output device 4.

The output device 4 is provided with the graphic matching means 35.
According to the average value sm of the similarities supplied from, the identification information for identifying the form of the “payment slip” stored in the classification storage area 22c of the dictionary storage unit 22 is output (step S9). Thus, the classification of the form is completed, and the processing of this flowchart ends.

As described above, in the form classifying apparatus of this embodiment, the position shift of the character string included in the image fetched from the image input device 1 and stored in the image storage unit 21 is first determined by the character The entire character string is roughly corrected by the column coordinate value correcting means 33, and further finely corrected in character units by the character coordinate value correcting means. For this reason, it is possible to quickly correct the positional deviation from the coordinates of the collation character string.

Further, a multi-tone image is fetched from the image input device 1, and characters are collated in accordance with a feature amount expressing each character in multi-tone. Therefore, each pixel in the image is 2
As in the case of value conversion, a character portion is not mistaken as a background portion, and a background portion is not mistaken as a character portion. For this reason, highly accurate character matching becomes possible.

The form classification apparatus of this embodiment can perform high-speed and high-accuracy classification of a form by utilizing the above-described character matching.

The present invention is not limited to those described in the above embodiments, and various modifications and applications are possible. Less than,
Modifications of the above embodiment applicable to the present invention will be described.

In the above embodiment, the case where the present invention is applied to the collation of a character string including one or more characters has been described. Specific examples of "character" in this case include "out",
He listed four kanji: “gold”, “den”, and “vote”. However, the "characters" to be collated in the present invention include, in addition to kanji, all kinds of characters for describing natural languages such as hiragana, katakana, and the alphabet, and further, figures such as mathematical symbols and star shapes. It can be applied including the above.

In the above embodiment, the processing for a horizontally written character string has been described. However, the present invention can be similarly applied to a vertically written character string.
In this case, the application of the vertical pixel projection profile and the horizontal pixel projection profile is opposite to that in the above embodiment.

In the above embodiment, a two-dimensional digital image having gradations is fetched from a form by the image input device 1 and stored in the image storage unit 21. However, the image to be processed may be captured by a digital still camera, for example, and transferred from the memory in the digital still camera to the image storage unit 21.

In the above embodiment, the form classifying apparatus does not have the function of rejection determination, but may have this function. FIG. 8 is a functional block diagram illustrating a functional configuration of a form classification device having a rejection determination function. This form classification device is substantially the same as the form classification device of FIG.
1 is different from that of the first embodiment.

The character string coordinate value correcting means 51 includes a rejection determination section 51a. The rejection determination unit 51a determines whether a character string is included in the image due to a change in the value of the generated pixel projection profile. Character string coordinate value correcting means 51
When it is determined that the character string is not included, the subsequent processes such as the correction of the coordinate value in the unit of the character string, the correction of the coordinate value in the unit of the character, and the comparison of the feature amount are stopped.

In the above embodiment, the data processing device 3
Is configured by a memory and a CPU, and the functions of the units 31 to 35 are realized by the CPU executing a program stored in the memory. On the other hand, as shown in FIG. 9, the program executed by the data processing device 7 is stored in a computer-readable recording medium such as a CD-ROM 6a and distributed, and the disk drive 6 is driven to store the program in the CD-ROM 6a. The read programs may be sequentially read into the memory of the data processing device 7 and executed by the CPU.

In the above embodiment, the case where the present invention is applied to the form classification device has been described. However,
The present invention can be applied to classification of documents other than forms. Also, the present invention, besides the classification of documents, for example,
Optical character reader (OCR: Optical Character)
Reader) or an authentication device for determining the authenticity of a printed document such as securities.

FIG. 10 is a functional block diagram showing a functional configuration of an OCR according to a modification of the embodiment of the present invention. In this OCR, the storage device 8 differs from the storage device 2 of the form classification device in FIG. 1, and the dictionary storage unit 81 has a code information storage area 81a instead of the classification storage area 22c. The output device 9 is different from that of the form classification device of FIG.

The code information storage area 81a stores code information corresponding to characters. Further, the graphic matching means 35 outputs the similarity not for the average value but for each character. Then, the output device 9 outputs the code information of each character stored in the code information storage area 81a.

With such a configuration, it is possible to read an image of a document on which a character string is written and convert each character in the character string into a character code at high speed and accurately.

FIG. 11 is a functional block diagram showing a functional configuration of an authentication apparatus according to a modification of the embodiment of the present invention. In this authentication apparatus, the storage device 10 is different from the storage device of the form classification device in FIG. 1, and the dictionary storage unit 101 has an allowable value storage area 101a instead of the classification storage area 22c.
The output device 11 is different from the output device 4 of the form classification device in FIG. 1 in that the output device 11 has an alarm unit 11a.

The allowable value storage area 101a stores an allowable value of the difference between the characteristic amount of the character string in the image and the characteristic amount of the collation character string. Further, the output device 11 issues an alarm from the alarm unit 11a when the similarity supplied from the graphic matching unit 35 is equal to or larger than the allowable value stored in the allowable value storage area 101a.

With this configuration, the authenticity of a document such as a security document in which a predetermined character string is described can be quickly and accurately determined.

[0112]

As described above, according to the present invention,
Since the positional deviation of the character string is roughly corrected on a character-string basis and further finely corrected on a character-by-character basis, the positional deviation of the character string can be corrected at a high speed. In addition, since character strings can be collated with feature amounts extracted from multi-gradation image data, highly accurate collation can be performed. By utilizing such character string matching, document classification, character reading,
Alternatively, the authenticity of a security or the like can be determined quickly and accurately.

[Brief description of the drawings]

FIG. 1 is a functional block diagram illustrating a functional configuration of a form classification device according to an embodiment of the present invention.

FIG. 2 is a flowchart showing an operation according to the embodiment of the present invention.

FIG. 3 is a flowchart showing an operation according to the embodiment of the present invention.

FIG. 4 is a flowchart showing an operation according to the embodiment of the present invention.

FIG. 5 is a diagram showing a specific example of a form.

FIGS. 6A to 6I are diagrams showing specific examples of the operation according to the embodiment of the present invention.

FIGS. 7A to 7I are diagrams illustrating a specific example of an operation according to the embodiment of the present invention;

FIG. 8 is a functional block diagram showing a functional configuration of a form classification device according to a modification of the embodiment of the present invention.

FIG. 9 is a functional block diagram showing a functional configuration of a form classification device according to a modification of the embodiment of the present invention.

FIG. 10 is a functional block diagram showing a functional configuration of an optical character reader according to a modification of the embodiment of the present invention.

FIG. 11 is a functional block diagram showing a functional configuration of an authentication apparatus according to a modification of the embodiment of the present invention.

[Explanation of symbols]

 Reference Signs List 1 image input device 2 storage device 3 data processing device 4 output device 5 data processing device 6 disk drive 6a CD-ROM 7 data processing device 8 storage device 9 output device 10 storage device 11 output device 11a alarm means 21 image storage unit 22 dictionary Storage unit 22a Coordinate storage area 22b Matching feature storage area 22c Classification storage area 31 Form position detection means 32 Coordinate conversion means 33 Text string coordinate value correction means 34 Text coordinate value correction means 35 Figure matching means 51 Text string coordinate value correction means 51a Rejection determination unit 81 Dictionary storage unit 81a Code information storage area 101 Dictionary storage unit 101a Allowable value storage area

Claims (6)

[Claims] An image storage means for storing multi-tone image data taken from a document of a predetermined format including a character string, coordinates at which each character of the character string should be located in the document, and the character string A dictionary storage unit that stores a feature amount to be included in the document and identification information for identifying the format of the document including the character string in association with each other; and the document in the image data stored in the image storage unit. A position detecting means for detecting a position of the character, and a coordinate system of coordinates where each character in a character string should be located in the document according to a detection result of the position detecting means, from the document coordinate system based on the document to the image. A coordinate conversion unit for converting the image data into an image coordinate system based on the data capture state; and the coordinates of the entire character string included in the image data stored in the image storage unit and the coordinate system converted by the coordinate conversion unit. documents Character string coordinate correcting means for correcting a deviation from coordinates at which a character string should be located in a character string unit; and each of the characters included in the character string whose coordinates have been corrected in character string units by the character string coordinate correcting means. Character coordinate correcting means for correcting, in character units, a deviation between the coordinates of a character and the coordinates where each character of the character string whose coordinate system has been converted by the coordinate converting means, and A graphic for extracting a characteristic amount of a character string in image data taken from the document whose coordinates have been corrected, and comparing the extracted characteristic amount with a characteristic amount to be included in the character string stored in the dictionary storage unit. A document classification apparatus comprising: a matching unit; and an output unit that outputs identification information stored in the dictionary storage unit according to a result of the matching by the graphic matching unit. 2. An image storage means for storing multi-tone image data taken from a document of a predetermined format including a character string, coordinates at which each character of the character string should be located in the document, and the character string A dictionary storage unit for storing a feature amount to be provided and a character code corresponding to the character string in association with each other; and a position detection for detecting a position of the document in image data stored in the image storage unit. Means, according to the detection result of the position detection means, based on the state of capturing the image data from the document coordinate system based on the document, the coordinate system of the coordinates where each character in the character string in the document should be located A coordinate conversion means for converting the image data into a converted image coordinate system, coordinates of the entire character string included in the image data stored in the image storage means, and a position of the character string in the document whose coordinate system has been converted by the coordinate conversion means. Everything A character string coordinate correcting means for correcting a deviation from a coordinate in a character string unit; and a coordinate of each character included in the character string whose coordinates are corrected in a character string by the character string coordinate correcting means, and the coordinate conversion. Character coordinate correcting means for correcting, in character units, a deviation from coordinates at which each character of the character string whose coordinate system has been converted by the means, and the document in which the coordinates of each character have been corrected by the character coordinate correcting means A graphic matching unit that extracts a feature amount of a character string in image data taken in from the storage device and compares the extracted feature amount with a feature amount that the character string should have in the dictionary storage unit; Output means for outputting a character code stored in the dictionary storage means in accordance with the result of the comparison by the means. 3. An image storage means for storing multi-tone image data taken from a document in a predetermined format including a character string; coordinates at which each character of the character string should be located in the document; A dictionary storage unit that stores a feature amount to be provided and an allowable range of the feature amount in association with each other; a position detection unit that detects a position of the document in image data stored in the image storage unit; According to the detection result of the position detecting means, the coordinate system of the coordinates where each character in the character string should be located in the document is an image based on the state of capturing the image data from the document coordinate system based on the document. Coordinate conversion means for converting to a coordinate system; coordinates of the entire character string included in the image data stored in the image storage means; and a character string to be located in the document whose coordinate system has been converted by the coordinate conversion means. Without coordinates A character string coordinate correcting means for correcting this in a character string unit, and a coordinate system in which the coordinates of each character included in the character string whose coordinates are corrected in units of the character string by the character string coordinate correcting means and the coordinate conversion means are used. Character coordinate correcting means for correcting, in character units, a deviation from coordinates at which each character of the converted character string is to be located; and the character coordinates corrected by the character coordinate correcting means are fetched from the document. A graphic matching unit that extracts a feature amount of a character string in image data, and compares the extracted feature amount with a feature amount that the character string should have, stored in the dictionary storage unit; An authenticity judging device, comprising: authenticity judging means for judging whether or not the difference in the feature amount falls within a range of an allowable range stored in the dictionary storage means. 4. A method of associating coordinates of each character of a character string in a document, a feature value of the character string, and identification information for identifying a format of a document including the character string. A method of classifying a document in a computer system having a stored dictionary, the method comprising: storing in a memory multi-gradation image data captured from a document in a predetermined format including a character string; Detecting a position of the document in data; and a document coordinate system based on the document, the coordinate system of coordinates where each character in a character string should be located in the document according to the detected position of the document. Converting the image data into an image coordinate system based on the state of capturing the image data, and the coordinates of the entire character string included in the image data stored in the memory and the image data before being converted to the image coordinate system. Correcting the deviation from the coordinates where the character string should be located in the written document in units of character strings; and converting the coordinates of each character included in the character string whose coordinates have been corrected in units of character strings into the image coordinate system. Correcting the deviation from the coordinates where each character of the character string should be located in the document on a character-by-character basis; and the characteristic amount of the character string in the image data imported from the document in which the coordinates of each character have been corrected Extracting the extracted feature value with the feature value to be possessed by the character string stored in the dictionary, and outputting the identification information stored in the dictionary according to the matching result. A document classification method comprising: 5. A computer system having a dictionary in which coordinates at which each character of a character string is to be located in a document, feature values to be included in the character string, and a character code corresponding to the character string are stored in association with each other. Storing a multi-gradation image data captured from a document in a predetermined format including a character string in a memory; and a position of the document in the image data stored in the memory. Detecting the position of the image data from the document coordinate system based on the document based on the coordinate system of the coordinates where each character in the character string should be located in the document according to the detected position of the document. Converting to the original image coordinate system; and coordinates of the entire character string included in the image data stored in the memory and character strings in the document converted to the image coordinate system. Correcting the deviation from the coordinates to be placed in units of character strings; and the coordinates of each character included in the character strings corrected in coordinates in units of character strings and the character strings in the document converted into the image coordinate system. Correcting the deviation from the coordinates where each character should be located in units of characters, and extracting the feature amount of the character string in the image data taken from the document in which the coordinates of each character have been corrected, and extracting the extracted features. Collating an amount with a characteristic amount to be possessed by the character string stored in the dictionary; and outputting a character code stored in the dictionary according to the collation result. Character reading method. 6. A computer system having a dictionary in which coordinates at which each character of a character string is to be located in the document, a feature value to be included in the character string, and an allowable range of the feature value are stored in association with each other. A method for determining the authenticity of a document, comprising the steps of: storing in a memory multi-tone image data captured from a document in a predetermined format including a character string; and storing the document in the image data stored in the memory. Detecting the position of the document, and, according to the detected position of the document, a document coordinate based on the document based on a coordinate system of coordinates where each character in the character string should be located in the document stored in the memory in advance. From the system to an image coordinate system based on the state of capture of the image data, and the coordinates of the entire character string included in the image data stored in the memory and the image coordinate system. Correcting the deviation from the coordinates where the character string should be located in the document in units of a character string; and converting the coordinates of each character included in the character string whose coordinates have been corrected in the unit of character string into an image coordinate system. Correcting the deviation from the coordinates where each character of the character string should be located in the written document, on a character-by-character basis; and characterizing the character string in the image data captured from the document in which the coordinates of each character have been corrected Extracting a quantity, and comparing the extracted feature quantity with a feature quantity that should be included in the character string stored in the dictionary; and comparing the difference in the matched feature quantity to a range of an allowable range stored in the dictionary. Determining whether or not there is an authenticity determination method.