JPH1027208A - Form collation method and apparatus - Google Patents

Abstract

(57) [Summary] [PROBLEMS] To provide a form collating apparatus for efficiently recognizing general-purpose forms. A ruled line of a form to be verified is extracted from an input image, and a reference line along the outer periphery of the ruled line is specified by a reference line specifying unit. The field extraction unit 13 generates format information indicating a field position for character entry from the extracted ruled line and reference line. The format information correction unit 16
The format information of the collation form registered in advance is corrected so as to match the format information of the collated form. The form selecting unit 17 selects a collation form in which the corrected format information corresponds within a certain range to the format information of the form to be collated, and outputs this as a collation result.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention compares a format of a plurality of collation forms registered in advance with a collated form included in a form input image in which fields for inputting characters and the like are formed. And a method for selecting a collation form that matches the format of the collated form.

[0002]

2. Description of the Related Art When performing automatic recognition of characters written on a form, the type of a form captured as an image is specified by collating with a plurality of collation forms registered in advance.
Normally, a field for entering a character or the like in the form image, for example, a character entry field is extracted, and characters in the extracted character entry field are cut out and subjected to character recognition processing. In this case, the extraction of the character entry field from the form image is performed by comparing the format information of the form and the format information of the collation form. This is a problem when collating. Therefore, conventionally, the above problem has been solved by using a special form printed with a reference mark for determining the position of the character entry field, and defining the position of the character entry field as a relative position from the reference mark. . In addition, a form ID for identifying a fixed form is printed in advance on a predetermined portion of the form, and the form ID is identified by recognizing the form ID with an optical sensor or the like.

[0003]

As described above, conventionally, the presence of the reference mark and the form ID is indispensable for solving the problem of the positional deviation and the difference in the form size that occur when the form image is input. . Therefore, when collating a form with character recognition, a special form on which a fiducial mark and a form ID are printed must be used, and a general-purpose form used in an office cannot be used. In addition, even if the images include the same form, if the resolution at the time of input is different, the recognition result is different. For example, when inputting at low resolution, some ruled lines may not be recognized, or conversely, rule lines that do not actually exist may be erroneously recognized due to the presence of dust or the like.

[0004] Therefore, an object of the present invention is to provide a general-purpose form having no reference mark or form ID, or a case where the resolution at the time of inputting a form image is different. An object of the present invention is to provide an improved matching method that always obtains the same matching result. Another object of the present invention is to provide a form collating apparatus capable of accurately selecting an optimal form from pre-registered forms for a form contained in an input image.

[0005]

According to the present invention, there is provided a form collating method for identifying a field area in which a field for writing a character or the like is formed from a collated form included in an input image. Matching each field area of the plurality of collation forms held in advance with the size of the field area of the identified collated form; and comparing each collated form with the matched field area size with the collated form. Whether there is a correspondence between individual fields of the form, whether there is a correspondence between the field of the collated form whose field area size matches and the integrated field consisting of a plurality of fields which are adjacent to each other in the field area of each collation form, Or, in the field area of each collated form and the collated form field where the size of the field area matches Whether correspondence between the integrated field comprising a plurality of fields which are adjacent to each other, determines,
Selecting a corresponding form for verification within a certain range;
Is executed on a computer device.

In this case, the presence or absence of the correspondence between the fields in the collation form and the collation form or between the field and the integrated field is determined by, for example, individual fields or integrated values represented by relative values from the reference position of the field area. The determination is made based on whether or not the field formation positions match each other.

According to another aspect of the present invention, there is provided a form collating apparatus for solving the above-mentioned problems, which is represented by a reference position of a field area in which a field for writing characters and the like is formed and a relative value from the reference position. Format information storage means for storing format information represented by the formation positions of individual fields for each collation form; a field area is specified from a collated form included in the input image, and the formation position of each field is determined by the field area Format information generating means for generating format information expressed as a relative position from the reference position of the above; for comparing each field area of each collation form with the field area of the collated form to match the sizes of both areas Format information comparing means for calculating the amount of deformation; individual format information for each collation form Format information correction means for correcting the calculated amount of deformation to generate corrected format information; and selecting a collation form in which the corrected format information matches within a certain range with the format information generated for the collated form. Collation means;

The collation means collates the corrected format information of the collation form with the format information generated for the collated form for each individual field. Field and a plurality of fields that are adjacent to each other in the field area of each collation form, or one field of each collation form and a plurality of fields that are mutually adjacent in the field area of the collated form. Configure.

The present invention also provides an instruction storage medium used when implementing the form collating apparatus on a computer device. The instruction storage medium is an instruction storage medium that is read by a computer device and that realizes a group of instructions to be executed by the computer device to realize a processing step for form collation on the computer device. Steps are included. (1) identifying a field area in which a field for writing a character or the like is formed from a collated form included in the input image; and (2) a field area of each of a plurality of collation forms stored in advance. (3) the presence or absence of a correspondence relationship between each matching form and the individual field of the collated form where the size of the field area is matched, and the collated form. Field and the integrated field consisting of multiple fields that are adjacent to each other in the field area of each collation form, or multiple fields that are adjacent to each other in the field area of each collation form and the field area of the collated form Judge whether there is a corresponding relationship with the integrated field consisting of -Option steps.

[0010]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, a form collating apparatus for inputting a form to be verified whose field area is surrounded by ruled lines and selecting a form matching a characteristic of the form to be verified from a plurality of forms for verification will be described as an example. An embodiment of the present invention will be described in detail.

FIG. 1 is a block diagram of a form collating apparatus according to this embodiment. The form collating apparatus 1 includes a form input unit 10, a ruled line extracting unit 11, a reference line specifying unit 12, a field extracting unit 13, a format information storing unit 14, a format information comparing unit 15, a format information correcting unit 1,
6, a form selection unit 17 and a collation result output unit 18.

Each processing unit is usually formed by a microprocessor unit of a computer device executing required processing in accordance with an instruction group (program and necessary data) stored in instruction storage means (not shown). However, the above-mentioned instruction group does not necessarily have to be integral with the computer device, and may be separate from the computer device. In this case, it is necessary to make the computer apparatus physically embodied on a flexible disk or a CD-ROM (compact disk type ROM) in a form readable and executable by the computer apparatus. If the start point and the end point can be grasped in a clear manner, the command group may be embodied on a communication medium.

First, the function of each processing unit of the form collating apparatus 1 will be outlined. The form input unit 10 passes, for example, an image captured from a general-purpose scanner to the ruled line extraction unit 11. This image is an image on which a form to be verified is formed. Ruled line extraction unit 11
Extracts a ruled line in the horizontal direction and the vertical direction from the captured image. The reference line specifying unit 12 specifies a reference line from the ruled line extracted by the ruled line extracting unit 11. The reference line will be described later. Field extractor 1
Reference numeral 3 extracts a field surrounded by the extracted ruled lines, for example, a character entry field, and passes the position and size to the form selection unit 17 in the form of relative coordinates from a reference line. The format information storage unit 14 stores format information about a collation form registered in advance. The format information is information indicating the characteristics of the form. In the present embodiment, the format coordinates include the reference coordinates of the form and the relative coordinates of each field in the form. Here, the reference coordinates are
The coordinates of the specific position of the reference line and the relative coordinates of the field mean the coordinates that specify the position of the field,
Details will be described later. The format information comparing unit 15 compares the format information of the collated form with the format information of each collation form in the format information storage unit 14 to derive a positional deviation and an enlargement / reduction ratio (amount of deformation) between them. This is passed to the format information correction unit 16. The format information correction unit 16 corrects the format information of each collation form in the format information storage unit 14 based on the deformation amount, and passes it to the form selection unit 17. Form selection unit 17
Selects the same or similar feature as the collated form from a plurality of collation forms whose format information has been corrected, and transfers it to the collation result output unit 18. The collation result output unit 18 sends information on the selected collation form to post-processing, for example, to character extraction means.

Next, an example of the operation of the form collating apparatus 1 will be described. Now, it is assumed that the image of the collated form is taken in from the form input unit 10. The collated form is already used,
The field is filled with characters. The captured image is input to the ruled line extraction unit 11, where the ruled lines are extracted. For the sake of simplicity, it is particularly assumed here that the entire ruled line is formed by a horizontal ruled line composed of a solid line or a broken line and a vertical ruled line.

First, a process of extracting a horizontal ruled line with a solid line will be described with reference to FIG. FIG. 2A shows a horizontal ruled line 21 when viewed on a small scale such that a pixel cannot be recognized.
FIG. 4 is a diagram showing an example of the above. In the figure, reference numeral 22 denotes a horizontal ruled line 2
FIG. 2B is an enlarged view of the end point of FIG. 1 when the part is viewed on a scale large enough to allow pixels to be recognized. While scanning the image in the horizontal direction, the ruled line extraction unit 11 determines that the black pixels are horizontally aligned with the ruled line threshold A as shown by any of the reference numerals 221 to 224 in FIG. A part distributed continuously over the above is found. Next, the portions where these portions are vertically adjacent to each other are found. When these parts are found, they are recognized as horizontal ruled lines. The horizontal ruled line recognized in this way is formed from a bunch of four black pixels 221 to 224 that are continuous in the horizontal direction, for example, as shown in FIG.

Next, the process of extracting a horizontal ruled line with a broken line will be described. FIG. 3A is a diagram showing an example of a horizontal ruled line in this case, and FIG. 3B is a partially enlarged view of the horizontal ruled line. As in the case of the solid line, the ruled line extraction unit 11 finds an area where black pixels are adjacent to each other while scanning the image in the horizontal direction, and obtains a circumscribed rectangle thereof. Next, the obtained vertical size of the circumscribed rectangle is equal to or smaller than the line width threshold B, and
A rectangle whose horizontal size is equal to or smaller than the dashed line threshold D is regarded as a part of the horizontal dashed line. Further, among the circumscribed rectangles regarded as a part of the horizontal broken line, those whose difference between the vertical center coordinates is equal to or smaller than the center difference threshold C and whose horizontal interval is equal to or smaller than the broken line interval threshold E are integrated. If the horizontal length of all the regions integrated in this manner is equal to or greater than the ruled line threshold value A, the region is recognized as a horizontal broken line and extracted.

As shown in FIG. 4A, a black pixel area (solid vertical line) crosses between a broken line 1 and a broken line 2 where the difference between the vertical center coordinates is equal to or less than the center difference threshold value C. In this case, it is determined whether one horizontal broken line 41 is divided by a black pixel portion or whether the broken lines 1 and 2 are independent broken lines. In this case, attention is paid to the gap area 42 between the broken lines shown in FIG. 4B, and the length of each of the white pixels existing between the broken line 1 and the black pixel portion and between the black pixel portion and the broken line 2 is indicated by the broken line. When the distance does not exceed the interval threshold value E, the two broken lines are regarded as one broken line 41 and the integration process is performed.

Finally, a process for specifying both ends of the broken line is performed. As an example, a process for specifying the right end will be described with reference to FIGS. 5A and 5B. FIG. 5A shows a state where a black pixel region exists in an end region 52 indicated by a broken line 51, and FIG. 5B is an enlarged view of the end region 52.
In this right end portion specifying process, it is determined whether or not a black pixel region exists while the length of the white pixel does not exceed the dashed line interval threshold value E. Update. This updating process is repeated until the length of the white pixel exceeds the broken line interval threshold value E.
Identify the right end of 1.

Note that by replacing the horizontal axis and the vertical axis in the horizontal ruled line extraction process of the ruled line extracting unit 11, the vertical ruled line can be extracted in the same procedure.
When the ruled line extracting unit 11 extracts a ruled line for the collated form, the result is sent to the reference line specifying unit 12.

Based on the ruled line extraction result sent from the ruled line extracting unit 11, the reference line specifying unit 12 generates a pair of ruled lines whose vertical coordinate positions are minimum and maximum among horizontal ruled lines, and A pair of ruled lines having the minimum and maximum horizontal coordinate positions are respectively specified as a horizontal reference line and a vertical reference line surrounding the field area. Each of the reference lines specified in this manner eventually follows the outermost periphery of the field region. The ruled line extraction result and the reference line identification result are sent to the format information comparing unit 14 and the format extracting unit 13, and are processed independently. Therefore, first, the processing contents of the format information comparing unit 14 will be described with reference to FIG.

The format information comparing unit 14 compares a plurality of collation forms stored in the format information storage unit 14 with the collated form, and adjusts the amount of displacement between the two forms and the size of the two forms. The amount of deformation is determined for the horizontal and vertical directions. 6A shows an example of a ruled line and a reference line (thick line) of a collation form stored in the format information storage unit 14. FIG. 6B shows a ruled line extraction process for a collated form. 2 shows an example of ruled lines and reference lines (thick lines) obtained by the processing of the unit 11 and the reference line identification unit 12. As shown in FIGS. 6A and 6B, the upper left of the screen is set as the reference coordinates (0, 0) here. In FIG. 6A, X ′ and Y ′ are the horizontal coordinates of the rule having the smaller horizontal coordinate for the vertical rule, and the vertical coordinates of the rule having the smaller vertical coordinate for the horizontal rule.
Similarly, in FIG. 6B, X and Y are the horizontal coordinates of the rule having the smaller horizontal coordinate for the vertical rule and the vertical coordinates of the rule having the smaller vertical coordinate for the horizontal rule, respectively. At this time, the horizontal displacement and the vertical displacement are given by the following equations.

[0022]

## EQU1 ## Horizontal displacement = X-X 'Vertical displacement = Y-Y'

Further, in FIG. 6A, W 'and H'
Are the spacing between two vertical reference lines and the spacing between two horizontal reference lines, respectively. Similarly, in FIG. 6B, W,
H is an interval between two vertical reference lines and an interval between two horizontal reference lines, respectively. At this time, the amount of deformation in the horizontal direction and the vertical direction for matching the reference line of the form, for example, each enlargement / reduction ratio is given by the following equation.

[0024]

## EQU2 ## Enlargement / reduction ratio in the horizontal direction = W / W 'Enlargement / reduction ratio in the vertical direction = H / H'

Thus, the format information comparing section 1
In 4, the amount of displacement and the amount of deformation are obtained for each collation form, and the results are sent to the format information correction unit 16. The format information correction unit 16 corrects the field coordinates (coordinates indicating the position of the field) in the format information based on the positional deviation amount and the deformation amount obtained by the format information comparison unit 14. One field (Fig. 6
The correction process will be described with reference to FIG. 6C showing the correspondence between the hatched portions (a) and (b).

In FIG. 6C, coordinates x 'and y' are represented by
The field coordinates in the horizontal direction and the field coordinates in the vertical direction of the field arbitrarily extracted from the collation form in the format information storage unit 14 are set. The field coordinates in each direction are both relative coordinates measured from the reference coordinates (0, 0). The format information correction unit 16 calculates the coordinates x ′ and y ′
Is corrected so that can be regarded as the field coordinates measured from the reference coordinates (0, 0) of the collated form. Assuming that the modified coordinates are coordinates x ″ and y ″, these are given by the following equations.

[0027]

X ″ = (W / W ′) × x ′ + X−X ′ y ″ = (H / H ′) × y ′ + Y−Y ′

Looking at this processing in FIG. 6C, the upper left coordinates (x1 ', y1') and the upper right coordinates (x2 ', y
2 ') is the upper left coordinate (x1 ",
y1 ") and the field specified by the upper right coordinates (x2", y2 "). In this way, the field coordinates are corrected for all the collation forms, and the result is transmitted to the form selection unit 18. Sent.

Next, the processing contents of the field extracting section 13 will be described with reference to FIGS. The field extraction unit 13 receives the ruled line extraction result and the reference line identification result,
The individual fields surrounded by ruled lines are extracted from the collated form. At this time, field coordinates of each field are generated. These field coordinates are relative coordinates from the reference coordinates. FIG. 7 shows the processing procedure of the field extraction, and FIG. 8 shows a specific example of the field extraction processing.

At the time of field extraction, first, as shown in FIG. 7, the focus of the horizontal ruled line is changed by moving in the vertical direction as a loop (1) (step S701). However, at the beginning of the process, the horizontal ruled line of interest is the horizontal ruled line with the smallest vertical coordinate. Next, the focused horizontal ruled line is moved to the right to change the focus of the intersection (step S701), and the process 1 is performed. In processing 1, attention is paid to an arbitrary intersection where the horizontal ruled line and the vertical ruled line intersect, and flag A and flag
gB is given as two distinguishable marks (step S
703). FIG. 8A shows, as a specific example, a state in which flagA and flagB are assigned to the noted intersection indicated by a circle.

Next, in the branch 1, while scanning on the horizontal ruled line, it is checked whether or not an intersection exists on the right side of the flag B (step S704). If there is an intersection on the right side of flagB (step S704: Yes), FIG.
As processing 2, the focus of the intersection is moved to the right intersection, flagB is changed to the focus intersection, and a new flag
gC is assigned to the noted intersection, and the process proceeds to branch 2. If f
If there is no intersection on the right side of lagB (step S7)
04: No), loop (2) is executed again.

In branch 2, it is checked whether or not an intersection exists below flagC (step S706). flagC
If there is an intersection under the line (Step S704: Ye
s), as illustrated in FIG. 8D, as the process 3, the focus of the intersection is moved to the lower intersection, the flag C is changed to the focus intersection, and the flag D is newly added to the focus intersection. Proceed to branch 3. If there is no intersection below flagC (step S704: No), branch 1 is executed again. In other words, as illustrated in FIG. 8C, an intersection having no intersection below is skipped, and attention is shifted to an intersection having an intersection below.

In branch 3, the horizontal ruled line is scanned leftward to check whether an intersection exists on the left side of flagD (step S708). If there is an intersection on the left side of flagD (step S708: Yes), FIG.
As shown in the example, the focus of the intersection is moved to the intersection to the left of flagD, the flagD is changed to the intersection of interest, and the process proceeds to branch 4. If there is no intersection on the left side of flagD (step S708: No), branch 2 is executed again. That is, as illustrated in FIG. 8 (e), an intersection having no intersection on the left side is jumped downward, and attention is shifted to an intersection having an intersection on the left side.

In branch 4, flagA is immediately above flagD.
It is checked whether or not exists (step S710). Then, when flagA exists immediately above flagD (step S708: Yes), the process proceeds to processing 5. On the other hand, f
When the flagA does not exist immediately above the flagD (step S708: Yes), the branch 3 is executed again. That is, as illustrated in FIG. 8 (g), an intersection where no flagA exists immediately above the intersection of interest jumps to the left and f.
Focus on the intersection where lagA exists.

In process 5, as illustrated in FIG. 8 (h), flagA, flagB, flagC, flagD
Then, a rectangular area (thick line) having four vertices assigned with is extracted as a field, and field coordinates for designating the extracted field are generated (step S711). Thereafter, the process proceeds to the loop (2) again, and the processes of steps S703 to S711 described above are repeated. The result of the field coordinates extracted through the above procedure is sent to the collation selection unit 17.

The form selecting unit 17 performs a field matching process between the collated form and the collation form based on the results sent from the format information correcting unit 16 and the field extracting unit 13. The procedure of the field matching process is as shown in FIG. 9. First, field matching (1) is performed based on the processing results of the format information correcting unit 16 and the field extracting unit 13. Here, the field (F ") of the format information after correction is compared with the field (F) of the form to be verified (step S91).
The coordinate values of (left, right, upper, lower) of each field F ″ and F are respectively expressed as (x ″ (l), x ″ (r), y ″ (t),
y "(b)), (x (l), x (r), y (t), y
When (b)) is set, the matching condition is expressed as follows.

[0037]

## EQU4 ## (A1) x "(l) -collation threshold G≤x (l) ≤
x "(l) + threshold G (A2) x" (r) -threshold G≤x (r) ≤x "
(R) + collation threshold G (A3) y ″ (t) −collation threshold G ≦ y (t) ≦ y ″
(T) + collation threshold G (A4) y ″ (b) −collation threshold G ≦ y (b) ≦ y ″
(B) + collation threshold G

The collation threshold value G is set in consideration of the resolution of capturing the image of the collated form. The above collation conditions are applied to all of the plurality of collation forms, and the number of mismatch fields is counted. Then, it is determined for each collation form whether or not the number of mismatched fields is zero (step S92),
If there is a collation form that is zero (step S9
2: Yes), collation match, that is, the collation form is regarded as a form having the same format as the collated form (step S93). On the other hand, if the number of unmatched fields is not zero (step S92: No), the process proceeds to field matching (2).

In the field matching (2), it is assumed that a plurality of fields are mistakenly combined into one field due to a break in the ruled line of the collated form in the field extracting unit 13, and the collation form is adjacent. The plurality of fields thus obtained are integrated, and the integrated field is collated with the fields of the form to be verified (step S94). In this step, a field F ″ that satisfies the following collation condition expression is obtained for all the fields F that did not match in the field matching (1).

[0040]

(B1) x ″ (l) ≧ x (l) −collation threshold G (B2) x ″ (r) ≦ x (r) + collation threshold G (B3) y ″ (t) ≧ y (t ) -Collation threshold G (B4) y ″ (b) ≦ y (b) + collation threshold G

When n fields F ″ satisfying the conditional expressions (B1) to (B4) are found, all of these n fields are integrated. Further, the integrated field and the field F are compared with the collation condition (A1). When all of (A4) to (A4) are satisfied, it is considered that the n fields F "coincide with the field F, and the numerical value" n "is subtracted from the number of unmatched fields obtained in step S92. As a result, if the number of mismatched fields becomes zero (step S95: Y
es), it is regarded as a form candidate that may have a matching format (step S96). If the value does not become zero even when this process is performed for all the fields (step S95: No), the process proceeds to field matching (3).

In the field matching (3), it is assumed that one field is erroneously divided into a plurality of fields and recognized by the field extraction unit 13 due to characters and noise in the form to be verified. By integrating multiple extracted fields that are adjacent to each other,
Field matching is performed (S97). In this step, the field F satisfying the following conditional expression is satisfied for each of the format fields F ″ determined as the mismatched field in the field matching (2).
Find all.

[0043]

(C1) x (l) ≧ x ″ (l) −collation threshold G (C2) x (r) ≦ x ″ (r) + collation threshold G (C3) y (t) ≧ y ″ (t )-Collation threshold G (C4) y (b) ≤ y "(b) + collation threshold G

When n fields F satisfying all the conditional expressions (C1) to (C4) are found, all the n fields are integrated. When the integrated field F ″ satisfies all of the collation conditions (A1) to (A4), it is considered that the field F ″ matches the n fields F, and “1” is determined from the number of mismatched fields in step S95. Subtract If the number of unmatched fields becomes zero (step S95: Yes), it is regarded as a form candidate that may have a format match (step S96). On the other hand, if it is not zero (S98: N
o), it is determined that there is no matching (step S99), and the field matching process ends. If there are a plurality of form candidates in step S96, the one with the smallest number of mismatched fields is selected, and this is set as the result of form collation.

As described above, in the present embodiment, the size of the field area of the collation form is modified to match the size of the field area of the collated form, and the fields included in the modified field area and the collated form are modified. Since the fields are compared with each other, it is not necessary to use a special form provided with a mark for identifying the form or a reference mark as in the related art, and a general-purpose form can be used. Further, when the collated form is captured in the form of an image, the collation can be performed correctly even if the field is erroneously recognized due to a break in ruled lines or the mixing of noise components.

When there is no existing format information, that is, when newly registering format information on a form, the processing shown in FIG. 10 is performed. When registering the format information, first, an image of a form in which characters or the like are not entered in the field is fetched from the form input unit 10 (step S101). The ruled line extracting unit 11 scans this image to extract horizontal ruled lines and vertical ruled lines (step S102). In addition, a vertical reference line, a horizontal reference line, and reference coordinates are specified from among the extracted ruled lines, and are sent to the field extraction unit 13 and the format information storage unit 16 (step S103). The field extracting unit 13 represents the field coordinates of the form as a relative value of the reference coordinates,
This is sent to the format information storage unit 16 (step S
104). In the format information storage unit 16, the reference coordinates and the field coordinates obtained in the above steps are registered as format information, and can be used at the time of the next form collation.

Although the embodiments of the present invention have been described in detail, the present invention is not limited to the above embodiments.
For example, in the present embodiment, the vertical reference line and the horizontal reference line of the collation form and the collated form are specified, and the size of the area surrounded by each of the former reference lines is reduced to the size of the area surrounded by each of the latter reference lines. Although the displacement amount and the enlargement / reduction ratio are changed so as to match the field size, the key point is that the size of the field area only needs to be matched in both forms, so that the field area deformation method may be arbitrary. Further, the field matching (2) processing (S94) and the field matching (3) processing (S97) in FIG. 9 do not necessarily have to be executed in this order, and the two processings may be made independent.

[0048]

As is apparent from the above description, according to the present invention, even when a general-purpose form is used as a form to be verified, differences in form size, misregistration, and resolution at the time of image capture are obtained. Can be performed robustly against changes in noise, contamination of noise, and the like, and this has the effect of expanding the applications. Further, by providing the form collation process as in the present invention at the preceding stage of the recognition system, the configuration of the recognition system can be simplified.

[Brief description of the drawings]

FIG. 1 is a block diagram of a form collating apparatus according to an embodiment of the present invention.

FIG. 2A is a diagram illustrating an example of a horizontal ruled line formed by a solid line on an image of a form to be verified according to the present embodiment, and FIG. 2B is an enlarged view of an end of the horizontal ruled line.

FIG. 3A is a diagram illustrating an example of a horizontal ruled line formed by a broken line on an image of a form to be verified according to the present embodiment, and FIG. 3B is an enlarged view of a part of the horizontal ruled line.

FIG. 4A is a diagram illustrating an example of a horizontal ruled line formed by two broken lines on a collated form image according to the embodiment;
3 is an enlarged view of a gap between the horizontal ruled lines.

5A is a diagram illustrating an example of a horizontal ruled line with a broken line having an end according to the present embodiment, and FIG. 5B is an enlarged view around the end.

6A is a diagram illustrating an example of a format of a collation form according to the embodiment, FIG. 6B is a diagram illustrating an example of a ruled line of a collated form image according to the embodiment, and FIG. FIG. 4 is a diagram for explaining processing of a format information correction unit according to the embodiment.

FIG. 7 is an explanatory diagram of a processing procedure of a field extracting unit according to the embodiment.

FIG. 8 is a view showing a specific example in each processing of the field extraction unit according to the embodiment.

FIG. 9 is a diagram illustrating a procedure of a form matching process according to the embodiment.

FIG. 10 is an explanatory diagram of a procedure of a registration process to a format information storage unit according to the embodiment;

[Explanation of symbols]

 REFERENCE SIGNS LIST 1 form collation device 10 form input unit 11 ruled line extraction unit 12 reference line identification unit 13 field extraction unit 14 format information storage unit 15 format information comparison unit 16 format information correction unit 17 form collation selection unit 18 collation result output unit

Claims (7)

[Claims] 1. A step of specifying a field area in which a field for writing a character or the like is formed from a collated form included in an input image, and a field area of each of a plurality of collation forms stored in advance. And matching the size of the field area of the identified document to be verified, and determining whether there is a corresponding relationship between each matching document and the individual field of the document to be verified whose size of the field region matches, Selecting a corresponding collation form in step (c), and executing on a computer device. 2. A step of specifying a field area in which a field for writing a character or the like is formed from a collated form included in an input image, and a field area of each of a plurality of collation forms stored in advance. And matching the size of the specified field area of the collated form, and an integrated field comprising a plurality of fields of the collated form whose field area sizes match and a plurality of adjacent fields within the field area of each collation form Determining the presence or absence of a correspondence relationship with, and selecting a corresponding collation form within a certain range, and executing on a computer device. 3. A step of specifying a field area in which a field for writing a character or the like is formed from a collated form included in an input image, and a field area of each of a plurality of collation forms stored in advance. And matching the size of the field area of the identified document to be matched, and integrating the fields of each matching document with the matched field area size and a plurality of fields adjacent to each other in the field area of the matched document. Determining the presence or absence of a correspondence relationship with a field, and selecting a matching form in a certain range, on a computer device. 4. A method according to claim 1, wherein each of the fields in the collation form and the collated form, or whether there is a correspondence between the field and the integrated field, is performed for each field or integrated field represented by a relative value from a reference position of the field area. 4. The form matching method according to claim 1, wherein the determination is made based on whether or not the forming positions match each other. 5. A collation form including format information including a reference position of a field area in which a field for writing a character or the like is formed, and a formation position of each field represented by a relative value from the reference position. Format information storage means stored for each field, and specifying the field area from the collated form included in the input image and forming individual fields represented by a reference position of the specified field area and a relative value from the reference position. Format information generating means for generating format information including a position, and comparing the format information of each collation form with the format information generated for the collated form, the size of the field area of the collation form and the collated form. Field information comparison means for calculating the amount of deformation for matching Format information correcting means for correcting the individual format information according to the calculated amount of deformation to generate corrected format information; and a collating unit for which the corrected format information matches the format information generated for the collated form in a certain range. A form collating apparatus, comprising: a collating means for selecting a form. 6. The collating means collates the corrected format information of the collation form with the format information generated for the collated form for each individual field. Match every field adjacent to each other in one field of the form and each field of the form for collation, or multiple fields adjacent to each other in the field area of each field of the form to be collated with each other in the field area of each form for collation 6. The form collating apparatus according to claim 5, wherein: 7. A computer readable by a computer device,
An instruction storage medium for realizing a processing step for form collation on the computer apparatus by materializing an instruction group to be executed by the computer apparatus, wherein the processing step comprises: Specifying a field area in which a field for writing a field is formed; and matching the size of each of the field areas of the plurality of collation forms held in advance with the specified field area of the collated form. Step and whether or not there is a correspondence relationship between each matching form and the individual field of the collated form where the size of the field area is matched, from multiple fields adjacent to each other in the field of the collated form and the field area of each collated form Whether there is a corresponding relationship with the integrated field or Determining whether or not there is a corresponding relationship with an integrated field consisting of a plurality of fields adjacent to each other within a field area of the collation form, and selecting a corresponding collation form within a certain range, Instruction storage medium to perform.