JP2000293625A - Slip processor and program recording medium therefor - Google Patents

Abstract

PROBLEM TO BE SOLVED: To surely specify the classification of plural kinds of slips by recognizing the features of the shapes of the slips by plural methods even at the time of character recognizing the plural kinds of slips by slip formats corresponding to the classification at once. SOLUTION: A CPU 1 recognizes the shapes of slip images read by an image scanner 7 for each kind based on preliminarily decided plural kinds of shape recognizing methods(slip size classification, corner cut classification, and hole classification), and decides the classification of the slips according to the combination of each recognized result. Then, the CPU 1 character-recognizes the slip images based on slip formats corresponding to the decided classification.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a form processing apparatus for reading a form image and recognizing characters, and a program recording medium therefor.

[0002]

2. Description of the Related Art Conventionally, this type of form processing apparatus requires a processing speed and a recognition rate in order to improve work efficiency. Therefore, a form format for defining a character recognition area for each form to be processed is required. It is registered and character recognition is performed based on this form format. here,
In the case where various forms are automatically fed and character recognition is performed, the type (classification) of the form is automatically specified. Things are known. That is, a method of classifying each form by recognizing the form size, form ID
Is printed on a form in advance, and the form ID is read and recognized to classify each form ID.
A method of classifying by evaluating a character recognition rate is known.

[0003]

However, the classification method based on the form size cannot be classified if different forms of the same size exist, and the classification method based on the form ID has the form ID printed in advance. Only the form is to be processed, and unless the form ID is added, it cannot be classified. Further, the classification method based on the recognition rate is unsuitable when the number of classifications is large because the number that can be classified at one time is limited due to the problem of processing speed.
An object of the first invention is to recognize a shape characteristic of a form by a plurality of methods even when characters of a plurality of forms are recognized at once in a form format corresponding to the classification. Is to be surely identified. A second object of the present invention is to identify a class by reliably recognizing the density of a specific area of a form even when performing character recognition of a plurality of forms at once in a form format corresponding to the classification. Is to be able to do it.

[0004]

The means of the present invention are as follows. The invention according to claim 1 is a reading means for reading a form image, a shape recognizing means for recognizing the form of the form image read by the reading means for each type based on a plurality of predetermined shape recognition methods, Classification determining means for determining the classification of the form based on a combination of the respective recognition results recognized by the shape recognition means, and a character for recognizing a form image based on a form format corresponding to the classification determined by the classification determining means And recognition means. The shape recognizing means includes a plurality of types of shape recognizing methods, a method of recognizing the size of the entire form, a method of recognizing a corner cut portion on the form, and a method of recognizing a hole formed at a specific position of the form. The shape of the form may be recognized by combining at least two of the methods. In the invention according to claim 1, when the form image is read,
The shape of this form image is recognized for each type based on a plurality of predetermined shape recognition methods, and the classification of the form is determined based on a combination of the respective recognition results, and the form is determined based on a form format corresponding to the determined classification. To recognize characters in the form image. Therefore, even when character recognition is performed on multiple forms at once in a form format corresponding to the classification, it is necessary to recognize the form characteristics of the form in multiple ways to identify the classification reliably. Can be.

According to a third aspect of the present invention, there is provided a reading means for reading a form image, a density recognizing means for recognizing the density of a specific area based on the form image read by the reading means, and a recognizing means for recognizing the density by the density recognizing means. Classification determining means for determining the classification of the form based on the determined density, and character recognition means for character recognizing a form image based on a form format corresponding to the classification determined by the classification determination means. is there. It is to be noted that a shape recognizing means for recognizing the form of the form based on the form image read by the reading means is provided, and the classification deciding means comprises a shape recognized by the shape recognizing means and a density recognized by the density recognizing means. The classification of the form may be determined by a combination of the above. According to the invention of claim 3, when the form data is read, the density of the specific area is recognized based on the form image, the classification of the form is determined based on the recognized density, and the determined classification is determined. The form image is recognized as a character based on the corresponding form format. Therefore, even when characters of a plurality of forms are recognized at once in a form format corresponding to the classification, the classification can be reliably specified by recognizing the density of a specific area of the form.

[0006]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS (First Embodiment) A first embodiment of the present invention will be described below with reference to FIGS. FIG. 1 is a block diagram showing the overall configuration of a form processing apparatus with an image scanner. The CPU 1 is a central processing unit that controls the overall operation of the form processing device according to various programs. The storage device 2 includes a storage medium 3 in which an operating system, various application programs, a database, character font data, and the like are stored in advance, and a drive system thereof. The storage medium 3 is fixedly provided or removably mountable, and includes a magnetic / optical storage medium such as a floppy disk, a hard disk, an optical disk, and a RAM card, and a semiconductor memory. The programs and data in the storage medium 3 are loaded into the RAM 4 under the control of the CPU 1 as needed. Further, the CPU 1 receives programs and data transmitted from other devices via a communication line or the like and stores them in the storage medium 3 or stored in a storage medium provided in other devices. Existing programs and data can be used via a communication line or the like. An input device 5, a display device 6, and an image scanner 7, which are input / output peripheral devices, are connected to the CPU 1 via a bus line, and the CPU 1 controls these operations according to an input / output program. Input device 5
Has a pointing device such as a keyboard and a mouse for inputting character string data and the like and for inputting various commands. The display device 6 is a liquid crystal display device, a CRT display device, a plasma display device, or the like. The image scanner 7 is a stationary scanner that reads a form image by scanning the form.

FIG. 2 is a partial block diagram of the storage device 2. The image memory 2-1 stores and holds a form image read by the image scanner 7, and the CPU 1 stores the form image in the image memory 2-1. Is analyzed to determine the classification of the form. At this time, the CPU 1 determines the form classification by recognizing the form shape using a plurality of predetermined shape recognition methods or using other recognition methods together. FIG. 3 exemplifies a form that can be used in this embodiment. In the illustrated example, nine forms are shown. FIG. 4 shows the contents of a heterogeneous form classification definition file 2-2 storing classification definition information for classifying the nine types of forms shown in FIG. 3 according to their characteristics. The form classification is determined with reference to the definition file 2-2. Here, the heterogeneous form classification definition file 2-2 stores definition information defining “form size classification”, “form shape classification”, “rectangular area density classification”, and “recognition rate classification” corresponding to the form type. In the configuration, CP
When determining the form classification, U1 specifies the classification hierarchically by sequentially searching each classification item in the heterogeneous form classification definition file 2-2 according to the arrangement order. That is, the nine types of forms have a size of “210 × 297”.
mm "," 210 x 148.5 mm "," 180 x 150 m
m), and is divided into three groups.
01, 002, and 003 belong to the group of “210 × 297 mm approximately A4 size”, and the forms 011, 012, 01
3 belongs to the group of “210 × 148.5 mm approximately A5 size”, and the forms 211, 022, and 023 are “180 ×
150mm substantially B6 size ". As described above, the nine types of forms are divided into three groups by "form size classification", and forms of the same size are classified by "form shape classification", "rectangular area density classification", and "recognition rate classification".

[0008] The "form form classification" means that the form is classified according to whether a corner portion on the form is cut or whether there is a stop hole at a predetermined position, and in the examples of FIGS. Are classified according to whether there is a cut in the upper right corner or a stop hole in the upper right. Also,
“Rectangular area density classification” means that classification is performed according to the density of a specific area (lower right rectangular area) on a form. In this example, the rectangular area density is “smaller than threshold”, Classification is made according to whether the value is “above the threshold”. "Recognition rate classification" means that the entire form is classified by evaluating the recognition rate when character recognition is performed. In this example, the classification is made according to whether or not to use the recognition rate classification. I have. In the example of FIG. 3, three types of forms belonging to the form size “210 × 297 mm” are:
Since there is no difference in the form shapes, the "form form classification" is not used, but is divided by the "rectangular area density classification" and the "recognition rate classification", and the form size is "210 x 148.5 m".
The three types of forms belonging to the form size “180 × 150 mm” are further divided into “forms”, “form shape classification (upper left corner cut presence / absence)” and “rectangular area density classification”. Shape classification (presence or absence of upper left stop hole) "and" rectangular region density classification ".

The rectangular area density definition file 2-3 stores definition information for defining a rectangular area on a form and a density threshold, and FIG.
It is a figure for explaining the data structure of -5. here,
In order to define a region including a printed portion on which a payee name or the like is printed among the regions on the form shown in FIG. 5A as a rectangular region for density determination, a rectangular region density definition file 2-
3 includes position information (horizontal,
Vertical information), width information, and height information. The position information, the width information, and the height information are in units of mm, and the position information is the distance in the horizontal and vertical directions from the reference point. In the rectangular area density definition file 2-3, "30%" is set as a density threshold. The form format file 2-4 stores the form format for each form type, and the CPU 1 stores the form format file 2-4 based on the form classification determined by referring to the heterogeneous form classification definition file 2-2. Access
A form format corresponding to the classification is read out, and the form image is recognized based on the form format. The form format defines the position and size of a character recognition area (see FIG. 5A) defined on the form as a rectangular area, and is arbitrarily registered in advance corresponding to various forms. Things.

Next, the operation of the form processing apparatus will be described with reference to the flow charts shown in FIGS. Here, a program for realizing each function described in these flowcharts is stored in the storage medium 3 in the form of a readable program code.
1 sequentially executes operations according to the program code. This is the same in other embodiments described later. FIG. 6 is a flowchart showing an operation when reading a form image and recognizing characters. First,
When the form image is read by the image scanner 7, the CPU 1 takes in the form image and stores it in the image memory 2-1 (step A1). Then, the size of the form image in the image memory 2-1 is analyzed (step A2). At this time, the four corners of the form are specified and the horizontal and vertical sizes are measured. When the form size is obtained by this, the heterogeneous form classification definition file 2-2
, A "form size classification" corresponding to the size is searched (step A3), and it is checked whether there is a corresponding classification (step A4). Here, if there is no corresponding classification, it is determined that the form is unusable, and the process proceeds to step A31 in FIG.
Although the process ends with an error, if there is a corresponding size classification, it is determined as a classification candidate and stored in the work area in the RAM 4 (step A5). Next, referring to the heterogeneous form classification definition file 2-2, it is checked whether "form form classification regarding corner cut (corner cut classification)" is set in the determined size group (step A6).
Since the “corner cut classification” is not set in the size group G1 having the form size of “210 × 297 mm” as shown in FIG. 4, the step A12 in FIG.
Then, it is checked whether or not the “form shape classification regarding stop holes (stop hole classification)” is set. In this case, the heterogeneous form classification definition file 2-2 is searched based on the determined size group to check whether or not the stop hole classification has been set. Proceeding to step A18, it is checked whether "rectangular area density classification" is set in the determined size group.

Here, since "rectangular area classification" is set for the size group G1, the specific area on the form image is specified by referring to the rectangular area density definition file 2-3 (step A19). The density in the area is measured (step A20). In this case, if a specific area is specified based on the start position information, the width information, and the height information in the rectangular area density definition file 2-3, the density in this area is detected. In the image, density detection is performed by measuring the occupation rate of black dots. Then, the measured density is compared with the “density threshold” set in the rectangular area density definition file 2-3, and according to the comparison result (less than or equal to the threshold and smaller than the threshold), different forms are used. With reference to the classification definition file 2-2, the type of form corresponding to the density classification is searched (step A21). As a result, it is checked whether there are a plurality of corresponding forms (step A22). If the measured density is “above the threshold value”, only one type of corresponding form “form 001” exists in the size group G1. Therefore, the classification is determined as the definitive classification. If the measured density is “smaller than the threshold”, there are two corresponding forms “form 002, 003” in the size group G1,
At this point, the classification cannot be determined, so the corresponding form type is determined as a classification candidate and stored in the work area in the RAM 4 (step A23).

If the corresponding classification can be determined by the “rectangular area density classification” in the size group G1 (in the case of the form 001), the process proceeds to step A28 in FIG. The format file 2-4 is searched, and the corresponding form format is read. Then, the form image is character-recognized according to this form format (step A29), and the recognition result is displayed and output on the candidate screen for one form (step A30). On the other hand, when there are a plurality of forms in the “rectangular area density classification” in the size group G1 and the classification cannot be determined (in the case of the forms 002 and 003), the process proceeds to step A24 in FIG. It is checked whether “recognition rate classification” is set in the determined group size / candidate classification. In this case, it is checked whether "use recognition rate classification" is set in the heterogeneous form classification definition file 2-2 corresponding to the form type "form 002, 003" stored and held in the work area of the RAM 4. . Now, FIG.
As shown in (2), since "use" is set in association with them, the process proceeds to step A25, and the type of the form for which "use recognition rate classification is used" is set to the heterogeneous form classification definition file 2-2. And accesses the form format file 2-4 based on the type, and reads out all the corresponding form formats. Then, the form image is character-recognized according to each form format (step A26), and the type of the form having the highest recognition rate among the recognition results is determined as the definitive classification (step A).
27). As a result, one of the “forms 002, 003” in the size group G1 is determined as the definitive classification by the “recognition rate classification”. Then, characters are recognized in accordance with the form format corresponding to the finalized classification, and the recognition results are displayed as candidates (steps A28 to A28).
A30).

Next, when another form is read, if the form size belongs to the size group G2 corresponding to "210.times.148.5 mm" (steps A1 to A4).
A5) As shown in FIG. 4, since "corner cut classification" is set in this size group G2, this is detected in step A6 and the process proceeds to step A7, where the four corners (corner portions) of the form image are deleted. It is analyzed to determine whether there is a corner cut (step A8). If there is, the type of the form corresponding to the corner cut classification is searched by referring to the different form classification definition file 2-2 (step A9). FIG. 9 is a flowchart showing the search processing in this case. First, "corner cut classification" in the determined size group for one item is extracted from the heterogeneous form classification definition file 2-2 (step B1). Then, the analysis result of the form image obtained in step A7 described above is compared with the "corner cut classification" extracted from the heterogeneous form classification definition file 2-2 (step B2), and it is determined whether the cut positions match. (Step B3) If not, the process proceeds to Step B5. If a match is detected, the corresponding form type is acquired from the heterogeneous form classification definition file 2-2 (Step B4). In this case, the heterogeneous form classification definition file 2-2
Since “upper left corner cut” is set as the corner cut classification corresponding to “form 011”, “form 011” is acquired as the form classification. Next, it is checked whether another “corner cut classification” is set in the determined size group (step B5). If there is, the process returns to step B1 and repeats the above operation. FIG. 11 shows types of forms according to the presence or absence of a corner cut and a corner cut position,
(A) is a case where there is a corner cut at the upper left position of a horizontally printed form on a landscape paper, (B) is a case where there is no cut,
(C) is a case where there is a corner cut at the upper right position of a vertically printed form on landscape paper. Note that among the nine types of forms shown in FIG. 3, only the form of the upper left cut shown in FIG. 11A is illustrated, but the forms are divided into five types according to the presence or absence of the corner cut and the corner cut position. Classification is possible.

Since no other "corner cut classification" has been set for the size group G2, the process of FIG. 9 ends at this point, and the process proceeds to step A10 of FIG. Here, as described above, as a result of searching for the type of form corresponding to “corner cut classification”, whether there is a plurality of applicable forms,
In other words, check whether the classification candidate was confirmed,
Now, in the size group G2, the classification candidates are narrowed down to one of the “form 011”, so that it is determined as the definitive classification, and the process proceeds to steps A28 to A30 in FIG. Also,
When a plurality of forms are searched, the type of each form is determined as a classification candidate, and stored in the work area in the RAM 4 (step A11). In this case, “forms 012 and 013” in the size group G2 are determined as classification candidates. Then, the process proceeds to step A12 in FIG. 7, but the classification candidates “forms 012 and 013” include “stop hole classification”.
Is not set, but since “rectangular area density classification” is set (step A18), a classification determination process (steps A19 to A22) corresponding to the density classification is executed.
In this case, regardless of whether the measured density is “smaller than the threshold” or “above the threshold”, the applicable form type is 1
Since there are only types, classification candidates are determined at this point. That is, if the density is equal to or higher than the density threshold value, “Form 0
12 ", if it is smaller than the threshold value," Form 013 "is determined as a definitive classification, and steps A28 to A30 in FIG.
Move on to

Next, when another form is read, if the form size belongs to the size group G3 corresponding to "180.times.150 mm" (steps A1 to A3).
5), as shown in FIG. 4, "size of cut hole" is set as the form shape classification instead of "corner cut classification" in this size group G3, so that step A in FIG.
At 12, this is detected, and classification determination processing according to the blind hole classification is performed (steps A13 to A17). That is, a predetermined portion (portion where a stop hole is formed) of the form image is analyzed (step A13), and the presence or absence of the stop hole is checked (step A14). If there is no stop hole, the process proceeds to step A18. Since it is detected, the type of the form corresponding to the stop hole classification is searched by referring to the heterogeneous form classification definition file 2-2 (step A15). FIG. 10 is a flowchart showing the search processing in this case. First, one item of “stop hole classification” in the determined size group is extracted from the heterogeneous form classification definition file 2-2 (step C).
1) Compare the analysis result of the form image with the extracted "stop hole classification" (step C2), and check whether the hole positions match (step C3). Here, if they do not match, the process proceeds to step C5, but if a match is detected, the corresponding form type is acquired from the heterogeneous form classification definition file 2-2 (step C4). Then, it is checked whether or not another “stop hole classification” is set in the determined size group (step C5). If there is, the process returns to step C1, and the above operation is repeated. FIG. 12 shows the types of forms according to the presence / absence of a stop hole and the position thereof.
(B) is the upper right position of the vertically printed form on the portrait paper when the stop hole is at the upper right position of the horizontally printed form on the landscape paper (lower right position if the printing direction is not concerned) In this case, there is a stop hole. Although only the case where there is a stop hole at the upper left position is exemplified in the nine types of forms shown in FIG. 3, the forms can be classified into five types according to the presence or absence of the stop hole and the position of the stop hole.

Now, in the heterogeneous form classification definition file 2-2, "top left stop hole" is specified as the stop hole classification corresponding to "form 022, 023", so there is a stop hole in the upper right part of the form. If the form classification is “form 02
2,023 ". Then, step A1 in FIG.
Then, it is checked whether there are a plurality of corresponding forms, that is, whether or not the classification candidate has been determined. If there is only one classification candidate, the classification candidate is determined as the finalized classification, and step A in FIG.
Move to 28-A30. Now, there are two types of “forms 022, 0
Since "23" is retrieved, the type of each form is determined as a classification candidate, and stored in the work area in the RAM 4 (step A17). Then, Step A18
, But this classification candidate “form 022, 023”
Since the rectangular area density classification is set in the, the classification determination processing (steps A19 to A22) according to the density classification is executed. In this case, regardless of whether the measured density is “smaller than the threshold” or “above the threshold”, there is only one applicable form type, and thus the classification candidate is determined at this time. That is, if the density is equal to or more than the density threshold value, “form 022”;
23 ”is determined as the definitive classification, and the step A2 in FIG.
It moves to 8-A30.

As described above, in this embodiment,
When the form image is read, referring to the heterogeneous form classification definition file 2-2, the shape of the form image is determined by a plurality of predetermined shape recognition methods (a form size classification, a corner cut classification, a recognition method by a blind hole classification). ), The classification of the form is determined based on a combination of the recognition results, and the form image is character-recognized based on the form format corresponding to the determined classification. Even when a form is character-recognized in a form format according to its classification, the classification is reliably specified by recognizing not only the form size but also other form characteristics in multiple forms. It is possible to expect a significant improvement in work efficiency. In this case, by sequentially executing a plurality of geometric recognition methods in a hierarchical manner, it becomes possible to narrow down the classification candidates, and it is also possible to determine the classification during the narrowing down.

In the first embodiment described above,
As the corner cut classification, the presence or absence of the corner cut and its position are detected so that the form is classified into five types of forms. However, if it is determined how many corner cuts are formed in the form, the position of the corner cut can be determined. And the number of recognizable classifications can be greatly increased by a combination of the numbers. This is the same for the stop holes.
Furthermore, the form size may be four or more,
Moreover, if the shape and size of the corner cut and the shape and size of the stop hole are recognized, the number of classifications can be further increased.

(Second Embodiment) Hereinafter, a second embodiment of the present invention will be described with reference to FIGS. This second
In the embodiment, the form type is determined only by the rectangular area density classification. In the above-described first embodiment, only one rectangular area is shown as a form to be classified and determined. However, in the second embodiment, the densities of two rectangular areas are determined, and a combination of the determination results is performed. The classification is determined by the following. FIG. 13A is a diagram showing a rectangular area density definition file 2-5 according to the second embodiment. The rectangular area density definition file 2-3 includes two types of rectangular areas No. 1 and No. 2 provided at predetermined positions on the form as shown in FIG.
2, the "area definition information" and the "density threshold value" for defining the area are stored. In addition,
Each area definition information includes a start position and width / height information as in the first embodiment. FIG. 13B shows a rectangular area density classification definition file 2-6 in which conditions according to the density threshold values corresponding to the rectangular areas No1 and No2 are defined for each form type (A, B, C). As a condition definition for each rectangular area, “smaller than the density threshold” and “above the density threshold” are defined for the set values in the rectangular area density definition file 2-5.

FIG. 15 is a flowchart showing the operation of the second embodiment. First, the area definition information corresponding to the rectangular area No. 1 is read from the rectangular area density definition file 2-5, and the corresponding area on the form image is specified as the density determination area (step D1). Then, the density of the specified rectangular area is measured in the same manner as in the first embodiment described above, and the “density threshold” corresponding to the measured density and the area No. 1 set in the rectangular area density definition file 2-5 is set. Are compared with each other to determine whether they are equal to or greater than the set value, and the determination result is stored and held in the work area of the RAM 4 (step D2). Then, it is checked whether or not the next rectangular area exists in the rectangular area density definition file 2-5 (step D3).
Since there is area definition information corresponding to No2, step D1
Then, the area definition information is read and the corresponding area is specified on the form image. Then, the density of this area No. 2 is measured and compared with the set value to determine whether the density is equal to or more than the set value.
The determination result is stored and stored (steps D2 and D2).
3). When the densities of the areas No. 1 and No. 2 have been determined in this way, the process proceeds to step D5, where the rectangular area density classification definition file 2-6 is accessed and its contents are read. That is, the condition definitions of the rectangular areas No. 1 and No. 2 corresponding to the first form A are read out of the condition definitions set for each form in the rectangular area density classification definition file 2-6, and the condition definitions and the areas No. 1 and No. 2 are read. Are compared with each other to check whether they match (step D6). Here, if any mismatch is detected, it is checked whether the next form exists in the rectangular area density classification definition file 2-6. In this case,
Since the presence is detected, the condition definition of the form B is read from the rectangular area density classification definition file 2-6 and compared with the determination result (step D6). In this way, the forms A, B,
The above operation is repeated while sequentially specifying C. As a result, if there is no corresponding form, the fact is notified (step D7), but if there is, the type of the form is determined as the fixed classification (step D8). FIG. 14 (A),
(B) and (C) show three types of forms classified according to whether or not there is print data in the rectangular areas No1 and No2.
(A) shows the case where there is printing in both areas, (B) shows the case where there is no printing in the rectangular area No1, and (C) shows the case where there is no printing in both areas. Although this example shows three types, it is possible to determine four types of classification simply by determining whether or not there is printing. When the form classification is determined in this way, the form format corresponding to the classification is read out, and the form image is recognized as a character in accordance with this form format.
This is the same as the embodiment.

As described above, in the second embodiment, the density of the specific area is recognized based on the form image, and the classification of the form is determined based on the recognized density. Since the form image is recognized by characters based on the form format to be used, even if multiple types of forms are recognized at a time in the form format corresponding to the classification, the classification can be specified reliably. It is possible to expect a significant improvement in work efficiency.

In the second embodiment described above,
Although the densities of the rectangular areas No1 and No2 are measured, the density determination area may be three or more or the entire form.
Furthermore, two types of “lower than the density threshold” and “smaller than the density threshold” are determined. However, three or more types of density levels are set, and a range within which the measured density is determined is determined. By doing so, the number of classes that can be determined can be increased even if there is only one type of density determination area.

[0023]

According to the first aspect of the invention, even when characters of a plurality of forms are recognized at a time in a form format corresponding to the classification, the form characteristic of the forms is recognized by a plurality of methods. Thus, the classification can be specified with certainty. According to the second aspect, even when characters of a plurality of forms are recognized at once in a form format corresponding to the classification, the classification is reliably specified by recognizing the density of a specific area of the form. can do.

[Brief description of the drawings]

FIG. 1 is a block diagram showing an overall configuration of a form processing apparatus.

FIG. 2 is a diagram showing a part of a storage device 2;

FIG. 3 is a diagram exemplifying types of forms;

FIG. 4 is a view for explaining the data structure of a heterogeneous form classification definition file 2-2.

5A is a diagram showing that a specific region on a form is set as a density determination region, and FIG. 5B is a diagram showing a data structure of a rectangular region density definition file 2-3.

FIG. 6 is a flowchart showing an operation when processing a form image to recognize characters.

FIG. 7 is a flowchart showing a form image process following FIG. 6;

FIG. 8 is a flowchart showing a form image process following FIG. 7;

9 is a flowchart detailing step A9 (search processing for searching for a form type corresponding to a corner cut classification) in FIG. 6;

FIG. 10 is a flowchart detailing step A15 (search processing for searching for a form type corresponding to a blind hole classification) in FIG. 7;

FIGS. 11A, 11B, and 11C are diagrams showing types of forms according to corner cuts.

FIGS. 12A, 12B, and 12C are diagrams showing types of forms according to stop holes.

13A is a diagram for explaining a data structure of a rectangular area density definition file 2-5, and FIG. 13B is a diagram for explaining a data structure of a rectangular area density classification definition file 2-6 in the second embodiment.

FIG. 14 shows (A), (B),
FIG. 4C is a diagram showing types of forms according to the density of a rectangular area.

FIG. 15 is a flowchart showing characteristic densities in the second embodiment.

[Explanation of symbols]

 1 CPU 2 Storage device 2-1 Image memory 2-2 Heterogeneous form classification definition file 2-3, 2-5 Rectangular area density definition file 2-4 Form format file 2-6 Rectangular area density classification definition file 3 Storage medium 4 RAM 5 input device 6 display device 7 image scanner

Claims (6)

[Claims] 1. A reading means for reading a form image, a shape recognizing means for recognizing a form of the form image read by the reading means for each type based on a plurality of predetermined shape recognizing methods, Classification determining means for determining the classification of the form based on a combination of the recognition results recognized by the means; character recognition means for character recognizing a form image based on a form format corresponding to the classification determined by the classification determining means; A form processing device comprising: 2. The form recognizing means includes a plurality of types of shape recognizing methods, a method for recognizing the size of the entire form, a method for recognizing a corner cut portion on the form, and a method for recognizing a hole formed at a specific position of the form. 2. The form processing apparatus according to claim 1, wherein the form is recognized by combining at least two of the methods. 3. A reading means for reading a form image, a density recognizing means for recognizing the density of the specific area based on the form image read by the reading means, and a density recognizing means based on the density recognized by the density recognizing means. A form processing apparatus comprising: a classification determining unit that determines a classification of the form; and a character recognizing unit that recognizes characters of the form image based on a form format corresponding to the classification determined by the classification determining unit. . 4. A form recognizing means for recognizing a form of the form based on the form image read by the reading means, wherein the classification determining means recognizes the shape recognized by the shape recognizing means and the density recognizing means. 4. The form processing apparatus according to claim 3, wherein the classification of the form is determined based on a combination with the determined density. 5. A recording medium having a program code read by a computer, comprising: a function of recognizing a shape of a read form image by type based on a plurality of predetermined shape recognition methods; A recording medium having a program code for realizing a function of determining a classification of a form based on a combination of results and a function of character recognizing a form image according to a form format corresponding to the determined classification. 6. A recording medium having a program code read by a computer, the function of recognizing the density of a specific area based on the read form image, and the classification of the form based on the recognized density. A recording medium having a program code for realizing a function of determining and a function of character recognition of a form image based on a form format corresponding to the determined classification.