JPH0944594A - Dividing method for area of document image and discriminating method for kind of multiple column - Google Patents

Abstract

PROBLEM TO BE SOLVED: To cope with irregular multiple column by making good use of column setting to divide the document image into areas when the multiple column is distinctive. SOLUTION: After the tilt of the inputted document image (101) is corrected (102), a compressed image is generated (103). Small areas are extracted (104) and classified into a character string candidate, etc., in a row direction (106). A column division blank part is extracted from the character string candidate small area by using a connection component of a long white run (10'). The kind of multiple column (one column, plural columns, or fee columns) of the input document is decided (108) and blank parts are selected according to the kind (109); and the small areas are integrated (110) and a document area is extracted.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a document image area dividing method and a column type determining method.

[0002]

2. Description of the Related Art There are roughly two methods for dividing an area of a document image. One of them is a cohesive text area (columns, etc.) that integrates small elements such as characters.
(Japanese Patent Application No. 3-128340).

The other is a method of detecting an edge portion of a region or a blank portion dividing the region and dividing the image according to the detected portion. For example, in the document image processing apparatus described in Japanese Patent Laid-Open No. 1-183783, the edge of a column is obtained from the peripheral distribution of the start position of the character string, and the same attribute (belonging column number) is assigned to the character strings belonging to the same column. The character areas are divided and extracted from the image by combining the character strings having the same attribute and vertically adjacent to each other. Further, in the character recognition device described in Japanese Patent Laid-Open No. 5-166001, the area division point is determined by the peripheral distribution in the horizontal and vertical directions.

[0004]

Since the former method is not bound by the shape of the area, it is possible to divide the area without a clear column, but since the performance is influenced by the integrated parameters, the clear step is defined. Even if there is, it can not be divided,
There is a problem that the characters are excessively divided when the space between the characters is wide.

In the latter method, the entire area may not be divided unless a part of the blank portion dividing the area can be detected, and the area may be inclined, the columns may be irregular, or the shape of the text area may be incorrect. There is a problem that the area cannot be divided even when the area is not rectangular.

It is an object of the present invention to use a document image area division method when the area division of the document image is clear, and to cope with the case where the column arrangement is irregular and the type of the column arrangement. It is to provide a discrimination method.

[0007]

In order to achieve the above object, according to the invention of claim 1, in a method of dividing an area of a document image, a plurality of small areas including a character string are extracted from the document image, A blank part or ruled line is detected from the plurality of small areas, and a column type including one column, a plurality of columns, and a free column is determined based on the detected blank part or ruled line, and the column type According to the above, the small area is integrated by using the blank portion, and the document image is divided into predetermined areas.

The invention according to claim 2 is characterized in that the type of column of the document image is determined based on the number and position of the blank portions or ruled lines.

According to the third aspect of the present invention, a blank portion or ruled line that divides the small area in a direction parallel to the character string is detected, the blank portion or ruled line is used to divide the image in parallel with the character string, and the division is performed. It is characterized in that the number of columns or types is obtained for each of the image portions thus created, and the results are integrated to determine the column type of the entire image.

According to a fourth aspect of the present invention, the detected blank portion is selected according to the type of column.

According to a fifth aspect of the present invention, the method for detecting the blank portion is a combination of a method of detecting a connected component of a long white run as a blank portion and a method of detecting a blank portion from a projection histogram of character elements. It is characterized by doing.

According to a sixth aspect of the present invention, the integration condition of the small areas is changed according to the column type.

According to a seventh aspect of the present invention, when the inclination of the document image is equal to or larger than a predetermined threshold value, the type of column is set to free column.

The invention according to claim 8 is characterized in that the column type includes a preset column type.

According to a ninth aspect of the present invention, in the method of discriminating the column type of a document image, a character string is included from the document image.
A document in which a plurality of small areas are extracted, blank portions or ruled lines are detected from the plurality of small areas, and one column, a plurality of columns, or a free column is included based on the number and position of the blank portions or ruled lines It is characterized by discriminating the column type of the image.

[0016]

DESCRIPTION OF THE PREFERRED EMBODIMENTS One embodiment of the present invention will be specifically described below with reference to the drawings. <Embodiment 1> FIG. 1 shows the structure of Embodiment 1 of the present invention.
In the figure, 101 is an image input means such as a scanner, 1
Reference numeral 02 is an image inclination correcting means, 103 is a means for generating a compressed image of an input image, 104 is a means for extracting a black pixel connected component (small area) from the compressed image, 105 is a row direction detecting means, and 106 is a small area. Categorizing means, 107 means for extracting column division blank part candidates, 108 means for discriminating the column type from the extracted column division blank part candidates, and 109, selecting column division blank part candidates according to the determined column type. Means to do, 11
0 is a means for obtaining a large cohesive text area (column, etc.) by integrating small areas using a column dividing line (column dividing blank part and solid line dividing line), and 111 is a parameter and various data in the process. And the like, a data storage unit for storing, etc., 112 a control unit for controlling each unit, and 113 a data communication path. Note that 102 to 110 and 112 can be realized by software on one processor.

FIG. 2 is a processing flowchart of the first embodiment. First, a document image is input using the image input means 101 (step 201). Next, the tilt correction means 10
The inclination of the input image is corrected by step 2 (step 2
02). As this correction method, for example, JP-A-5-35
The method described in Japanese Patent No. 914 is used. If it is known in advance that the input image has no inclination, this inclination correction processing is not necessary and the inclination correction means 102 may not be provided.

The image compression means 103 generates a compressed image (step 203). For example, if the input image has a resolution of about 400 DPI, it is vertically and horizontally compressed to 1/8 (that is,
If there is at least one black pixel among the 8 × 8 pixels, the pixel after compression is made black), so that normally adjacent characters are fused, but lines and regions are not fused. The black pixel connected component of the compressed image is a character, a character string in which characters are fused, a line segment, a table, a figure (or a part of the figure), or the like. The small area extraction unit 104 extracts this connected component as a small area (step 204).

Then, the row direction detecting means 105 detects the row direction (step 205). As a method of detecting the row direction, for example, the method described in Japanese Patent Laid-Open No. 5-314309 is used. It is not necessary to detect when the row direction is designated in advance. Therefore, in that case, the row direction detecting means 105 may not be provided. However, in that case, a line direction designating means is required. Further, it is not always necessary to obtain the row direction at this point, and the row direction may be obtained before or immediately after the generation of the compressed image. However, the sub-region classification, which is the next process, requires the row direction, so it must be known earlier.

When the row direction is known, the small area classification means 10
6 is a coordinate axis with the character string direction as the horizontal direction, and the small area is a character string candidate, a figure, or a field separator (solid line).
Etc. (step 206). The height of a small area and the black pixel density are used for classification. Since the degree of pixel fusion due to compression differs depending on the inter-character distance, a small area of a character string candidate may be composed of one character.

An embodiment will be described below in which the character line direction is horizontal. The column-divided blank portion candidate detection unit 107 extracts a column-divided blank portion candidate from the character string candidate small area (step 2).
07). The details will be described with reference to FIGS. 3 to 13 and FIG. FIG. 3 is a processing flowchart for extracting a column division blank part candidate. First, the image is divided into several bands in the horizontal direction (character line direction). For that purpose, a horizontal dividing line is extracted (step 301). For example, assuming that a small area (that is, one classified as a character string) is arranged as shown in FIG. 4, a connected component of a white run that is long in the horizontal direction is obtained. The result is shown in FIG. Reference numerals 501 and 502 in FIG. 5 indicate connected components of this long white run (not shown, but long white runs can be obtained in other interline portions as well as 501). Among the long white orchid connected components, those having a sufficient height (that is, larger than a predetermined threshold value) are extracted. This is the horizontal blank space. In the example of FIG. 5, the connected component 501 has a height smaller than a predetermined threshold, and the connected component 502 has a height larger than the predetermined threshold.
This height threshold may be a fixed value determined in advance, or the tallest white-run connected component with the highest height (50 in FIG. 5).
It may be a ratio of 2) to the height. Alternatively, it may be linked to the average height of a small area corresponding to a character string.

If there is a horizontal solid line field separator in step 206, this is also treated in the same manner as the horizontal division blank part.

The horizontal dividing line or the blank portion obtained by the above processing divides the whole into several horizontal band areas (step 302). In the example of FIG. 6, the whole is divided into a band area 602 and a band area 603 by the horizontal division blank portion 601. Hereinafter, a column division blank part candidate is extracted for each band area. Further, the number of steps is determined for each band area (steps 303 and 304).

FIG. 7 is a detailed processing flowchart of step 304 in FIG. First, a white run that is long in the vertical direction is generated in the same manner as described with reference to FIG.
1) Then, the connected component of this is obtained (step 702). This will be described with reference to FIG. In the band area 801, long white runs 802 and 803 are generated, and their connected components are obtained. Of the connected components, only the widest one is extracted. Whether or not the width is sufficiently wide may be determined by a predetermined threshold value or may be determined by the ratio to the widest one among the detected long-white run connected components. Further, the width threshold may be changed according to the degree of separation of the small area corresponding to the character string.

The degree of separation of a small area indicates the degree of inter-character distance, and is used to judge whether the inter-character distance is wide or narrow depending on how much the characters are fused by image compression. In other words, if the number of small areas corresponding to character strings is long in the horizontal direction, it means that there is a lot of fusion between characters, and the degree of separation is low, and it can be determined that the distance between characters is narrow. In such a case, a long-white orchid is unlikely to appear, so even a narrow-white long-run connected component is extracted. On the contrary, if there are few horizontally long small areas, the degree of separation between characters is high, and long white runs are likely to appear.
At this time, only the long white orchid connected component having a wide width is extracted.

The number of stages is temporarily obtained from the long-white run connected components extracted as described above (step 703). Referring to FIG. 9, the extracted long-white orchid connected component is 901,
902. It is detected how many stages there are at the division position. That is, if the number of columns is 2 (that is, there are two text areas), the connected component must be present at (or around) the position indicated by the position 903. If the number of stages is 3, the connected component should be present at the position 904, and if the number of stages is 4, the connected component should be present at the position 905.

Therefore, it is checked how many divided positions each of the extracted connected components has in the number of columns. The processing flowchart is shown in FIG. First turn off all flags
(Step 1001). Next, an unchecked long-white run connected component to be processed is selected. If not, the process ends (steps 1002 and 1003). Next, if the connected component is near the half point from the left in the band area, 1
The / 2 point flag is turned on and the process proceeds to the processing of the next connected component (steps 1004 and 1005).

Thereafter, it is checked whether or not the point is near the 1/3, 2/3, 1/4, and 3/4 points, and if it is near the point, the flag is turned on (steps 1006 to 1013). If it does not correspond to any place, turn on the irregular flag
(Step 1014). At this time, it is possible to adopt a method of setting a flag by using not only the long-white run connected component but also a solid line field separator. In that case, the vertical solid line separator is also processed in steps 1002 and 1003.

The position of the long white run connected component (and the solid line field separator) extracted by the above processing is found in the band area. Next, the number of steps of the band area is obtained from this position flag. This will be described with reference to the process flowchart of FIG. 12 (corresponding to step 703 of FIG. 7 by combining FIGS. 10 and 12).

If only the 1/2 point flag is ON, the number of stages is 2 (steps 1201, 1202). 1
If only the / 3 point flag and the 2/3 point flag are ON, the number of stages is 3 (steps 1203 and 120).
4). If only the 1/4 point flag and the 3/4 point flag are turned on together with the 1/2 point flag, the number of stages is 4 (steps 1205 and 1206). In cases other than the above, determination is made based on the above-described degree of separation (determined by the ratio or absolute number of horizontally long character-equivalent small areas) (step 120
7). If the degree of separation is high and the distance between characters is expected to be wide, the number of columns is set to 1 because the document is often created by a word processor (step 120).
8). In other cases, the number of stages is undefined (step 1
209). Other methods can be adopted for this stage number determination processing. For example, if the irregular flag is ON, the number of stages is always undefined.

Returning to FIG. 7, when the number of stages is detected by the above-described processing, one of the connected components that is a candidate for the stage division blank part is selected (step 704). When the number of stages is indefinite or when the number of stages is 1, the number of stage division blank part candidates is set to 0. In other cases (that is, the stage numbers 2 to 4), all the wide connected components used for the stage number detection are set as stage division blank part candidates. Since the connected component has a width, it is better to set the divided blank part at the position of the center line or at the position of the longest run among the runs that form the connected component in the small area integration in step 210. Time handling becomes easier. As described above, in steps 701 to 704, the extraction processing of the stage division blank part candidate using the long white run is performed.

Next, the step division blank part candidate extraction processing using the marginal distribution histogram in steps 705 to 709 will be described. First, a marginal distribution histogram in which the number of small areas is obtained for each width (class) to some extent is created (step 705). FIG. 11 is a histogram 1 of the peripheral distribution of a small area.
101 is shown. Next, the line head is detected (step 70).
6). This is performed by detecting a class in the histogram 1101 that is more frequent than the next-neighboring class by a threshold value or more. In FIG. 11, 1102 and 1103 correspond to the head of the line. Here, the threshold value may be a fixed value, or may be normalized by the number of small areas in the band area or the total frequency of the histogram. In addition, as shown in FIG. 11, the frequency distribution does not necessarily form a clear valley, so that the frequency distribution may be compared not with the next class but with the next class. If the class to the left of the detected point (1102, 1103) having a large frequency difference is less frequent by the number of consecutive thresholds or more,
The point becomes a blank part candidate (step 706).

For example, in FIG. 11, the area to the left of 1102 is 1
The frequency is low for two consecutive classes, 104 and 1105. 11
Similarly, the frequency on the left side of 03 is also low at 1106 and 1107.
Therefore, the candidates for the blank part are 1105-1104 and 1107.
It becomes -1106.

Then, using the detected blank part candidates,
The number of stages is determined (step 708). First, the division position flag is set in the same manner as in FIG. However, the irregular flag is not used here. When each flag is set, the number of stages is determined. FIG. 13 is a detailed flowchart of step 708 of FIG.

First, the 1/4, 1/2, and 3/4 flags are set to O.
If it is N, the number of stages is set to 4 (step 1301,
1302). At this time, the difference from step 703 is that
This means that the 1/3 flag and the 2/3 flag may be turned on.

If the 1/3 and 2/3 flags are ON, the number of stages is set to 3 (steps 1303 and 1304). 1
If the / 2 flag is ON, the number of stages is set to 2 (steps 1305 and 1306). If none of the above, the number of stages is undefined if the number of blank part candidates is greater than or equal to the threshold value, and otherwise the number of stages is set to 1 (step 1
307, 1308, 1309).

Since the number of steps is detected by the above processing, the division blank part candidate is selected in the same manner as in step 704. Among the blank part candidates used in the step number detection of step 708, only the blank part candidates existing near the division position of the detected step number are set as the column division blank part candidates (step 70).
9). If the number of stages is indefinite or is 1, the stage division blank part candidate is not extracted (it is not extracted in step 709, so there is a possibility that it is extracted in step 704).

Next, the consistency of the number of steps obtained in steps 703 and 708 is checked to determine the number of steps in the band area (step 710). The determination of the number of stages follows the determination rule as shown in FIG. When the number of stages is determined, only the stage division blank part candidates that match the number of stages are left (step 711). For example, the blank part candidate that was determined to have three columns in step 703 and remained so as to match with it is step 71.
As a result of the stage number matching check with 0, if it is determined that the stage number is indefinite, it is removed. When it is determined that the number of stages is 2 in step 703 and the number of stages is 4 in step 708, it is determined that the number of stages is 4, as shown in FIG. 22, and therefore the blank part candidate obtained from the long run is also the blank part candidate obtained from the marginal distribution. Will be adopted together. As described above, in step 207, the extraction processing of the column division blank part candidate area is performed.

Returning to FIG. 2, the column type discriminating means 108.
Determines the type of column (step 208). This classifies an input document into three types, one column, a plurality of columns, and a free column, depending on the number of columns calculated for each band area. An example will be described with reference to FIG. Reference numerals 1401 to 1403 are band areas, and reference numerals 1404 to 1406 are extracted stage division blank area candidates. Here, the number of sets is 1401 = 1,1402
Becomes 3, and 1403 becomes 2. According to FIG. 15, the column type of the entire document is determined.

First, the height H of the entire document is measured (step 1501). This amount H may be the total height of the band area. In the latter case, when there is a ruled line or a chart between the band areas and a gap is generated, it is possible to prevent the ratio from the total of the heights of the various band areas, which will be obtained later, cannot be obtained correctly. Next, the sum of the heights of the strip regions (which are called multi-stage strip regions) determined to have two to four stages (Total)
1) and the total height (Total 2) of the indeterminate step zone area in which the number of steps is indefinite is calculated (steps 1502 and 15).
03).

When the respective total values are obtained, they are sequentially compared with the threshold value. First, if Total 1 / H is larger than the threshold value 1, it is determined that the input document has a plurality of columns (steps 1504 and 1505). Otherwise Tot
When al 1 / H is larger than the threshold 2 (<threshold 1), it is determined that the input document is a free column (step 15).
06, 1507). Otherwise, (Total 1
When + Total 2) / H is larger than the threshold value 3, it is determined that the input document is a free column (step 150).
8, 1509). If the above conditions are not satisfied, it is determined that the input document is a single column (step 151).
0).

Subsequently, the column division blank portion distinction means 109 is
Selection of blank space candidate for column division is performed (step 20).
9). This is in accordance with the document column type determined in step 208. If it is one column, the column division blank part candidate is removed. In the case of multiple columns and free columns, the column division blank part candidate is used as it is as the column division blank part.

The small area integrating means 110 integrates the small areas corresponding to the character strings by utilizing the blank space of the column division and extracts a coherent text area (corresponding to the column or a part thereof) (step 210). ). For how to integrate small areas,
A method is used in which areas adjacent to each other in the row direction are integrated into a row (or a part thereof), and rows (or a part thereof) are integrated in a direction perpendicular to the row direction to form an area ( For example, the method described in Japanese Patent Application No. 3-128340).

When the small areas are integrated in the row direction, the column division blank portion obtained in step 209 is used as in the case of the existing field separator (division line). Do not combine the small areas beyond the dividing line and the dividing blank area. Alternatively, the integration condition is set to be strict in the vicinity of the step division blank part, and only those closer to each other are integrated. Also, this integrated parameter is changed depending on the type of column. That is, in the case of one column, they are integrated even if they are far apart, and in the case of a free column, only those close to each other are integrated. In the case of multiple columns, even if they are separated as in the case of one column, they should be integrated as long as they do not exceed the dividing line and the blank portion for dividing columns, or the dividing line and the blank portion for dividing columns up to the distance equivalent to the column width. Try to integrate unless it exceeds.

<Second Embodiment> FIG. 16 shows the structure of a second embodiment. 17 is a processing flowchart of the second embodiment. In this embodiment, the inclination correcting means 102 of the first embodiment is replaced with the inclination detecting means 1602, and the other configuration is the same as that of the first embodiment. In step 1702, the inclination of the image is detected, and when the inclination angle is larger than the threshold value, it is difficult to detect the blank portion. Therefore, the blank portion detection in step 1708 and subsequent steps is not performed (step 1707), and the column type is selected. It is processed as a free column (step 1712). Others are the same as those in the first embodiment, and thus the description thereof is omitted. Note that 1602 to 1610 and 1612 can be realized by software on one processor.

<Third Embodiment> FIG. 18 shows the configuration of the third embodiment, and FIG. 19 is a processing flowchart thereof. In this embodiment, in addition to the configuration of the first embodiment, the column type indicating means 1
It is configured by adding 814. If the column type instructed by the column type instructing means 1814 is one column, the column division blank part detection processing and the column type determination processing in and after step 1909 are not performed (step 190).
8).

If the column type instructed by the column type instruction means 1814 is a non-single column, step 1
In 910, when the discrimination is performed as shown in FIG.
A place determined as a column is determined as a free column. Others are the same as in the first embodiment. Note that 1802 to 1810 and 1812 can be realized by software on one processor.

<Fourth Embodiment> FIG. 20 shows the structure of a fourth embodiment, and FIG. 21 is a processing flowchart thereof. The fourth embodiment is different from the structure of the first embodiment in that the column dividing blank portion separating means 109 is provided.
In this embodiment, the small area integrating means 110 is removed to determine the column type of the document image. Therefore, in the fourth embodiment, when the column type is determined in step 2108, the process ends. Others are the same as in the first embodiment. Note that 2002 to 2008 and 2010 can be realized by software on one processor.

[0049]

As described above, according to the first aspect of the invention, since the region division is performed only by using the stage division lines having a high consistency as the column set, the highly accurate region division processing is performed. It can be performed.

According to the second aspect of the present invention, it is possible to accurately determine the column type, and it is possible to perform highly accurate area division processing.

According to the third aspect of the present invention, even if the number of columns is different for each part of the image, the type of columns can be obtained with high accuracy, so that highly accurate area division processing can be performed.

According to the fourth aspect of the invention, since it is not necessary to use the area division blank portion having low reliability, it is possible to perform the area division processing with high accuracy.

According to the fifth aspect of the present invention, since it is possible to detect the step division blank portion with high accuracy, it is possible to perform the area division processing with high accuracy.

According to the sixth aspect of the present invention, by adopting the processing according to the type of column, one column is less likely to be erroneously separated, and in the case of a plurality of columns, undivided areas are reduced. It is possible to perform highly accurate area division processing.

According to the seventh aspect of the present invention, even if the input image is inclined and there is a problem in detecting a blank portion, it is possible to secure a certain area division processing capability.

According to the eighth aspect of the invention, by narrowing down the types of columns, it is possible to more accurately determine the types of columns, and it is possible to perform highly accurate area division processing.

According to the ninth aspect of the present invention, it is possible to accurately determine the type of column.

[Brief description of drawings]

FIG. 1 shows a configuration of a first exemplary embodiment of the present invention.

FIG. 2 is a processing flowchart of the first embodiment.

FIG. 3 is a detailed flowchart of step 207 of FIG.

FIG. 4 is a diagram showing an example of classified small areas.

FIG. 5 is a diagram showing an extracted horizontal blank portion.

FIG. 6 is a diagram in which the entire image is divided into strip regions by horizontal division blank portions.

FIG. 7 is a detailed flowchart of step 304 of FIG.

FIG. 8 is a diagram for extracting a white run connected component that is long in the vertical direction from a band area.

FIG. 9 is a diagram illustrating the positions of long-white run connected components in a band area.

10 is a part of a detailed flowchart of step 703 of FIG.

FIG. 11 shows an example of a peripheral distribution histogram of a small area.

12 is a part of a detailed flowchart of step 703 of FIG.

FIG. 13 is a detailed flowchart of step 708 of FIG. 7.

FIG. 14 is a diagram illustrating the number of columns for each band area.

FIG. 15 is a detailed flowchart of step 208 of FIG.

FIG. 16 shows a configuration of a second embodiment of the present invention.

FIG. 17 is a processing flowchart of the second embodiment.

FIG. 18 shows a configuration of a third embodiment of the present invention.

FIG. 19 is a process flowchart of the third embodiment.

FIG. 20 shows a configuration of a fourth embodiment of the present invention.

FIG. 21 is a processing flowchart of the fourth embodiment.

FIG. 22 shows a rule for determining the number of stages.

[Explanation of symbols]

 101 Image Input Means 102 Inclination Correcting Means 103 Image Compressing Means 104 Small Area Extracting Means 105 Row Direction Detecting Means 106 Small Area Classifying Means 107 Column Dividing Blank Area Candidate Extracting Means 108 Column Type Type Discriminating Means 109 Column Dividing Blank Part Separate Means 110 Small Area integration means 111 Data storage unit 112 Control unit 113 Data communication path

Claims (9)

[Claims] 1. A method of dividing an area of a document image, wherein a plurality of small areas including a character string are extracted from the document image, a blank portion or ruled line is detected from the plurality of small areas, and the detected blank portion is detected. Alternatively, based on the ruled line, the type of column including one column, a plurality of columns, and a free column is discriminated, and the small areas are integrated using the blank portion according to the column type, and the document image is displayed. An area dividing method for a document image, characterized by dividing the area into predetermined areas. 2. The method of dividing a document image according to claim 1, wherein the column type of the document image is determined based on the number and position of the blank portions or ruled lines. 3. A blank portion or ruled line that divides the small area in a direction parallel to the character string is detected, an image is divided in parallel with the character string by the blank portion or ruled line, and each divided image portion is divided. 2. The method for dividing an area of a document image according to claim 1, further comprising: determining the number or types of columns and integrating the results to determine the type of columns of the entire image. 4. The document image area dividing method according to claim 1, wherein the detected blank portion is selected according to the column type. 5. The method for detecting a blank portion is characterized by using a method of detecting a connected component of a long white run as a blank portion and a method of detecting a blank portion from a projection histogram of character elements. The method for dividing an area of a document image according to claim 1. 6. The method for dividing an area of a document image according to claim 1, wherein the integration condition of the small areas is changed according to the type of the column. 7. The method of dividing an area of a document image according to claim 1, wherein when the inclination of the document image is equal to or larger than a predetermined threshold value, the column type is a free column. 8. The method of dividing an area of a document image according to claim 1, wherein the column type includes a column type designated in advance. 9. A method of discriminating a column type of a document image,
Extracting a plurality of small areas including a character string from the document image,
A blank part or ruled line is detected from the plurality of small areas, and one column set, a plurality of column sets, based on the number and position of the blank part or ruled line,
A method of discriminating a type of a document image, which includes discriminating a type of a document image including a free column.