JP2001331763A - Method for recognizing table - Google Patents

Abstract

PROBLEM TO BE SOLVED: To precisely separate a group of characters in a tubular document having no ruled lines (or where ruled lines are deleted remarkably) and to precisely recognize table structure even when a table is tilted. SOLUTION: Characters whose pitches are within a fixed value in a table are integrated to obtain the group of the characters and the average of the pitches of the characters in the group of the characters is obtained to re-divide the group of the characters based on the average. Next, a group of characters where the mutual positional relation between the re-divided groups of the characters is within fixed inclination are registered as a row. The groups of characters having a relation where coordinates in the horizontal direction of the re-divided groups of the characters are overlapped with each other are registered as a column. Then, the groups of the characters of respective rows and respective columns obtained like above are recognized to prepare data which can be read into a table.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION In recent years, demand for document recognition processing as input peripheral equipment has been increasing. The present invention relates to a table recognition method capable of recognizing a table without ruled lines with higher accuracy in the document recognition processing.

[0002]

2. Description of the Related Art When a character recognition result of a tabular document is used in, for example, spreadsheet software, a table structure is recognized, and a character group recorded in a column and a row of a character group in the table is recognized. Needs to be recognized. In the case of a table format document having ruled lines, the ruled line structure is usually recognized, and the character group of each row and each column is identified from the ruled line structure. On the other hand, when recognizing a tabular document having no ruled lines as shown in FIG. 11A, conventionally, the character spacing is determined based on the average value of the character size, the character pitch, and the like. Has been used to obtain a character group, and when the vertical and horizontal coordinates of the character group overlap each other, the character group is registered as a row or a column.

[0003]

In the table of FIG. 11 (a), a character group can be accurately estimated by integrating characters at a fixed character interval as described above.
In a table such as that shown in (b), when character groups are estimated at the above-mentioned constant character interval, a character group that should originally be a plurality of character groups, such as an area in the drawing, may be recognized as one character group. When the table is inclined as shown in FIG. 11C, a plurality of rows may be recognized as one row if the row interval is small. The present invention has been made in view of the above circumstances, and the present invention provides a method for recognizing a table in a tabular document in which there is no ruled line or the ruled line is largely deleted.
Provided is a table recognition method that can accurately separate a character group that cannot be separated by the conventional method, and that can accurately recognize a table even when the table is inclined. Aim.

[0004]

FIG. 1 is a diagram for explaining the outline of the present invention. As shown in FIG.
The table is recognized as follows. A character group is obtained by integrating characters in a table with a character spacing within a certain range. Find the average value of the character spacing of the characters in the above character group,
The character group is subdivided based on the average value. A character group in which the positional relationship between the subdivided character groups is within a certain inclination is registered as a line. A character group having a relationship in which the horizontal coordinates of the subdivided character groups overlap each other is registered as a column. When recognizing a table in which the ruled lines are partially omitted, the ruled line included in the image is removed, and then the above table recognition is performed. In the present invention, since the table is recognized as described above, even if a character group with a narrower interval is mixed in the table than the space between other character groups, these character groups are used as a plurality of character groups. Can be accurately recognized. In addition, even a table having a slope can be recognized accurately.

[0005]

FIG. 2 is a processing block diagram of a first embodiment of the present invention. Note that the present invention can be realized by a normal computer system including a CPU, a memory, an external storage device, an input / output device, a scanner for reading an image, a recording medium reading device, a communication interface, and the like. The program for performing the character recognition process of the present invention is stored in the memory. When the program is executed, the program is read into a memory, and the image read by the scanner or the like is subjected to character recognition by the character recognition process of the present invention. Is output from the input / output device.

Next, a first embodiment of the present invention will be described. This embodiment is applied to, for example, recognition of a tabular document having no ruled line as shown in FIG. In this embodiment, as shown in FIG. 2, first, the character group extracting unit 11 obtains rectangular coordinates surrounding the character group. In other words, for a range given as a table without ruled lines, rectangles surrounding connected components are extracted from the connectivity of the pixels forming the characters, and then the rectangles are integrated from the overlap and positional relationship between the rectangles to estimate the characters. Then, a rectangle surrounding the character is extracted. Next, a character group is extracted by integrating the extracted characters from the coordinates of the rectangle by integrating the characters having a length (character interval) between the characters within a certain threshold value,
Find the rectangular coordinates surrounding the character group. FIG. 3B shows a character group extracted from the table without ruled lines shown in FIG. As shown in FIG.
"Month) to (May)" are extracted as one character group because the character interval is short.

Next, the character group subdivision section 12 performs a character group subdivision process. That is, as shown in the flowchart of FIG. 7 described later, the character groups extracted by the character group extraction unit 11 are extracted one by one, the character spacing between adjacent characters is calculated, and the average Wa is calculated. Then, it is checked whether each character is apart from Wa by a certain length or more, and if so, the character group is subdivided. With respect to the table of FIG.
FIG. 3C shows a character group extracted by the character group subdivision process. As shown in the figure, “(January) to (May)” are subdivided as character groups. Next, the row extracting unit 13 performs a row extracting process. In the row extraction unit 13,
As shown in the flowchart of FIG. 8 described later, lines are extracted in consideration of the inclination between the character groups. That is, first, the character groups are sorted in order from the left, and one extracted number is set to i in the order in which the character groups are sorted. If there is no line at this point, a new line is created and the character group i is registered in that line. For example, in the example of FIG. 3C, “sum” is at the leftmost position, so a new row is registered.

When the line is registered as described above, the inclination between the character group i and the character group i closest to the character group i in the horizontal direction among the character groups existing in the line is obtained and already extracted. The row number (g) giving the minimum slope and the slope (θmi
n) is stored. Here, as the inclination, as shown in FIG. 4, a center point may be obtained from the rectangular coordinates surrounding the character group, and a line connecting the center points with a horizontal line may be used. next,
Check whether the inclination θmin is smaller than a threshold value (θth),
If θmin is larger than θth, a new row is created. If θmin is smaller than the threshold (θth), the character group i is registered in the g-th line. This process is performed for all the character groups, and the line extraction is completed. By extracting the lines as described above, the lines and the character groups are associated with each other, for example, as shown in FIG.

Next, a column extraction section 14 performs column extraction. In the column extraction, for example, a character group having a relationship in which the coordinates in the horizontal direction overlap is registered as a column. That is, the horizontal direction is X
When the axis and the vertical direction are the Y axis, a character group in which the coordinate values in the X axis direction of the rectangle surrounding the character group overlap (a character group in which at least a part of the coordinate values in the X axis direction have the same value) is registered as a column. I do. For example, in the example of FIG. 3C, since the coordinate values in the X-axis direction of the character groups in the area surrounded by the broken line in FIG. 5 overlap, the character groups in each area are registered in the same column. . Thereby, for example, as shown in FIG. 6B, the columns are associated with the character groups. The table data creation unit 15 performs character recognition for each character group, creates cell data from the relationship between rows and columns, and registers the character group in the cell data. That is, data that can be written in a table as shown in FIG. 6C is created from the character group recognition result, the row extraction result shown in FIG. 6A, and the column extraction result shown in FIG. 6B. You.

Next, the character group subdivision processing in the character group subdivision section 12 will be described with reference to the flowchart of FIG. First, the character groups extracted by the character group extraction unit 11 are extracted one by one, and the characters (number: Nc) included in the extracted character group i are sorted in order from the left based on the coordinates. Then, the character interval between adjacent characters is obtained, and the average Wa is obtained. That is, as shown in FIG. 7, the number of character groups is set to Ng, and i = 1 (step S1). Then, it is checked whether i ≦ Ng, and if i> Ng, the process is terminated. If i ≦ Ng, Nc is set to the number of characters of the i-th character group (step S3). And i
The characters included in the th character group are sorted in order from the left (step S4). Next, Wa is set to the average character spacing of the character group of the i-th character (step S5).

Next, the distance between the character j and the next character j + 1 is determined in the sorted order, and it is checked whether the character j is separated from Wa by a certain length or more. Subdivides the character group. At this time, the number N of character groups is increased by one, and the number of the character group after the (i + 1) th is increased by one. Then, characters 1 to j are registered in the i-th character group, and rectangular coordinates including the registered characters are obtained. Next, characters from j + 1 to Nc are newly registered in the (i + 1) th character group. Get the rectangular coordinates containing the registered character.
Then, i is incremented by one, and the subdivision process of the next character group is repeated, and the subdivision process is performed for all the character groups. That is, as shown in FIG. 7, j = 1 and j <
Nc is checked (step S7). If j ≧ Nc, i = i + 1 is set in step S15 and the process returns to step S2. If j <Nc, the character j and the character j +
The interval of 1 is set to W (step S8), and it is checked whether or not the character interval W> the average character interval Wa × [constant value] (step S9). If the above conditions are not satisfied, step S
At 10, j = j + 1 is set and the process returns to step S7. When the character spacing W> the average character spacing Wa × [constant value],
The number of character groups is increased by 1 (step S11), and the numbers of the i + 1th and subsequent character groups are increased by 1. Further, the first to j-th characters are registered in the i-th character group, the number of characters and the rectangular coordinates are calculated (step S13), and j + 1 to j + 1 are registered in the (i + 1) -th character group.
The Nc-th character is registered, and the number of characters and rectangular coordinates are calculated (step S14). When the above processing is completed, step S
At 15, i = i + 1, and the process returns to step S2.

Next, the row extracting process in the row extracting section 13 will be described with reference to the flowchart of FIG. In FIG. 8, first, character groups are sorted in order from the left (step S
1) Next, θth is set as a threshold of inclination (preset), Ng is set as the number of character groups, C is set as the number of lines (set to 0 at first) (step S2), and i is set to 0 (step S2).
3). Next, it is checked whether i is smaller than Ng (the number of character groups), and if not, the process is terminated. If i is smaller than Ng, it is checked whether C = 0 (step S).
5). If C = 0, go to step S10, add 1 to C, and register character group i in the C-th line (step S1).
1). Thereby, for example, in the example of FIG. 3C, the character group “total” is registered (since the character group “total” is at the leftmost position). The first line is registered as described above.

Next, 1 is added to i (step S12), and the process returns to step S4. As described above, of the character groups existing on the line, the character group closest to the character group i in the horizontal direction and the character group i , And the row number (g) and the slope (θmin) that give the smallest slope among the already extracted rows are stored.
That is, it is determined whether or not i is smaller than Ng (the number of character groups) in step S4.
In step 5, it is checked whether C = 0. Since C is not 0 this time, the procedure goes to step S6, where j = 1, g = 0 (g is a row number), θmin is set to infinity, and it is checked whether j ≦ C (step S7). Since the first time is j ≦ C, the process goes to step S13, where the character group i of the character group on the j-th line
Then, the inclination θ between the character group closest to the horizontal direction and the character group i is obtained. In step S14, it is checked whether | θ | <θmin, and if | θ | <θmin, step S15
And set g = j, θmin = | θ |
In step 16, 1 is added to j, and the process returns to step S7. Also, | θ
If | <θmin, go to step S16, add 1 to j, and return to step S7. By performing the above processing,
Line number (g) that gives the minimum slope among the lines already extracted
And the inclination (θmin) are obtained.

Next, the inclination θmin is a threshold (θth)
A check is made to see if the value is smaller than θth, and if θmin is greater than θth, a new row is created. Also, θmin is a threshold (θt
h) If smaller, the character group i is registered in the g line.
That is, in step S7, it is checked whether j ≦ C, and if not j ≦ C, the process goes to step S8 and θmin
Check whether <θth. If θmin <θth is not satisfied, the process proceeds to step S10, where 1 is added to C, and the character group i is registered in the Cth line (steps S11 and S12). As a result, the character group i is registered on a new line. Also, θmin
If <θth, the character group i is registered in the g-th line in step S9, 1 is added to i in step S12, and the process returns to step S4. Thereby, the character group i is added to the already registered line. By performing the above processing for all the character groups, each line is extracted, and FIG.
Each line is associated with each character group as shown in FIG.

FIG. 9 is a processing block diagram of a second embodiment of the present invention. This embodiment shows an embodiment for recognizing a table in which ruled lines are largely omitted as shown in FIG. 10. The ruled line extracting unit 16 and the ruled line extracting unit 16 shown in the block diagram of FIG. The removal unit 17 is added. In FIG. 9, first, a ruled line extracting unit 16 extracts ruled lines from a table area, and obtains ruled line information such as its length and position. As the ruled line extraction processing, a known method described in JP-A-9-50527 can be used. Next, the pixels forming the ruled line are erased from the image by the ruled line removing unit 17. The following processing is the same as in the first embodiment, and the character group extracting unit 11 obtains rectangular coordinates surrounding the character group. Next, the character group subdivision unit 12 performs the character group subdivision processing as described in the flowchart of FIG. Next, the row extracting unit 13 performs a row extracting process. As described in the flowchart of FIG. 8 described above, the line extracting unit 13 extracts a line in consideration of the inclination between the character groups.

Next, a column is extracted in a column extracting section 14. The table data creation unit 15 performs character recognition for each character group, and registers the character group in each column and each row. As a result, data that can be written to the table is created as shown in FIG. Then, cell data is created from the relationship between rows and columns, and a character group is registered in the cell data. As described above, in the present embodiment, a table can be recognized even for a table in which ruled lines remain partially. When ruled lines approach each other in the recognized table cells in the up, down, left, and right directions, by adding ruled line information to the cells, when reproducing the table in an application such as a word processor, the table including the ruled lines is displayed. Can be reproduced.

[0017]

As described above, in the present invention,
The following effects can be obtained. (1) It is possible to accurately separate a character group that was difficult to separate by the conventional method of obtaining a character group by integrating characters at a fixed character interval. For this reason, it is possible to accurately recognize a table without ruled lines. (2) Since a character group in which the positional relationship between the character groups is within a certain inclination is registered as a line, the table can be recognized with high accuracy even when the table is inclined. (3) By recognizing the table by removing the ruled lines, the table can be accurately recognized even if the ruled lines are partially omitted.

[Brief description of the drawings]

FIG. 1 is a diagram illustrating an outline of the present invention.

FIG. 2 is a processing block diagram of the first embodiment of the present invention.

FIG. 3 is a diagram illustrating an example of a table to be recognized in an embodiment of the present invention.

FIG. 4 is a conceptual diagram illustrating a method of extracting a tilt between character groups.

FIG. 5 is a conceptual diagram of a column extraction process according to the embodiment of the present invention.

FIG. 6 is a diagram illustrating an example of a row extraction result and a column extraction result, and data that can be written in a table obtained from the result.

FIG. 7 is a flowchart of a character group subdivision process according to the embodiment of the present invention.

FIG. 8 is a flowchart of a row extracting process according to the embodiment of the present invention.

FIG. 9 is a processing block diagram of a second embodiment of the present invention.

FIG. 10 is a diagram showing an example of a table in which ruled lines are largely omitted.

FIG. 11 is a diagram showing an example of a table whose structure can be recognized by a conventional method and an example of a table whose structure cannot be recognized.

[Explanation of symbols]

 11 Character Group Extraction Unit 12 Character Group Subdivision Unit 13 Row Extraction Unit 14 Column Extraction Unit 15 Table Data Creation Unit 16 Rule Line Extraction Unit 17 Rule Line Removal Unit

Claims (6)

[Claims] 1. A method for recognizing a table structure from a table image without ruled lines, wherein a character group is obtained by integrating characters having a fixed character spacing within the table, and a character spacing of characters in the character group is obtained. The character group is subdivided based on the average value, and the character group in which the positional relationship between the subdivided character groups is within a certain slope is registered as a line. A table recognition method characterized by registering, as a column, a character group having a relationship in which horizontal coordinates of the character groups overlap each other. 2. The table recognition method according to claim 1, wherein the table recognition according to claim 1 is performed after removing a ruled line included in the image. 3. A program for recognizing a table from a table image having no ruled lines, wherein the program obtains a character group by integrating characters in the table with a character spacing of less than a predetermined value. A process of re-dividing the character group based on the average value of the character spacing of the character group, and registering a character group in which the positional relationship between the re-divided character groups is within a certain slope as a line, A table recognition program for causing a computer to execute a process of registering, as a column, a character group having a relationship in which the horizontal coordinates of the subdivided character groups overlap each other. 4. A program for recognizing a table structure according to claim 3, wherein a ruled line included in the image is removed before performing the table structure recognition. 5. A recording medium recording a program for recognizing a table from a table image without ruled lines, wherein the program integrates characters in the table with a character interval within a certain range to obtain a character group. An average value of the character spacing of the characters in the character group is obtained, the character group is subdivided based on the average value, and a character group in which the positional relationship between the subdivided character groups is within a certain inclination is registered as a line. A recording medium storing a table recognition program, wherein a character group having a relationship in which the horizontal coordinates of the subdivided character groups overlap each other is registered as a column. 6. A recording medium storing a table structure recognition program according to claim 3, wherein the ruled line included in the image is removed before the table structure recognition is performed.