JP2012190313A - Image processing device and program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide an image processing device capable of generating a feature character string reflecting a content of a document.SOLUTION: An appearance frequency order candidate extraction section 32 extracts candidates of a constitution character string for constituting the feature character string based on an order of appearance frequencies in a document of character strings extracted by a character string extraction section 310. An arrangement order candidate extraction section 34 extracts candidates of the constitution character string for constituting the feature character string based on an arrangement of the character strings in the document extracted by the character string extraction section 310. A feature character string extraction section 36 selects, based on character string order information from the appearance frequency order candidate extraction section 32 and arrangement order information from the arrangement order candidate extraction section 34, two or more character strings from the candidates of the character string included in at least one of the character string order information and the arrangement order information and connects them together to generate the feature character string.

Description

  The present invention relates to an image processing apparatus and a program.

Patent Document 1 discloses an information processing apparatus that generates a file name by adding a predetermined delimiter between character strings of input index information.
Patent Document 2 discloses an image reading apparatus that reads a document image, recognizes characters from the read image, and uses a character string having a high appearance frequency as a file name for the document based on the recognition result.
Patent Document 3 discloses a system that executes inference by inputting layout information, font size information, and appearance frequency information of a target document using knowledge including rules for each document format.

Japanese Patent Laid-Open No. 10-289137 JP 2006-211261 A JP 2006-309347 A

  An object of the present invention is to provide an image processing apparatus capable of generating a characteristic character string reflecting the contents of a document.

  According to the first aspect of the present invention, a character string extracting unit that extracts a plurality of character strings from reading information obtained by a reading unit that reads a document, and a character in the document from the character string extracted by the character string extracting unit. First extraction means for extracting one or more first character string candidates constituting a characteristic character string related to the document based on the appearance frequency of the string, and the character string extracted by the character string extraction means, Second extraction means for extracting one or more second candidates for character strings constituting the characteristic character string based on the arrangement of the character strings; the first candidate extracted by the first extraction means; An image processing apparatus comprising: a feature character string generation unit that selects and connects two or more character strings from at least one of the second candidates extracted by the two extraction units and generates the feature character string.

  According to the second aspect of the present invention, the characteristic character string generation unit selects two or more character strings having different meanings from each other, or two or more character strings having no same meaning from each other, and The image processing apparatus according to claim 1, wherein a characteristic character string is generated.

  According to a third aspect of the present invention, the characteristic character string generation unit determines an order in which two or more selected character strings are connected based on attributes of each of the two or more selected character strings. The image processing apparatus according to Item 1.

  The present invention according to claim 4 is characterized in that the characteristic character string generation means selects and connects two or more character strings so that the number of characters of the characteristic character string is within a predetermined number. An image processing apparatus.

  According to a fifth aspect of the present invention, the first extracting means weights a character string composed of a plurality of words for the extracted first candidate by weighting a character string composed of one word. 2. The image processing apparatus according to claim 1, wherein the characteristic character string generation unit preferentially selects a first candidate having a large weight by the first extraction unit.

  According to a sixth aspect of the present invention, in the image processing apparatus according to the first aspect, the characteristic character string generation unit determines an order in which the selected character strings are connected based on the type of document.

  According to a seventh aspect of the present invention, in the feature character string generation means, when both the first candidate and the second candidate do not include a character string related to the document type, the character related to the document type The image processing apparatus according to claim 1, wherein the characteristic character string is generated so as to include a string.

  According to an eighth aspect of the present invention, there is provided a character string extraction step for extracting a plurality of character strings from read information obtained by a reading means for reading a document, and characters in the document from the character strings extracted in the character string extraction step. A first extraction step for extracting one or more first character string candidates constituting a character string related to a document based on the appearance frequency of the column, and a character string extracted in the character string extraction step, A second extraction step of extracting one or more second candidates of character strings constituting the characteristic character string based on the arrangement of the character strings; the first candidate extracted in the first extraction step; 2. A feature character string generation step for selecting and connecting two or more character strings from at least one of the second candidates extracted in the two extraction steps to generate the feature character string. An image processing program for executing the up to the computer.

  According to the first aspect of the present invention, it is possible to provide an image processing apparatus capable of generating a characteristic character string reflecting the contents of a document.

  According to the second aspect of the present invention, in addition to the effect obtained by the first aspect of the present invention, it is possible to prevent generation of characteristic character strings in which words having the same meaning are duplicated.

  According to the third aspect of the present invention, in addition to the effect obtained by the first aspect of the present invention, it is possible to generate a character string that has a good appearance as compared with the case where the present configuration is not provided.

  According to the fourth aspect of the present invention, in addition to the effect obtained by the first aspect of the present invention, it is possible to generate a character string having a good appearance as compared with the case where the present configuration is not provided.

  According to the fifth aspect of the present invention, in addition to the effect obtained by the first aspect of the present invention, it is easier to include a compound word in the characteristic character string as compared with the case where the configuration is not provided. Can do.

  According to the present invention of claim 6, in addition to the effect obtained by the present invention of claim 1, it is possible to generate a characteristic character string reflecting the contents of the document.

  According to the present invention of claim 7, in addition to the effect obtained by the present invention of claim 1, it is possible to generate a characteristic character string reflecting the contents of the document.

  According to the eighth aspect of the present invention, it is possible to provide an image processing program capable of generating a characteristic character string reflecting the contents of a document.

It is a figure which illustrates the hardware constitutions of the image processing apparatus concerning this embodiment. 2 is a processing program that operates in the image processing apparatus illustrated in FIG. 1. It is a figure which shows the structure of the appearance frequency order candidate extraction part shown in FIG. It is a figure which shows the structure of the arrangement | positioning order candidate extraction part shown in FIG. It is a figure which shows the structure of the characteristic character string production | generation part shown in FIG. It is a figure which illustrates classification standard information. It is a flowchart which shows the process of a processing program. It is a flowchart which shows the process of a processing program. FIG. 4 is a diagram illustrating an example of a document that is a processing target of the image processing apparatus according to the present embodiment.

FIG. 1 is a diagram illustrating a hardware configuration of an image processing apparatus 2 according to the present embodiment.
As illustrated in FIG. 1, the image processing apparatus 2 includes a control device 21 including a calculation unit 212 such as a CPU and a storage unit 214 such as a memory, a communication device 22, a recording device 24, and a user interface device (UI). Device) 25, a printing device 26, and an image reading device 27.

The UI device 25 includes a display device such as an LCD (Liquid Crystal Display) display device or a CRT (Cathode Ray Tube) display device, and a keyboard / touch panel.
The printing device 26 is, for example, a printer, and prints character data or image data on a recording medium such as paper.
The image reading device 27 is a scanner or the like, for example, and reads an image or the like from a recording medium such as a document, and converts it into, for example, read information in a bitmap format.
That is, the image processing apparatus 2 has a hardware configuration part as a computer capable of information processing and communication with other image processing apparatuses or terminals.
In the following drawings, substantially the same components and processes are denoted by the same reference numerals.
In the present embodiment, the image processing apparatus 2 includes the printing apparatus 26 and the image reading apparatus 27. However, the image processing apparatus may be, for example, a PC that does not include the printing apparatus and the image reading apparatus. In this case, the image processing apparatus may be connected to the image reading apparatus via a LAN (Local Area Network) or the like.

FIG. 2 is a diagram showing the configuration of the processing program 3 that operates in the image processing apparatus 2 shown in FIG.
As shown in FIG. 2, the processing program 3 includes a document reading information receiving unit 302, an automatic generation necessity specifying unit 304, a character number setting unit 306, a layout analysis unit 308, a character string extraction unit 310, a language determination unit 312, a document classification. 314, classification reference storage unit 316, appearance frequency order candidate extraction unit 32, arrangement order candidate extraction unit 34, and characteristic character string extraction generation unit 36.
The processing program 3 is supplied to the image processing apparatus 2 via, for example, the storage medium 240 (FIG. 1), loaded into the storage unit 214, and installed on the image processing apparatus 2 on the OS (not shown). It is executed by specifically using the hardware resources of the processing device 2.
In the present embodiment, the processing program 3 is realized by software, but all or part of the processing program 3 may be realized by hardware such as an FPGA (Field Programmable Gate Array). .

FIG. 3 is a diagram illustrating a configuration of the appearance frequency order candidate extraction unit 32 illustrated in FIG. 2.
As shown in FIG. 3, the appearance frequency order candidate extraction unit 32 includes a frequency calculation unit 322, a composite character string determination unit 324, a character string arrangement determination unit 326, a character string rank determination unit 328, and a rank reference storage unit 330. The
FIG. 4 is a diagram illustrating a configuration of the arrangement order candidate extraction unit 34 illustrated in FIG. 2.
As illustrated in FIG. 4, the arrangement order candidate extraction unit 34 includes a character string position determination unit 342, a character string size determination unit 344, an arrangement order candidate determination unit 326, and a scoring reference storage unit 348.

FIG. 5 is a diagram showing a configuration of the characteristic character string generation unit 36 shown in FIG.
As shown in FIG. 5, the characteristic character string generation unit 36 includes an arrangement order candidate storage unit 360, an appearance frequency order candidate storage unit 362, a document type character string storage unit 364, an arrangement order candidate division unit 366, and an arrangement order candidate selection unit. 368, an appearance frequency order candidate selection unit 370, a constituent character string determination unit 372, a synonym dictionary database (DB) 374, a character string attribute determination unit 376, and a character string concatenation unit 382.

In the processing program 3 (FIG. 2), the document reading information receiving unit 302 receives the reading information (document reading information) obtained from the image reading device 27, and the received document reading information is extracted from the arrangement analyzing unit 308 and the character string extraction. It is stored so as to be provided for processing by the unit 310.
The automatic generation necessity designation unit 304 is configured so that the image processing apparatus 2 automatically generates a characteristic character string related to a document corresponding to the document reading information received by the document reading information receiving unit 302, or the user performs, for example, It is specified whether to create the UI device 25 by operating it.
Specifically, the user designates whether the image processing device 2 automatically generates a feature character string or the user creates a feature character string by operating the UI device 25 and uses it. The information (automatic generation necessity information) generated by the user operating the UI device 25 is received by the automatic generation necessity specifying unit 304.
Then, the automatic generation necessity specifying unit 304 outputs the received automatic generation necessity information to the characteristic character string generation unit 36.
Here, the “character string” is a character string for a human to identify a document. For example, when the document is converted into electronic data (electronic file) or the like, the electronic data or the electronic data is converted. The name of a path folder (directory) or the like to be stored.

The character number setting unit 306 sets the number of characters (length) of the characteristic character string.
Specifically, for example, the user specifies the number of characters (set number of characters) of the characteristic character string set by operating the UI device 25, and the information (character number information) generated by the operation is set to the number of characters. Part 306 accepts.
Then, the character number setting unit 306 outputs information indicating the number of set characters corresponding to the character number information (set character number information) to the characteristic character string generation unit 36.

The layout analysis unit 308 analyzes the document reading information, classifies natural images such as characters, tables, and photographs, CG (Computer Graphics), or paintings included in the document (object classification), and associates position information with each of them. .
Further, the arrangement analysis unit 308 outputs information (arrangement information) indicating the analysis result to the appearance frequency order candidate extraction unit 32, the arrangement order candidate extraction unit 34, and the document classification unit 314.
Here, the arrangement information is information indicating which object (natural image such as a character, a table, or a photograph, a CG or a painting) is included at which position and in which position in the document corresponding to the document reading information. is there.
This “placement information” includes, for example, position information indicating the position of each object and scale information indicating the scale (size, area, etc.) of each object.

Here, for example, the position information may indicate an absolute position such as position coordinates, or may indicate a relative positional relationship with another character string or the like.
Similarly, the size information may indicate an absolute size of the object, such as a font or an occupied area, or may indicate a relative size between other objects. Alternatively, it may indicate a difference from the average value of the object size.
The above-described classification by the layout analysis unit 308 is performed by, for example, detecting various lines, frame lines, ruled lines, or color information included in the document, edge detection, and pattern matching. However, it is not restricted to these methods.

The character string extraction unit 310 analyzes the document reading information by using, for example, an OCR (Optical Character Recognition) function, and converts the character string included in the document into a predetermined character string alone by, for example, morphological analysis. Extract in a form with meaning.
Here, the character recognition refers to generating character data (character string) by specifying the character by comparing the image data of the character obtained by reading with a pattern stored in advance.
The morpheme analysis is, for example, classifying one sentence into morphemes (the smallest language unit having meaning) based on information on grammatical rules stored in advance and a dictionary in which words are registered. This is a process for discriminating the part of speech of a morpheme.
In this morphological analysis process, the language of the character string is also determined (for example, whether the character string is Japanese, English, or another language).
Further, the character string extraction unit 310 outputs the extracted character strings to the appearance frequency order candidate extraction unit 32, the arrangement order candidate extraction unit 34, and the document classification unit 314.

The language determination unit 312 analyzes the character string extracted by the character string extraction unit 310 and determines in which language the document is configured.
Specifically, for example, the language determination unit 312 determines the language having the highest ratio in each character string in the document (that is, the language that appears most frequently in the document) as the language of the document, and determines the determination result. The indicated information (language information) is output to the characteristic character string generation unit 36.
In this embodiment, the language determination unit 312 determines the language of the document by analyzing the character string extracted by the character string extraction unit 310. However, for example, the user operates the UI device 25. The language of the document may be determined by manual input or selection from a list.

Based on the arrangement information from the arrangement analysis unit 308 and the information from the character string extraction unit 310, the document classification unit 314 performs document manuscript according to information (classification standard information) related to the classification standard stored in the classification standard storage unit 316. Determine the type.
Further, the document classification unit 314 generates information indicating the determination result (document type information) and outputs it to the characteristic character string generation unit 36.
The classification reference storage unit 316 stores the classification reference information illustrated in FIG.

FIG. 6 is a diagram illustrating classification standard information.
The classification reference information is information (table) indicating the relationship between the document type and the condition for determining the document type, and indicates the number of points given to each document type when each condition is met. ing. Each condition and each document type are not limited to those illustrated in FIG.
The document classification unit 314 (FIG. 2) determines each condition based on the information from the layout analysis unit 308 and the information from the character string extraction unit 310, and if the conditions match, a predetermined score is assigned to each document type. To grant.
Then, the document classification unit 314 determines the document type with the highest total score as the document type related to the document reading information.
For example, a case will be described in which a character string “application form” exists in the upper center of the manuscript, the size (font size, etc.) is equal to or larger than a predetermined size, and a table exists.
In this case, in the example shown in FIG. 6, the total score for the type “application” is 35 points, the total score for the type “approval” is 5 points, and the type “design drawing”. The total score is 5 points.
Therefore, the document classification unit 314 determines that the “application” having the highest score is the type of the document.

Note that the information indicating the position such as “upper” and “center” in the example of FIG. 6 may be represented by absolute position information such as position coordinates, or a relative position with other character strings or the like. It may indicate a relationship.
Further, when the automatic generation necessity information received by the automatic generation necessity designation unit 304 does not indicate that the characteristic character string is automatically generated by the image processing apparatus 2 (that is, when the user selects). The document classification unit 314 may be configured not to perform processing.
Furthermore, in the present embodiment, the document classification unit 314 determines the type of document, but the user may specify the type of document.

The appearance frequency order candidate extraction unit 32 extracts a character string (constituent character string) candidate constituting the characteristic character string based on the order of appearance frequency of the character string extracted by the character string extraction unit 310 in the document. The constituent character string candidates extracted by the appearance frequency order candidate extraction unit 32 are referred to as appearance frequency order candidates.
In the appearance frequency order candidate extraction unit 32, the frequency calculation unit 322 (FIG. 3) calculates the number of appearances (appearance frequency) for each character string extracted by the character string extraction unit 310, and the character string and the character string. Are associated with the appearance frequency and output to the character string rank determination unit 328.

The composite character string determination unit 324 determines whether each character string extracted by the character string extraction unit 310 is a composite character string, for example, by morphological analysis.
Further, when it is determined that the character string is a composite character string, the composite character string determination unit 324 determines information indicating that the character string is a composite character string (composite character string information) as a character string rank determination. To the unit 328.
Here, the “composite character string” is a character string composed of a plurality of words.
For example, since the character string “market size” includes two words “market” and “scale”, it is determined as a composite character string.

The character string arrangement determination unit 326 determines whether each character string extracted by the character string extraction unit 310 is a character string included in a predetermined object based on the arrangement information.
When it is determined that the character string is included in a specific object, the character string arrangement determining unit 326 sends information indicating that fact (specific arrangement character string information) to the character string rank determining unit 328. Output.

Based on the information from the frequency calculation unit 322 and the composite character string information from the composite character string determination unit 324, the character string order determination unit 328 includes information (ranking information) related to the ranking criteria stored in the ranking reference storage unit 330. The order of the character strings is determined according to the reference information).
Further, the character string rank determination unit 328 generates information indicating the determination result (character string rank information) and outputs the information to the characteristic character string generation unit 36.
The rank reference information stored in the rank reference storage unit 330 indicates, for example, a reference that gives a higher score for each character string as the appearance frequency is higher.
Further, the rank reference information may indicate that when a certain character string is a character string related to the composite character string information, the number of points given to the character string is increased.

Further, the ranking reference information may indicate that when a certain character string is a character string related to the specific arrangement character string information, the number of points given to the character string is reduced.
For example, when the character string arrangement determining unit 326 determines that a certain character string is included in the table in the document, the ranking reference information may indicate that the score given to the character string is 0.
Further, for example, when the character string arrangement determining unit 326 determines that a certain character string is included in the table in the document, the ranking reference information is displayed in the table from the character string appearance frequency calculated by the frequency calculating unit 322. It may indicate that the number of the character strings determined to be included is subtracted.

The processing of the character string rank determination unit 328 and the rank reference information will be described with a specific example.
For example, the ranking reference information gives 10 points to the character string with the first appearance frequency, gives 8 points to the character string with the second appearance frequency, and gives 4 points to the character string with the third appearance frequency. It is assumed that 3 points are given to the character string having the fourth appearance frequency, and that the number of points given when the character string is a composite character string is increased by 5 times.
For example, the calculation result of the frequency calculation unit 322 is
“Scale”: 10 pieces, “Market”: 8 pieces, “Market scale”: 4 pieces, “Scale expansion”: 3 pieces.

In this case, depending on the appearance frequency, the rank and score of the appearance frequency of each character string is:
1st place: “Scale” (10 points), 2nd place: “Market” (8 points), 3rd place: “Market size” (4 points), 4th place: “Scale expansion” (3 points)
It is.
Here, since the character strings “market scale” and “scale expansion” are determined to be composite character strings by the composite character string determination unit 324, the character string rank determination unit 328 determines the character strings “market scale” and “scale”. The number of points given to “enlargement” is increased 5 times.
Therefore, the character string rank determination unit 328
1st place: “market size” (20 points), 2nd place: “scale expansion” (15 points), 3rd place: “scale” (10 points), 4th place: “market” (8 points)
Is determined.

The arrangement order candidate extraction unit 34 (FIG. 2) extracts constituent character string candidates constituting the characteristic character string based on the arrangement of the character string extracted by the character string extraction unit 310 in the document. The constituent character string candidates extracted by the arrangement order candidate extraction unit 34 are referred to as arrangement order candidates.
In the arrangement order candidate extraction unit 34, the character string position determination unit 342 (FIG. 4) determines the position of each character string extracted by the character string extraction unit 310 based on the arrangement information from the arrangement analysis unit 308.
Furthermore, the character string position determination unit 342 associates each character string with position information regarding the character string, and outputs the associated character string position determination unit 342 to the arrangement order candidate determination unit 326.

The character string size determination unit 344 determines the size of each character string extracted by the character string extraction unit 310 based on the arrangement information from the arrangement analysis unit 308.
Furthermore, the character string scale determination unit 344 associates each character string with the scale information related to the character string and outputs the associated character string scale determination unit 344 to the arrangement order candidate determination unit 326.

Based on the information from the character string position determining unit 342 and the information from the character string size determining unit 344, the arrangement order candidate determining unit 326 is information on the ranking criteria stored in the scoring criterion storage unit 348 (scoring criteria). The order based on the arrangement of each character string is determined according to (information).
Further, the arrangement order candidate determination unit 326 generates information indicating the determination result (arrangement order information) and outputs the information to the characteristic character string generation unit 36.
The scoring standard information stored in the scoring standard storage unit 348 indicates, for example, the number of points given to a character string when the position of each character string is relatively upward and relatively central in the document. Indicates to increase.
In addition, the scoring reference information indicates that when the scale of each character string is relatively large in the document, for example, the font of the character string is large, the number of points assigned to the character string is increased.

The process of the arrangement order candidate determination unit 326 and the point allocation reference information will be described with specific examples.
For example, if the scoring reference information indicates that 10 points are given to a character string above a predetermined position in the document, and 5 points are given to a character string located in the center in the horizontal direction from the predetermined position in the document. To do.
Further, for example, 10 points are assigned to a character string having a scale that is 5 times or more of the average value of the character strings in the original, and the scoring reference information is 2 times or more and less than 5 times the average value of the character strings. Suppose that 8 points are given to a character string of a certain scale.
Further, for example, the position of the character string “estimate” in the manuscript is above the predetermined position and in the center in the horizontal direction from the predetermined position, and the font size of the character string “estimate” is the average font size. Suppose that it is 5 times.

On the other hand, for example, in the manuscript, the position of the character string “market” is below the predetermined position but at the center in the horizontal direction from the predetermined position, and the font size of the character string “market” is the average font size. Suppose that it is 3 times.
In this case, the score given to the character string “estimate” is 10 + 5 + 10 = 25 points, and the score given to the character string “market” is 0 + 5 + 8 = 13 points.
Therefore, the arrangement order candidate determination unit 326
1st place: “Estimate” (25 points) 2nd place: “Market” (13 points)
Is determined.

The feature character string extraction generation unit 36 is based on the character string rank information from the appearance frequency order candidate extraction section 32 and the arrangement rank information from the arrangement order candidate extraction section 34, and at least one of the character string rank information and the arrangement rank information. Two or more character strings are selected from the character string candidates included in and connected to generate a characteristic character string.
In the characteristic character string generation unit 36, the arrangement order candidate storage unit 360 (FIG. 5) stores the arrangement order information from the arrangement order candidate extraction unit 34.
Character string order information from the appearance frequency order candidate storage unit 362 and the appearance frequency order candidate extraction unit 32 is stored.
Document type character string storage unit 364 stores a character string (document type character string) corresponding to the document type included in the document type information from document classification unit 314.

If there is a character string having a character number longer than the character number set by the character number setting unit 306 among the character strings related to the arrangement order information stored in the arrangement order candidate storage unit 360, the arrangement order candidate dividing unit 366 For example, the character string alone is divided into a form having a predetermined meaning by morphological analysis.
The arrangement order candidate selection unit 368 selects each of the character strings (arrangement order candidates) included in the arrangement order information in descending order, and outputs the selected character strings to the constituent character string determination unit 372.
The appearance frequency order candidate selection unit 370 selects the character strings included in the character string order information (appearance frequency order candidates) in descending order, and outputs them to the constituent character string determination unit 372.

The arrangement order candidate selection unit 368 receives automatic generation necessity information indicating that the user creates the characteristic character string by operating the UI device 25 from the automatic generation necessity designation unit 304. A list in which arrangement order candidates included in the arrangement order information are arranged in descending order may be transmitted to the UI device 25 such as a display device.
Similarly, the appearance frequency order candidate selection unit 370 receives automatic generation necessity information indicating that the user creates a characteristic character string by operating the UI device 25 from the automatic generation necessity designation unit 304. In addition, a list in which the appearance frequency order candidates included in the character string order information are arranged in descending order may be transmitted to the UI device 25 such as a display device.
The UI device 25 displays a list in which arrangement order candidates are arranged in descending order and a list in which appearance frequency order candidates are arranged in descending order.
In this case, when the user operates the UI device 25, an arrangement order candidate and an appearance frequency order candidate constituting the characteristic character string are selected.

The constituent character string determination unit 372 compares the arrangement order candidate selected by the arrangement order candidate selection unit 368 with the appearance frequency order candidates selected by the appearance frequency order candidate selection unit 370, and each corresponds to the document reading information. It is determined whether or not it is suitable as a constituent character string constituting a characteristic character string related to a document to be printed. Specific processing will be described later.
The synonym dictionary DB 374 stores a synonym dictionary including a list of combinations of character strings that are synonyms, for example.
The character string attribute determining unit 376 determines the attribute of the character string, for example, by morphological analysis.
Here, the attribute of the character string may distinguish, for example, the type of part of speech such as a noun, a verb or an adjective. When the character string is a noun, it distinguishes a common noun or proper noun, etc. Furthermore, when the character string is a proper noun, a name or a place name indicating a specific thing other than a person such as a person name or a corporate name may be distinguished.
Further, when the attribute is a place name, the attribute may distinguish a country name or a region name.

The constituent character string determination unit 372 selects the arrangement order candidate selected by the arrangement order candidate selection unit 368 and the appearance frequency selected by the appearance frequency order candidate selection unit 370 based on the synonym dictionary stored in the synonym dictionary DB 374. It is determined whether or not the order candidates are synonymous with each other (decision 1-1).
Further, the constituent character string determination unit 372 determines whether or not both the arrangement order candidate and the appearance frequency order candidate include character strings that are synonyms based on the synonym dictionary stored in the synonym dictionary DB 374. (Judgment 1-2).
Furthermore, the constituent character string determination unit 372 determines whether the arrangement order candidates or the appearance frequency order candidates are the same character strings, whether the arrangement order candidates or the appearance frequency order candidates include the same character string, Further, it may be determined whether any one of the arrangement order candidate or the appearance frequency order candidate includes the other character string (determination 1-3).

When it is determined that at least one of the above determinations 1-1 to 1-3 is correct, the constituent character string determination unit 372 discards the appearance frequency order candidates while retaining the arrangement order candidates as they are as constituent character string candidates. To do.
Then, the constituent character string determination unit 372 controls the appearance frequency order candidate selection unit 370 so as to select a new appearance frequency order candidate.
In this case, the appearance frequency order candidate selection unit 370 selects the appearance frequency order candidate having the highest rank among the unselected appearance frequency order candidates from the character string order information stored in the appearance frequency order candidate storage unit 362, The data is output to the constituent character string determination unit 372.
Then, the constituent character string determination unit 372 performs the determinations 1-1 to 1-3 for the placed arrangement order candidate and the newly selected appearance frequency order candidate in the same manner as described above.

The constituent character string determination unit 372 outputs the arrangement order candidates and the appearance frequency order candidates to the character string attribute determination unit 376.
The character string attribute determining unit 376 determines whether or not the attribute of the arrangement order candidate and the attribute of the appearance frequency order candidate are the same, and outputs information indicating the determination result to the constituent character string determining unit 372.
The constituent character string determination unit 372 determines whether or not the placement order candidate attribute and the appearance frequency order candidate attribute are the same based on the information indicating the determination result from the character string attribute determination unit 376 (determination). 2).

When it is determined that the above determination 2 is correct, the constituent character string determination unit 372 discards the appearance frequency order candidates while retaining the arrangement order candidates as they are as constituent character string candidates.
Then, the constituent character string determination unit 372 controls the appearance frequency order candidate selection unit 370 so as to select a new appearance frequency order candidate.
In this case, the appearance frequency order candidate selection unit 370 selects the appearance frequency order candidate having the highest rank among the unselected appearance frequency order candidates from the character string order information stored in the appearance frequency order candidate storage unit 362, The data is output to the constituent character string determination unit 372.
Then, the constituent character string determination unit 372 performs the above determination 2 on the placed arrangement order candidate and the newly selected appearance frequency order candidate as described above.

When it is determined that all of the above determinations 1-1 to 1-3 and the above determination 2 are negative, the constituent character string determining unit 372 uses these arrangement order candidates and appearance frequency order candidates as constituent character strings. Output to the character string concatenation unit 382.
The constituent character string determination unit 372 may determine whether the document type character string stored in the document type character string storage unit 364 is included in the placement order candidate and the appearance frequency order candidate.
In this case, when it is determined that one of the arrangement order candidate and the appearance frequency order candidate includes the document type character string, the constituent character string determination unit 372 performs the above determinations 1-1 to 1-3 and the above determination 2. Regardless, the placement order candidate or the appearance frequency order candidate including the document type character string may be output to the character string concatenation unit 382 as a constituent character string.
When it is determined that neither the placement order candidate nor the appearance frequency order candidate includes the document type character string, the constituent character string determination unit 372 determines the document type in addition to the placement order candidate and the appearance frequency order candidate. A character string may be output to the character string concatenation unit 382 as a constituent character string.

Furthermore, the constituent character string determination unit 372 may determine a specific attribute placement order candidate or appearance frequency order candidate as a constituent character string in accordance with the document type character string.
For example, when the document type character string is “application”, the arrangement order candidate or the appearance frequency order candidate having the attribute “person name” is configured regardless of the above-described judgments 1-1 to 1-3 and the above-mentioned judgment 2. You may output to the character string connection part 382 as a character string.
Furthermore, the constituent character string determination unit 372 may appropriately change the determination criterion of the constituent character string to be determined based on the language information from the language determination unit 312.

In addition, when the arrangement order candidate is not extracted by the arrangement order candidate extraction unit 34, the constituent character string determination unit 372 controls the appearance frequency order candidate selection unit 370 so as to select another appearance frequency order candidate.
In this case, the appearance frequency order candidate selection unit 370 selects the appearance frequency order candidate having the highest rank among the unselected appearance frequency order candidates from the character string order information stored in the appearance frequency order candidate storage unit 362. And output to the constituent character string determination unit 372.
Then, similarly to the above, the constituent character string determination unit 372 performs the above determinations 1-1 to 1-3 and the above determination 2 on the original appearance frequency order candidate and the newly selected appearance frequency order candidate. May be.
In addition, according to the determinations in the above determinations 1-1 to 1-3 and the above determination 2, the constituent character string determination unit 372 retains the placement order candidates, but the appearance frequency order candidates are retained and the placement order. A candidate may be newly selected.

First, the character string concatenation unit 382 determines the number of constituent character strings that constitute the characteristic character string.
Specifically, the character string linking unit 382 receives a plurality of constituent character strings from the constituent character string determining unit 372 and calculates the total number of characters (total number of characters) of these constituent character strings.
Further, the character string concatenation unit 382 receives the set character number information from the character number setting unit 306.
Then, the character string concatenation unit 382 determines whether the total number of characters of the constituent character strings is within the set number of characters indicated by the set character string.
When the total number of characters of the constituent character string is within the set number of characters, the character string concatenating unit 382 controls the constituent character string determining unit 372 and the appearance frequency order candidate selecting unit 370 so as to select another appearance frequency order candidate. To do.

In this case, the appearance frequency order candidate selection unit 370 selects the appearance frequency order candidate having the highest rank from the unselected appearance frequency order candidates from the character string order information stored in the appearance frequency order candidate storage unit 362, The data is output to the constituent character string determination unit 372.
Then, the constituent character string determination unit 372 performs the above determinations 1-1 to 1-3 and the above determination 2 for the constituent character string that has been output to the character string concatenation unit 382 and the newly selected appearance frequency order candidates. If the newly selected appearance frequency order candidate is determined as a constituent character string, the constituent character string (appearance frequency order candidate) is output to the character string concatenation unit 382.

On the other hand, when the total number of characters of the constituent character string is not within the set number of characters, the character string concatenating unit 382 discards the constituent character string received from the constituent character string determining unit 372 immediately before.
Through the above processing, the number of constituent character strings constituting the characteristic character string is determined.
The above processing will be specifically described with an example.
For example, the set number of characters is 20, “Address change application” (7 characters) is selected as the placement order candidate, and the appearance frequency order candidate # 1 “Nishi-Yokohama-shi” (5 characters) in order as the appearance frequency order candidate ) Appearance frequency order candidate # 2 “Relocation date” (3 characters), Appearance frequency order candidate # 3 “Household” (3 characters), and Appearance frequency order candidate # 4 “Housemate” (3 characters) To do.
In this case, the total number of characters of the arrangement order candidates and the number of characters of the appearance frequency order candidates # 1 to # 3 is 18 characters, and the total number of characters of the arrangement order candidates and the appearance frequency order candidates # 1 to # 4 is 21 characters. It is.

Therefore, in this case, the character string concatenation unit 382 discards the constituent character string (appearance frequency order candidate # 4) “cohabitant” received from the constituent character string determination unit 372 immediately before, and arranges the arrangement order candidate and the appearance frequency order candidate. # 1 to # 3 are determined as constituent character strings to be concatenated.
Therefore, in this case, the character string coupling unit 382 determines the number of constituent character strings constituting the characteristic character string as four (placement order candidates and appearance frequency order candidates # 1 to # 3).

When generating a characteristic character string, a delimiter such as “-(hyphen)” or “_ (under bar)” may be inserted between a plurality of constituent character strings.
In this case, the character string concatenation unit 382 may add the number of insertion symbols to the total number of characters in the comparison between the total number of characters and the set number of characters.
Further, in the above embodiment, when the total number of characters is within the set number of characters, the constituent string determining unit 372 and the appearance frequency order candidate selection so that the character string concatenation unit 382 selects another appearance frequency order candidate. Although the unit 370 is controlled, the character string coupling unit 382 may control the constituent character string determination unit 372 and the arrangement order candidate selection unit 368 so as to select another arrangement order candidate.

Next, the character string concatenation unit 382 determines the order in which a plurality of constituent character strings are concatenated.
The character string concatenation unit 382 concatenates the constituent character string that is the candidate for the appearance frequency and the constituent character string that is the candidate for the arrangement order to the beginning of the characteristic character string.
Here, when there are a plurality of appearance frequency order candidates, the character string concatenation unit 382 connects the appearance frequency order candidates with a higher appearance frequency so that they appear earlier.
In addition, when the arrangement order candidates are not determined as the constituent character strings, the character string concatenation unit 382 connects the appearance frequency order candidates with a higher appearance frequency in front.

Through the above processing, the character string concatenation unit 382 concatenates a plurality of constituent character strings to generate a characteristic character string.
Further, the character string coupling unit 382 transmits the generated characteristic character string to the UI device 25, and the characteristic character string is displayed on the UI device 25.
Note that when the character string concatenation unit 382 accepts the document classification character string as the constituent character string, the character string connection character string 382 may use the document classification character string as the head of the characteristic character string.
In addition, when the character string linking unit 382 accepts an arrangement order candidate or an appearance frequency order candidate including a document type character string as a constituent character string, the character string connection unit 382 selects an arrangement order candidate or an appearance frequency order candidate including the document type character string. It may be the beginning of the feature character string.

The character string concatenation unit 382 determines the concatenation order of the constituent character strings based on the attributes of the constituent character strings regardless of whether the constituent character strings are placement order candidates or appearance frequency order candidates. May be.
For example, if there is a configuration character string of attribute “place name” and a configuration string of attribute “person name”, the configuration string of attribute “location name” is concatenated before the configuration string of attribute “person name”. Also good.
Further, the character string concatenation unit 382 may determine the concatenation order according to the language of the document regardless of whether the constituent character string is an arrangement order candidate or an appearance frequency order candidate.
For example, when there is a constituent character string of the attribute “country name” and a constituent character string of the attribute “region name”, and the language information from the language determination unit 312 indicates “Japanese”, a constituent character string of the attribute “country name”. Are combined before the component string of the attribute "Region name", and the language information indicates "English", the component string of the attribute "Region name" is concatenated before the component string of the attribute "Country name" You may make it do.
Further, the character string concatenation unit 382 determines the concatenation order of the constituent character strings based on the manuscript type character strings regardless of whether the constituent character strings are placement order candidates or appearance frequency order candidates. Also good.

7A and 7B are flowcharts (S10) showing the processing of the processing program 3.
In step 102 (S102), document reading information receiving unit 302 receives document reading information obtained by reading a document.
In step 104 (S104), the layout analysis unit 308 generates layout information based on the document reading information, and the character string extraction unit 310 extracts a character string.

In step 106 (S106), the appearance frequency order candidate extraction unit 32 extracts appearance frequency order candidates based on the order of appearance frequencies in the document.
In step 108 (S108), the arrangement order candidate extraction unit 34 extracts arrangement order candidates based on the arrangement of the character strings in the document.
In step 110 (S110), the automatic generation necessity designation unit 304 determines whether or not the image processing apparatus 2 automatically generates the characteristic character string, and determines that the setting for automatic generation is made. If so, the process proceeds to S120, and if not (that is, if the user has made settings for example by operating the UI device 25), the process proceeds to S112.
In step 112 (S112), the UI device 25 displays the list of appearance frequency order candidates extracted by the appearance frequency order candidate extraction unit 32 and the arrangement order candidates extracted by the arrangement order candidate extraction unit 34.
In step 114 (S114), by selecting the appearance frequency order candidate and the arrangement order candidate by the user, the characteristic character string is selected, and the process ends.

In step 120 (S120), the document classification unit 314 determines the type of document according to the classification reference information.
In step 122 (S122), the characteristic character string generation unit 36 determines whether or not the number of arrangement order candidates is not 0. If not, the process proceeds to S126. Proceed to S124.
In step 124 (S124), the characteristic character string generation unit 36 determines whether or not the number of appearance frequency order candidates extracted is not 0. If not, the process proceeds to S130. Ends.

In step 126 (S126), the arrangement order candidate selection unit 368 of the characteristic character string generation unit 36 selects the arrangement order candidates in descending order.
In step 128 (S128), the characteristic character string generation unit 36 determines whether or not the number of appearance frequency order candidates extracted is not 0. If not, the process proceeds to S130. Ends.
In step 130 (S130), the appearance frequency order candidate selection unit 370 of the characteristic character string generation unit 36 selects the appearance frequency order candidates in descending order.

In step 142 (S142), the constituent character string determination unit 372 of the characteristic character string generation unit 36 determines whether or not the selected arrangement order candidate and appearance frequency order candidate are synonymous with each other. If it is determined that there is, the process proceeds to S146, and if it is determined that it is not a synonym, the process proceeds to S142.
In step 144 (S144), the constituent character string determination unit 372 of the characteristic character string generation unit 36 determines whether the attribute of the placement order candidate and the attribute of the appearance frequency order candidate are the same, and the attributes are the same. If it is determined that they are, the process proceeds to S146. If it is determined that they are not the same, the process proceeds to S148.
In step 146 (S146), the characteristic character string generation unit 36 discards the appearance frequency order candidates that are the processing targets.

In step 148 (S148), the character string concatenation unit 382 of the characteristic character string generation unit 36 determines whether or not the total number of characters of the constituent character string is within the set number of characters. The process proceeds to S130, and if not, the process proceeds to S150.
In step 150 (S150), the character string concatenation unit 382 of the characteristic character string generation unit 36 discards the constituent character string received from the constituent character string determination unit 372 immediately before.
In step 152 (S152), the character string concatenation unit 382 of the characteristic character string generation unit 36 determines the order in which a plurality of constituent character strings are concatenated, concatenates the plurality of constituent character strings, and ends the process. .

Hereinafter, the processing of the image processing apparatus 2 according to the present embodiment will be described with specific examples.
FIG. 8 is a diagram illustrating an example of a document to be processed by the image processing apparatus 2 according to the present embodiment.
In the manuscript illustrated in FIG. 8, a character string “application form” is written in the upper center in a font larger than other character strings, and the character string “application form” is smaller on the right side below the character string “application form”. The font is larger than the text string in the main text, and the text "Mayor of XXX City" is written.
Therefore, the arrangement order candidate extraction unit 34 sequentially extracts the arrangement order candidate # 1 “application form” and the arrangement order candidate # 2 “Mayor XXX City”.
In addition, the appearance frequency order candidate extraction unit 32, in descending order of appearance frequency, appearance frequency order candidate # 1 “ground use”, appearance frequency order candidate # 2 “applicant”, appearance frequency order candidate # 3 “Fujitaro” ”, Appearance frequency order candidate # 4“ XXX city ”, appearance frequency order candidate # 5“ application ”, and appearance frequency order candidate # 6“ mayor ”.

The characteristic character string generation unit 36 first selects the arrangement order candidate # 1 “application form” and selects the appearance frequency order candidate # 1 “ground use”.
The feature character string generation unit 36 does not include the arrangement order candidate # 1 “application” and the appearance frequency order candidate # 1 “ground use”, which are not synonymous with each other and do not include the same synonym, and these attributes are also mutually exclusive. Judged to be different.
Therefore, the characteristic character string generation unit 36 determines the arrangement order candidate # 1 “application” and the appearance frequency order candidate # 1 “ground use” as constituent character strings.

When the set number of characters is 15, the total number of characters of the arrangement order candidate # 1 “application” and the appearance frequency order candidate # 1 “ground use” is 10 characters. Then, appearance frequency order candidate # 2 “applicant” is selected.
Since the arrangement order candidate # 1 “application” and the appearance frequency order candidate # 2 “applicant” both include the character string “application”, the characteristic character string generation unit 36 determines that the arrangement order candidate # 1 “application”. The appearance frequency order candidate # 2 “applicant” is determined to be synonymous with each other.
Therefore, the appearance frequency order candidate # 2 “applicant” is discarded.

Next, the characteristic character string generation unit 36 selects the appearance frequency order candidate # 3 “Taro Fuji”.
The characteristic character string generation unit 36 has the same arrangement order candidate # 1 “application form” and appearance frequency order candidate # 1 “ground use” and appearance frequency order candidate # 3 “Taro Fuji” that have already been determined. It is not a word and does not include the same synonym, and it is determined that these attributes are also different from each other.
Therefore, the characteristic character string generation unit 36 newly determines the appearance frequency order candidate # 3 “Taro Fuji” as a constituent character string.

Since the total number of characters of the arrangement order candidate # 1 “application”, the appearance frequency order candidate # 1 “ground use”, and the appearance frequency order candidate # 3 “Fujitaro” is 14 characters, the characteristic character string generation unit 36 selects appearance frequency order candidate # 4 “XXX city”.
The characteristic character string generation unit 36 includes the already determined placement order candidate # 1 “application”, appearance frequency order candidate # 1 “ground use” and appearance frequency order candidate # 3 “Fuji Taro”, and appearance frequency order candidates. It is determined that # 4 “XXX city” is not synonymous with each other, does not include the same synonym, and these attributes are also different from each other.
However, the number of characters of placement order candidate # 1 “application”, appearance frequency order candidate # 1 “ground use”, appearance frequency order candidate # 3 “Fujitaro”, and appearance frequency order candidate # 4 “XX city” The total is 18 characters, exceeding the set number of characters.
Therefore, the appearance frequency order candidate # 4 “XXX city” is discarded, and the arrangement order candidate # 1 “application”, the appearance frequency order candidate # 1 “ground use”, and the appearance frequency order candidate # 3 “Fujitaro” Is determined as a constituent character string.

Next, the feature character string generation unit 36 starts with placement order candidate # 1 “application”, followed by appearance frequency order candidate # 1 “ground use” and appearance frequency order candidate # 3 “Fujitaro”. In order, these constituent character strings are concatenated.
Therefore, the characteristic character string generation unit 36 generates a characteristic character string “Application form ground use Fuji Taro”.

2 ... Image processing device,
3 ... Processing program,
302: Document reading information receiving unit,
304 ... Automatic generation necessity designation part,
306 ... Number of characters setting section,
308 ... Placement analysis unit,
310 ... character string extraction unit,
312 ... Language determination unit,
314: Document classification section,
316: Classification criteria storage unit,
32 ... Appearance frequency order candidate extraction unit,
322 ... Frequency calculation unit,
324... Composite character string determination unit,
326... Character string arrangement determination unit,
328 ... Character string rank determination unit,
330 ... ranking reference storage unit,
34 ... arrangement order candidate extraction unit,
342 ... character string position determination unit,
344: Character string scale determination unit,
326 ... Arrangement order candidate determination unit,
348 ... Scoring standard storage unit,
36... Character string extraction unit,
360 ... arrangement order candidate storage unit,
362 ... Appearance frequency order candidate storage unit,
364 ... Document type character string storage unit,
366 ... arrangement order candidate division unit,
368 ... Placement order candidate selection unit,
370 ... Appearance frequency order candidate selection unit,
372: Configuration character string determination unit,
374 ... synonym dictionary DB,
376 ... character string attribute determination unit,
382 ... character string concatenation part,

Claims (8)

A character string extracting means for extracting a plurality of character strings from the read information obtained by the reading means for reading an original;
First extraction means for extracting one or more first character string candidates constituting a characteristic character string related to a document from the character string extracted by the character string extraction means based on the appearance frequency of the character string in the document; ,
Second extraction means for extracting one or more second candidates for character strings constituting the characteristic character string from the character strings extracted by the character string extraction means based on the arrangement of the character strings in the document;
Two or more character strings are selected from at least one of the first candidate extracted by the first extracting means and the second candidate extracted by the second extracting means, and the characteristic character string is generated. An image processing apparatus comprising: a characteristic character string generation unit. 2. The feature character string generation unit selects two or more character strings having different meanings from each other, or two or more character strings that do not include words having the same meaning from each other, and generates the feature character string. An image processing apparatus according to 1. The image processing apparatus according to claim 1, wherein the characteristic character string generation unit determines an order in which the two or more selected character strings are connected based on attributes of the two or more selected character strings. The image processing apparatus according to claim 1, wherein the characteristic character string generation unit selects and connects two or more character strings so that the number of characters of the characteristic character string is within a predetermined number. The first extraction means weights the extracted first candidate so that the weight of the character string composed of a plurality of words is larger than the weight of the character string composed of one word,
The image processing apparatus according to claim 1, wherein the characteristic character string generation unit preferentially selects a first candidate having a large weight by the first extraction unit. The image processing apparatus according to claim 1, wherein the characteristic character string generation unit determines an order of connecting the selected character strings based on a type of document. The feature character string generation unit may include the feature character string so as to include a character string related to the document type when neither the first candidate nor the second candidate includes a character string related to the document type. The image processing apparatus according to claim 1, wherein a sequence is generated. A character string extraction step for extracting a plurality of character strings from the read information obtained by the reading means for reading the document;
A first extraction step of extracting one or more first character string candidates constituting a characteristic character string related to a document from the character string extracted in the character string extraction step based on the appearance frequency of the character string in the document; ,
A second extraction step of extracting one or more second character string candidates constituting the characteristic character string from the character string extracted in the character string extraction step based on the arrangement of the character string in the document;
Two or more character strings are selected from at least one of the first candidate extracted in the first extraction step and the second candidate extracted in the second extraction step, and the characteristic character string is generated. An image processing program for causing a computer to execute a characteristic character string generation step.

Images