JP2006277104A - Image reading device, extraction method for dictionary registration object word/phrase and program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an image reading device for extracting not only words but also their attribute information from a document. <P>SOLUTION: This image reading device is provided with an image reading means for reading the image of an original, and for generating input image data, a layout analyzing means for generating layout information from the input image data, an image dividing means for dividing the input image data into a plurality of small regions based on the layout information, a data base in which an identifier for specifying the small region having a title character string/image among the plurality of small regions and its title character string/image and an identifier for specifying the small region having an information character string/image and its information character string/image are stored so as to be associated with each other, an information extracting means for extracting registration object words and phrases from the small region having the title character string/image, and for extracting the attribute information from the small region having the information character string/image based on the information stored in the data base and an output means for outputting the dictionary registration object words and phrases. <P>COPYRIGHT: (C)2007,JPO&INPIT

Description

  The present invention relates to a technique for generating and updating a dictionary by extracting words and idioms from a document.

  With the spread of activities across borders such as global information distribution and economic activities in recent years, the demand for translation from one language (for example, English) to another language (for example, Japanese) is increasing. However, when a translation is requested from a trader (translator), the price is generally high and it takes time. Therefore, the demand for an automatic translation (machine translation) apparatus that automatically translates using a machine such as a computer is increasing.

  Dedicated dictionaries such as proper noun dictionaries and technical term dictionaries are used to accurately translate proper nouns such as company names, personal names, and product names in the text, as well as technical terms in specific technical fields. It is common. These dictionaries are often created by human hands. Also, especially for information that is constantly updated, such as proper nouns, it is necessary to update the dictionary (maintenance), but this is also generally done manually.

By the way, there may be a document in which information is arranged according to a predetermined layout, such as a book in which company information is described or a certain form owned by each company. It would be convenient if information could be extracted from such a document and used to update the dictionary. As a technique for extracting information from a document having a predetermined layout in this way, there is a technique described in Patent Document 1. Patent Document 1 discloses a technique for acquiring information necessary for creating a delivery slip based on the layout of an application form and creating a slip or updating a customer database.
Japanese Patent Application Laid-Open No. 5-298341

However, according to the technique described in Patent Document 1, there is a problem that although it is possible to extract words from a document, it is difficult to extract additional attribute information.
The present invention has been made in view of the above circumstances, and an object thereof is to provide an image reading apparatus that can extract additional attribute information in addition to words from a document.

In order to solve the above problems, the present invention provides an image reading unit that reads an image of a document and generates input image data, a layout analysis process for the input image data generated by the image reading unit, and layout information Layout analyzing means for generating the image, image dividing means for dividing the input image data into a plurality of small areas based on the layout information generated by the layout analyzing means, and a heading character string among the plurality of small areas Alternatively, a first identifier that identifies a small area having a heading image and its heading character string or heading image, and a second identifier that identifies a small area having an information character string or information image, and its information character string and information image And a layout database stored in association with each other and a first stored in the layout database Information extracting means for extracting a registration target phrase from a small area specified by an identifier, and attribute information of the dictionary registration target phrase from a small area specified by a second identifier stored in the layout database; and the information extraction An image reading apparatus having output means for outputting a dictionary registration target phrase extracted by the means is provided.
According to this image reading apparatus, information (registration target words and their attribute information) necessary for automatically updating a dictionary is automatically output from a document based on the layout of the document.

In a preferred aspect, the image reading apparatus includes a definition storage unit that stores definitions of a heading character string or heading image, an information character string or an information image, and the plurality of small subordinates according to the definitions stored in the definition storage unit. Among the regions, small region specifying means for specifying a small region having a heading character string or a heading image, and a small region having an information character string or an information image, and information on the small region specified by the small region specifying unit And a database updating means for updating the contents of the layout database.
According to this image reading apparatus, since the layout database is automatically updated, a layout database corresponding to the layout of the document to be processed can be automatically created.

Further, the present invention provides an image reading step for reading an image of a document and generating input image data, a layout analysis step for performing layout analysis processing on the input image data to generate layout information, and the layout information. An image dividing step of dividing the input image data into a plurality of small areas, and a first identifier for identifying a small area having a heading character string or a heading image and the heading character string or It is specified by the first identifier stored in the layout database that stores the header image, the information character string or the small identifier having the information image, and the information identifier and the information image stored in association with each other. The registration target word / phrase is specified from the small area by the second identifier stored in the layout database. There is provided a method for extracting a dictionary registration target phrase including an information extraction step of extracting attribute information of the dictionary registration target phrase from a small area and an output step of outputting the dictionary registration target phrase extracted in the information extraction step. .
The present invention also provides a program for causing a computer device to execute the above-described dictionary registration target phrase extraction method.

Hereinafter, an embodiment of the present invention will be described with reference to the drawings.
FIG. 1 is a block diagram showing a functional configuration of a dictionary update system 1 according to an embodiment of the present invention. The image reading unit 10 reads an image of the document DOC and generates input image data. The layout analysis unit 20 performs a layout analysis of the image data and extracts layout information. The area dividing unit 30 divides the input image data into small area image data based on the layout information. The area dividing unit 30 extracts small areas corresponding to the attribute information and those corresponding to the dictionary registration target phrase from the small area based on the information in the layout database. The attribute information extraction unit 40 extracts attribute information from the small area specified to correspond to the attribute information. The registration target word / phrase extraction unit 60 extracts a dictionary registration target word / phrase from a small area identified as corresponding to the dictionary registration target word / phrase. The dictionary data registration unit 50 registers the extracted dictionary registration target phrase and attribute information in the dictionary DIC.

  FIG. 2 is a diagram illustrating a configuration of the dictionary update system 1. The dictionary update system 1 includes a multifunction device 100 and a server 200. The MFP 100 and the server 200 are connected via a network 300 such as the Internet, a LAN (Local Area Network), and a WAN (Wide Area Network). In FIG. 2, only one multi-function device 100 and one server 200 are shown in order to avoid complicated drawing. However, the dictionary update system 1 includes a plurality of multi-function devices 100 or a plurality of servers 200. May be included.

  FIG. 3 is a diagram illustrating a hardware configuration of the multifunction peripheral 100. The multifunction peripheral 100 mainly includes a control system including a CPU (Central Processing Unit) 110 and the like, an image reading system 160 that reads an image of a document, and an image forming system 170 that forms an image on a sheet (recording material). The CPU 110 has a function of controlling each component of the multifunction peripheral 100 by reading and executing a control program stored in the storage unit 120. The storage unit 120 includes a ROM (Read Only Memory), a RAM (Random Access Memory), an HDD (Hard Disk Drive), etc., and various programs such as a control program and a translation program, and various data such as image data and text data. Remember. The display unit 130 and the operation unit 140 are user interfaces. The display unit 130 is configured by, for example, a liquid crystal display, and displays a message to the user, an image showing a work status, and the like according to a control signal from the CPU 110. The operation unit 140 includes a numeric keypad, a start button, a stop button, a touch panel installed on a liquid crystal display, and the like, and outputs a signal corresponding to a user operation input and a display screen at that time. The user can input an instruction to the multifunction device 100 by operating the operation unit 140 while viewing an image or message displayed on the display unit 130.

  The I / F 150 is an interface for transmitting and receiving control signals and data to and from other devices. By connecting to a public telephone line, for example, via the I / F 150, the multi-function device 100 can perform FAX transmission / reception. Alternatively, by connecting to a network such as the Internet via the I / F 150, the multi-function device 100 can also send and receive e-mail messages. Alternatively, it functions as a printer by receiving image data from a computer device connected via a network and forming an image on a sheet.

  The image reading system 160 includes a document conveying unit 161 that conveys a document to a reading position, an image reading unit 162 that optically reads a document at a reading position and generates an analog image signal, and converts the analog image signal into digital image data. And an image processing unit 163 that performs necessary image processing. The document feeder 161 is a document feeder such as an ADF (Automatic Document Feeder). The image reading unit 162 includes a platen glass on which an original is placed, an optical device such as a light source and a CCD (Charge Coupled Device) sensor, and an optical system such as a lens and a mirror (all not shown). The image processing unit 163 includes an A / D conversion circuit that performs digital / analog conversion, and an image processing circuit that performs processing such as shading correction and color space conversion (all not shown).

  The image forming system 170 includes a sheet conveying unit 171 that conveys a sheet to an image forming position, and an image forming unit 172 that forms an image on the conveyed sheet. The paper transport unit 171 includes a paper tray for storing paper, a transport roller for transporting the paper from the paper tray to a predetermined position one by one (all not shown). The image forming unit 172 includes, for example, a photosensitive drum on which toner images of YMCK colors are formed, a charger that charges the photosensitive drum, an exposure device that forms an electrostatic image on the charged photosensitive drum, and YMCK on the photosensitive drum. It has a developing device or the like that forms toner images of the respective colors (all are not shown).

  The above components are connected to each other by a bus 190. For example, when image data is generated from a document by the image reading system 160 and an image is formed on a sheet by the image forming system 170 according to the generated image data, the multifunction peripheral 100 functions as a copying machine. When the image reading system 160 generates image data from a document and outputs the generated image data to another apparatus via the I / F 150, the multifunction peripheral 100 functions as a scanner. When the image forming system 170 forms an image on a sheet in accordance with image data received via the I / F 150, the multifunction peripheral 100 functions as a printer. When the image reading system 160 generates FAX data from a document and transmits the generated FAX data to the FAX receiving device via the I / F 150 and the public telephone line, the multifunction peripheral 100 functions as a FAX transmitter. Alternatively, when the image reading system 160 generates image data from a document, and further generates text data from the image data by character recognition processing and generates a translation of the text data by executing a translation program, the multi-function device 100 Functions as a scan translator. Although not shown, a plurality of computer devices are connected to the multifunction device 100 via the I / F 150. Users of these plurality of computer devices can use the multifunction device 100 as a printer, a FAX transceiver, or the like by transmitting and receiving data to and from the multifunction device 100 via their computer devices. Alternatively, by setting a document directly on the multifunction device 100, the multifunction device 100 can be used as a copying machine, a FAX transceiver, or the like.

  FIG. 4 is a diagram illustrating a hardware configuration of the server 200. The CPU 210 executes a program stored in the ROM 220 or the HDD 250 using the RAM 230 as a work area. The HDD 250 is a storage device that stores various programs and data. The user can input data to the server 200 by operating the keyboard 260 and the mouse 270. The server 200 is connected to the multifunction device 100 via the I / F 240 and can transmit and receive data to and from the multifunction device 100. The display 280 displays an image and a message indicating the execution result of the program under the control of the CPU 210. These components are connected to each other by a bus 290. The HDD 250 is a translation server that stores a translation program and a dictionary DIC and has a function of performing translation.

  FIG. 5 is a flowchart showing the operation of the dictionary update system 1. When the power (not shown) is turned on, the CPU 110 of the multifunction peripheral 100 reads the control program from the storage unit 120 and executes it. When the control program is executed, the CPU 110 controls the display unit 130 to display a menu screen. At this time, the multi-function device 100 waits for a user operation input. Similarly, in the server 200, when the power (not shown) is turned on, the CPU 210 reads the control program from the HDD 250 and executes it. When the control program is executed, the CPU 210 enters a data reception waiting state. When the CPU 110 of the multifunction peripheral 100 and the CPU 210 of the server 200 execute the control program, the dictionary update system 1 has each function shown in FIG.

  FIG. 6 is a diagram illustrating a document DOC used in dictionary update processing in the present embodiment. The document DOC is, for example, company information, and various information such as a company name, a business type, a head office location, a URL, and a securities code are arranged according to a predetermined layout. The document DOC is divided into a plurality of small areas by ruled lines, for example. In each chapter area, a character string (hereinafter referred to as “headline character string”) that specifies the type of information described in the small area. For example, [business type], [company name], [head office], [URL] , [Securities code], etc.) and information content (hereinafter referred to as "information character string". For example, manufacturing industry, ABC Industrial Co., Ltd., 100-0000 Tokyo xx ward yy1-1, http://www.xxx .yyy.co.jp /, 0000).

  A description will be given with reference to FIG. 5 again. The user sets the document DOC on the ADF or the platen glass, and operates the keyboard 260 and the mouse 270 of the server 200 to perform an operation input that instructs execution of the dictionary update process. The execution instruction for the dictionary update process includes information for specifying the multifunction peripheral 100. When the execution of the dictionary update process is instructed, the CPU 210 of the server 200 outputs a signal instructing the image reading to the multifunction peripheral 100 specified by the execution instruction of the dictionary update process.

  Upon receiving a signal for instructing image reading, the CPU 110 of the multifunction peripheral 100 controls the image reading system 160 to read the image of the document DOC and generate input image data (step S110). The CPU 110 stores the generated input image data in the storage unit 120. When the document DOC is composed of a plurality of pages, the image data of the plurality of pages is added to the information indicating the page number and stored in the storage unit 120.

  Next, the CPU 110 performs a well-known layout analysis process on the input image data and extracts layout information (step S120). Through the layout analysis process, the input image data is divided into a plurality of small areas. The layout information includes, for example, information such as the coordinates of the vertices of the small areas in the two-dimensional orthogonal coordinate system of each small area and the character size in each small area. CPU 110 stores the extracted layout information in storage unit 120.

  Next, the CPU 110 divides the input image data into a plurality of small area image data based on the layout information (step S130). The small area is given an identifier for distinguishing it from other small areas. The CPU 110 stores the small area image data, layout information, and an identifier in the storage unit 120 in association with each other.

  Next, the CPU 110 searches for a heading character string from each small area (step S140). This process is performed as follows. First, the CPU 110 performs character recognition processing on each of the small area image data to generate text data. CPU 110 stores the generated text data in storage unit 120. CPU 110 retrieves a heading character string from the text data of each small area. The heading character string to be searched may be input by the user when inputting a dictionary update processing instruction, or a database, table, function, or the like that defines the heading character string to be searched in the storage unit 120 or the HDD 250. May be stored in advance. Here, character strings “[company name]” and “[business type]” are defined as headline character strings to be searched.

  Next, the CPU 110 updates the layout database (step S150). That is, the CPU 110 corresponds to a headline character string found by the search, an identifier for specifying a small area in which the character string is found, and the headline character string to a dictionary registration target phrase (for example, proper noun or technical term). Are stored in the storage unit 120 as a layout database in association with each other. The definition of the relationship between the heading character string and whether the heading character string corresponds to the dictionary registration target phrase or attribute information may be input by the user when inputting an instruction for dictionary update processing, A database, table, function, or the like that defines this relationship may be stored in the unit 120 or the HDD 250 in advance. Here, the relationship [company name] = dictionary registration target phrase and [business type] = attribute information is defined.

  Next, the CPU 110 extracts dictionary registration target phrase information and attribute information (step S160). The CPU 110 refers to the layout database and extracts the dictionary registration target phrase from the small area text data corresponding to the dictionary registration target phrase and the attribute information from the small area corresponding to the attribute information. The CPU 110 stores the extracted dictionary registration target phrase and attribute information in the storage unit 120 in association with each other. Next, CPU 110 determines whether extraction of dictionary registration target phrase information and attribute information has been completed for all pages of document DOC (step S170). If the processing has not been completed for all pages (step S170: NO), the CPU 110 repeatedly executes the processing of steps S160 to S170 until the processing is completed for all pages.

  When processing has been completed for all pages (step S170: YES), the CPU 110 updates the dictionary (step S180). That is, the CPU 110 transmits the dictionary registration target phrase and attribute information stored in the storage unit 120 to the server 200 that is the transmission source of the execution instruction for the dictionary update process. CPU 210 of server 200 adds the received information to dictionary DIC stored in HDD 250. When the dictionary DIC is not stored in the HDD 250, the CPU 210 newly generates a dictionary DIC based on the received information.

  FIG. 7 is a diagram illustrating the contents of the dictionary DIC. In this example, “Japanese company name” is recorded as a dictionary registration target phrase, and “English company name” and “business type” are recorded as attribute information thereof. Thus, according to the present embodiment, a dictionary in which dictionary registration target words and their attribute information are associated can be automatically created and updated.

<Modification>
The present invention is not limited to the above-described embodiment, and various modifications can be made.
In the above-described embodiment, the mode of searching for a heading character string and an information character string from a small area has been described. It is also possible to search for graphics created in this manner. That is, a heading graphic and an information graphic may be used instead of the heading character string and the information character string.

  In the above-described embodiment, a mode has been described in which the server 200 outputs a dictionary update instruction, the MFP 100 extracts information necessary for updating the dictionary, and transmits the information to the server 200. However, the division of functions between the multifunction peripheral 100 and the server 200 is not limited to that described in the above embodiment. The server 200 may execute part or all of the functions described as the functions of the multifunction peripheral 100 in the above-described embodiment. Conversely, part or all of what has been described as the function of the server 200 in the above-described embodiment may be executed by the multifunction peripheral 100. For example, the multi-function device 100 may have all the functions of the dictionary update system 1 described in the above embodiment. This is effective when the multifunction peripheral 100 itself stores the translation program and dictionary in the storage unit 120 and has a function as a translator.

  In the above-described embodiments, the aspect in which the multifunction peripheral 100 has the above-described functions has been described. However, instead of the multifunction peripheral, an image reading device such as a scanner having no image forming function may be used.

  In the above-described embodiment, the aspect in which the multifunction device 100 automatically updates the layout database has been described. However, the multifunction device 100 may not automatically update the layout database. In other words, the layout database may be predetermined and unchanged. Alternatively, the layout database may be input when the user instructs execution of the dictionary update process.

It is a block diagram which shows the function structure of the dictionary update system 1 which concerns on one Embodiment. It is a figure which shows the structure of the dictionary update system. 2 is a diagram illustrating a hardware configuration of the multifunction machine 100. FIG. 2 is a diagram illustrating a hardware configuration of a server 200. FIG. 3 is a flowchart showing the operation of the dictionary update system 1. It is a figure which illustrates document DOC. It is a figure which illustrates the contents of dictionary DIC.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 ... Dictionary update system, 10 ... Image reading part, 20 ... Layout analysis part, 30 ... Area division part, 40 ... Attribute information extraction part, 50 ... Dictionary data registration part, 60 ... Registration object phrase extraction part, 100 ... Multifunction machine 110 ... CPU, 120 ... storage unit, 130 ... display unit, 140 ... operation unit, 150 ... I / F, 160 ... image reading system, 161 ... original conveying unit, 162 ... image reading unit, 163 ... image processing unit, 170: Image forming system, 171: Paper conveying unit, 172 ... Image forming unit, 190 ... Bus, 200 ... Server, 210 ... CPU, 220 ... ROM, 230 ... RAM, 240 ... I / F, 250 ... HDD, 260 ... Keyboard, 270 ... Mouse, 280 ... Display, 290 ... Bus, 300 ... Network

Claims (4)

Image reading means for reading an image of a document and generating input image data;
Layout analysis means for performing layout analysis processing on the input image data generated by the image reading means and generating layout information;
Image dividing means for dividing the input image data into a plurality of small regions based on layout information generated by the layout analyzing means;
Among the plurality of small areas, a first identifier for identifying a small area having a heading character string or a heading image and a second area for identifying the small area having the heading character string or heading image and an information character string or information image A layout database that stores the identifiers and their information character strings and information images in association with each other;
The registration target word / phrase from the small area specified by the first identifier stored in the layout database, and the attribute information of the dictionary registration target word / phrase from the small area specified by the second identifier stored in the layout database. Information extracting means for extracting;
An image reading apparatus comprising: output means for outputting a dictionary registration target phrase extracted by the information extraction means. Definition storage means for storing definitions of a heading character string or heading image and an information character string or information image;
A small area specifying means for specifying a small area having a heading character string or a heading image and a small area having an information character string or an information image among the plurality of small areas according to the definition stored in the definition storage means; ,
The image reading apparatus according to claim 1, further comprising database update means for updating contents of the layout database based on information on the small area specified by the small area specifying means. An image reading step for reading an image of a document and generating input image data;
A layout analysis step of performing layout analysis processing on the input image data and generating layout information;
An image dividing step of dividing the input image data into a plurality of small regions based on the layout information;
Among the plurality of small areas, a first identifier for identifying a small area having a heading character string or a heading image and a second area for identifying the small area having the heading character string or heading image and an information character string or information image And the second identifier stored in the layout database from the small area specified by the first identifier stored in the layout database that stores the identifier and its information character string and information image in association with each other. An information extraction step of extracting attribute information of the dictionary registration target phrase from the small area specified by
A dictionary registration target word / phrase extraction method comprising: an output step of outputting the dictionary registration target word / phrase extracted in the information extraction step. Computer equipment,
An image reading step for reading an image of a document and generating input image data;
A layout analysis step of performing layout analysis processing on the input image data and generating layout information;
An image dividing step of dividing the input image data into a plurality of small regions based on the layout information;
Among the plurality of small areas, a first identifier for identifying a small area having a heading character string or a heading image and a second area for identifying the small area having the heading character string or heading image and an information character string or information image And the second identifier stored in the layout database from the small area specified by the first identifier stored in the layout database that stores the identifier and its information character string and information image in association with each other. An information extraction step of extracting attribute information of the dictionary registration target phrase from the small area specified by
An output step of outputting the dictionary registration target phrase extracted in the information extraction step.

Images