JP2008269216A - Document image acquisition device - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a document image acquisition device capable of accurately classifying document images used for a presentation acquired as image data by using the document image acquisition device for every announcement for the documents regardless of their document formats. <P>SOLUTION: The document image acquisition device comprises an image sampling section 403 for acquiring the document image on the basis of a display output signal of the document in which the display contents are changed in response to progress of an explanation of a presenter, a VGA signal ON/OFF detection section 401 for detecting existence of the display output signal, and an announcement alternation determination section 410 for determining an end of the document images per each presenter for the document images acquired by the image sampling section 403 on the basis of detected results of the VGA signal ON/OFF detection section 401. <P>COPYRIGHT: (C)2009,JPO&INPIT

Description

  The present invention generally relates to a document image management system. More specifically, presentation document images acquired as image data using a presentation document image acquisition device or the like are classified for each presentation, and the title of the presentation and the name of the presenter are indexed for the classified document images. The present invention relates to a presentation document image acquisition apparatus capable of identifying and selectively displaying a document image based on the index information by giving it as information.

  In a presentation performed at a meeting or the like, a plurality of presentation documents (slides) are often displayed by a plurality of presenters. Normally, these meeting materials are only displayed on the spot via a projector, etc. If you want to refer to the contents again, you need to ask the presenter to provide the original presentation document. .

  A presentation document image acquisition device (hereinafter simply referred to as a document image acquisition device) captures a video signal sent from a presenter's PC to a projector or the like by using a distributor to obtain a video on the presenter's PC monitor. The same video is formed and acquired as image data. If this apparatus is used, although there is a restriction of image data, it is convenient because presentation materials can be automatically acquired and stored.

  However, when a presentation is made by a plurality of presenters, a large number of document images are accumulated. Therefore, even if those thumbnail images are displayed, who displayed each document image and what kind of image was displayed. There is a problem that it is not easy to grasp whether the title is related to the title. For this reason, conventionally, after manually classifying document images for each presentation unit, the title and the name of the presenter are often given as index information. However, when there are a large number of target document images, there is a problem that it takes a lot of labor for the user.

  Here, automation of two operations is desired. The first is an operation of dividing the accumulated document image for each presentation. The second is an operation of indexing the presenter's name and title to a group of document images classified by presenter or topic. Although not directly applied to the document image acquisition apparatus, the following conventional techniques can be applied to automating these two operations.

The following segmentation technologies are available for presentation units.
(A) A group of presentations consists of continuous document display, and pays attention to the fact that there is a certain time interval until the next presentation starts. Are classified as presentation breaks.
(B) Recording the activation operation of the presentation software and the time when the operation was performed, and associating the operation with the display document (see Patent Document 1).
(C) Focusing on the fact that the first page of the presentation document is a normal cover page, the title extraction process and the layout process are used to identify whether the page is a cover page. In order to obtain the discriminant function, a genetic algorithm, a neural network, and discriminant analysis are used (see Patent Document 2). Then, the document is divided into presentation units based on the identified cover.
(D) The presentation is segmented according to the time schedule described in the schedule management tool or the like.
(E) Focusing on the fact that presentation documents often use different formats (design templates) for each presenter, a group of document images having a common format is determined as a group of presentation documents.

Extraction techniques such as title and presenter name include the following.
(A) For a document in which a region such as a title is fixed, a character string in the fixed region is extracted as a title (see Patent Document 3).
(B) A character string to be extracted is marked by a color marker or a border, and the marked area is extracted by OCR (see Patent Document 4).
(C) A structural model that expresses the correspondence between the layout structure for coping recognition and the logical structure such as “title”, “author”, and “text” is defined, and the region segmentation result of the document image is matched with the structural model. Thus, the “title”, “author”, and “text” areas are identified and extracted (see Patent Document 5).
(D) “Likeness of title” is calculated from various features of a rectangular frame surrounding a character and a character string, and those exceeding a threshold are extracted as titles (see Patent Document 6).
JP 2002-109099 A JP 2001-28041 A JP-A-64-46873 JP-A-1-150964 JP-A-5-342326 JP-A-9-134406

  A first object of the present invention is to classify a document image such as a presentation acquired as image data using a document image acquisition device or the like with high accuracy for each presentation in any document format. The object is to provide a document image acquisition apparatus.

  A second object of the present invention is to provide a document image acquisition apparatus capable of accurately assigning index information for each presenter to a classified document image.

  In order to solve the problem, the document image acquisition apparatus according to the first aspect of the present invention acquires an image unit based on a display output signal of a document whose display contents change according to the progress of the presenter's explanation. And detecting means for detecting the presence / absence of a display output signal, and determining means for determining a delimiter of the document image for each presenter with respect to the document image acquired by the acquiring means based on the detection result of the detecting means.

  The document image acquisition apparatus according to the second aspect of the present invention includes a classification unit that classifies the document image acquired by the acquisition unit based on the determination result of the determination unit according to the first aspect.

  The document image acquisition apparatus according to the third aspect of the present invention is characterized in that the classification means according to the second aspect inserts an image indicating a delimiter for each presenter between document images. is there.

  The document image acquisition apparatus according to a fourth aspect of the present invention is characterized in that the classification means according to the second aspect identifies a break between a document image group of a presenter unit and a document image group of another presenter unit. To do.

  According to a fifth aspect of the present invention, there is provided a document image acquisition apparatus, an extraction unit that extracts attribute information for each presenter from a document image acquired by the acquisition unit of the first aspect, and a classification unit that classifies the attribute information extracted by the extraction unit. Providing means for assigning to the document image group of the classification result as index information.

  According to a sixth aspect of the present invention, there is provided a document image acquisition apparatus, an acquisition unit that acquires a document image based on a display output signal of a document whose display content changes according to the progress of an explanation by a presenter, meeting holding information, and an agenda Input means for inputting at least one of the records, first extraction means for extracting attribute information for each presenter including at least one of the presenter name and title name from the input content of the input means, and first extraction means And determining means for determining the delimiter of the document image for each presenter with respect to the document image acquired by the acquiring means.

  A document image acquisition apparatus according to a seventh aspect of the present invention includes a classification unit that classifies the document image acquired by the acquisition unit based on the determination result of the determination unit according to the sixth aspect.

  A document image acquisition apparatus according to an eighth aspect of the present invention includes: a second extraction unit that extracts attribute information for each presenter including at least one of a presenter name and a title name from a document image acquired by the acquisition unit; Means for assigning the attribute information extracted by the means to the document image group of the classification result of the classification means as index information.

  A document image acquisition apparatus according to a ninth aspect of the present invention is such that the document image in the first to eighth aspects is a slide image.

  According to the configurations of the first and sixth aspects of the present invention, it is possible to divide by presenter regardless of the document format, and thus it is possible to classify the presenter unit with high accuracy. According to the configurations of the fifth and eighth aspects, index information can be given to the classified document images with high accuracy in units of presenters.

(First embodiment)
DESCRIPTION OF EXEMPLARY EMBODIMENTS Hereinafter, exemplary embodiments of the invention will be described with reference to the drawings.
The document image acquisition apparatus according to the first embodiment of the present invention is an apparatus for classifying the document image group acquired by the document image acquisition apparatus in units of presentation (from the start of the presentation to the end of the presentation). First, the document image acquisition apparatus according to the first embodiment of the present invention will be outlined.

  The document image acquisition device (display screen recording device) 4 according to the first embodiment receives a screen image displayed on the display of the presenter's personal computer (PC), that is, a display output signal (video signal, for example, RGB signal). It is a device that captures and records. In FIG. 1, the display screen recording device 4 is described. As shown in FIG. 1, RGB signals sent from the presentation PC 1 to the display device 3 such as a projector are distributed by the signal distributor 2 and taken into the display screen recording device 4. As will be described in detail later, it is possible to form a screen image from RGB signals in a display screen recording apparatus and store it (refer to JP-A-2006-127518 for details).

A method for identifying a change in presentation, that is, a method for identifying a break of a document image in units of presenters will be described.
FIG. 2 is a diagram for explaining identification of a change of presentation (identification of a document image break for each presenter).
In general, when a presentation (explanation) is given at a conference, the presenter displays the created presentation material on the display of his notebook PC and displays it on the notebook PC to display it to those who participated in the conference. And a projector are connected with a VGA cable, a display command to an external display device 3 is executed, and RGB signals are sent to the display device 3 to be displayed on the display device 3.

  In the case of giving a presentation with his / her notebook PC 1, the cable connection or wireless connection between the display device 3 and the PC 1 is involved. In FIG. 2, three presenters (Exhibit A, Otsu, and Rin) are making presentations sequentially using their own notebook PCs. At this time, the state of the VGA signal (RGB signal) is as follows: The period from the time when the presenter's presentation ends (t1) to the time when the presenter's presentation starts (t2) is stopped.

  Similarly, the period from the time point when the presenter's presentation ends (t3) to the time point when the presenter's presentation starts (t4) is stopped. Thus, if the stop state of the VGA signal is detected and recorded, the document image recorded and accumulated by the display screen recording device can be divided in association with the presentation change.

  In other words, if the presentation change is performed during these stopped states, the document image acquired during the period from t0 to t1 is displayed on the presenter and the document image acquired during the period from t2 to t3 is displayed on the presenter. The document image acquired during the period from t4 to t5 can be associated with the presenter.

  As shown in FIG. 2, when an RGB signal stop state (within a certain period of time) is detected, a delimiter image or the like is generated and recorded together with a document image in order to clearly indicate that a presentation change has occurred. It is also possible to accumulate.

A detailed configuration of the document image acquisition apparatus (display screen recording apparatus) according to the present embodiment will be described.
FIG. 3 shows a basic configuration of the display screen recording apparatus 4 according to the embodiment of the present invention. Here, each functional block described in FIG. 3 is realized by a processing device such as a CPU or a program executed therein, a semiconductor storage device such as a DRAM, or a storage device such as a hard disk. The same applies to the following description.

The VGA signal ON / OFF detection unit 401 is an example of a detection unit according to the present invention, and detects an ON / OFF state of a VGA signal distributed and transmitted from a VGA signal distributor (not shown).
The time recording unit 402 records the start time of the ON / OFF state of the VGA signal detected by the VGA signal ON / OFF detection unit 401. Furthermore, the acquisition time of the change image extracted by the change image extraction unit 404 (to be described later), and the announcement change time when the announcement change determination unit 410 determines that the announcement change has occurred are counted and recorded.

The image sampling unit 403 is an example of an acquisition unit of the present invention. When a VGA signal is detected, the image sampling unit 403 acquires the VGA signal at a predetermined time period, digitizes the signal, forms an image, Temporarily save to a storage area that does not.
The change image extraction unit 404 compares the image newly formed by the image sampling unit 403 with the image extracted immediately before it, and if there is a difference between them, the newly formed image is selected as the storage target image. To do.

  The text information extraction unit 405 is an example of the extraction unit of the present invention, and extracts text information by applying a character recognition process (OCR) to a document image selected as a storage target image. This processing is performed to search document images recorded and accumulated based on text information, and to assign the presenter's name and title of presentation described in the document image as an index to the classified document image group. To do.

The thumbnail generation unit 406 generates a thumbnail image of the document image selected as the storage target image.
The index file creation unit 407 creates an index file that associates the storage target image and its acquisition time, the presentation change and the time when it occurred, text information extracted from the storage target image, and a thumbnail image (for the file structure). Will be described later).

  The presentation change determination unit 410 is an example of a determination unit of the present invention, and determines whether or not there is a change of presentation based on the VGA signal ON / OFF state detected by the VGA signal ON / OFF detection unit 401 and their time information. To do. In other words, when the VGA signal was observed to change two times, the change from the ON state to the OFF state (signal stop change) and the subsequent change from the OFF state to the ON state (signal resumption change), an announcement change occurred. Determined. It should be noted that an interval time between the signal stop change and the next signal restart change is counted, and two changes of the signal stop change and the signal restart change are observed within a certain temporal threshold range (for example, 10 minutes). It is determined that a presentation change has occurred, but if two changes of signal stop change and signal restart change are observed within the range exceeding the threshold, it is determined that a new meeting is started and the presentation material is displayed there It is also possible to define When it is determined that an announcement change has occurred, the announcement change determination unit 410 records the announcement change event and its time (signal stop time and signal restart time), and sends the information to the announcement change identification image generation unit 411. .

  The presentation change identification image generation unit 411 is an example of a classification unit according to the present invention, and creates an image that clearly indicates a change of presentation. This image may be, for example, an image including a message such as "Here the presenter has changed", and by displaying these images sandwiched between the thumbnail images of the document images that have been sequentially acquired, Enable users to see at a glance.

  The data storage unit 408 includes a document image extracted by the change image extraction unit 404, a thumbnail image generated by the thumbnail generation unit 406, a presentation replacement identification image generated by the presentation replacement identification image generation unit 411, and an index file generation unit 407. Stores information such as created index files.

  The network IF unit 409 is an interface for connecting the display screen recording device 4 to a network such as the Internet. The user can specify the URL of the display screen recording device 4 described here from his / her computer, access the information stored in the data storage unit 408 in the device, and display them.

An example of an index file in the present embodiment will be described next.
FIG. 4 shows a configuration example of the index file created by the index file creation unit 407. Rstart (event ID 0001) in the table is recorded when the main body switch of the display screen recording apparatus of the present invention is turned on and a VGA signal is first detected. Each time an event occurs, the index file is updated by adding event data to the data area.

A display example of thumbnail images in which the change of presentation (delimiter of document image in units of inventor) is clearly described in the present embodiment will be described.
FIG. 5 is a display example of a document image provided to the user via the Internet. In accordance with the recording order, the thumbnail images are arranged side by side from the upper left to the lower right, and the announcement replacement identification image “presenter has changed” is inserted at the place where the presenter has changed. By performing such display, it is possible to easily identify from which document to which document is included in the same presentation. Of course, if the thumbnail image is clicked, the image can be enlarged and displayed as in the existing technology.

  Note that this display example is merely an example.For example, the edge color of the thumbnail image and the background color of the thumbnail image are changed for each presentation without inserting the presentation change identification image, or the thumbnail images are arranged for each presentation. It may be corrected.

  In the embodiments described so far, the acquired document images are displayed by being classified by presentation unit (presenter unit). If the name of the presenter and the title of the presentation (presentation title) can be assigned as attribute information to each document image group separated into presentation units, search and classification based on these attribute information becomes possible. In addition, it is convenient to display these attribute information by simply hovering the mouse pointer over the thumbnail image. Hereinafter, a display screen recording apparatus improved so as to extract the name of the presenter and the title of the presentation from the document image and assign it to each document image group separated into presentation units will be briefly described.

FIG. 6 is a functional block diagram of the improved document image acquisition device (display screen recording device 4B). The same components as those in the document image acquisition apparatus shown in FIG. 3 are denoted by the same reference numerals, and the description thereof is omitted.
The method of classifying documents displayed at the time of presentation based on the change of the ON / OFF state of the VGA signal for each presentation unit is the same as the method described in the display screen recording apparatus 4 described above.

Next, the flow of title / presenter name extraction processing will be described.
The title / presenter name extraction unit 412 is an example of the first or second extraction unit of the present invention, and the title of the presentation is separated for each group of document images separated in the presentation unit accumulated in the data storage unit 408. And the process of extracting the name of the presenter. The presentation document differs from other documents in the format of the cover and other pages. Typically, the title is written in the largest character string in the middle and upper area of the cover document image. In the document page, the section title is described by the largest character string at the top of the document image.

  Therefore, after detecting the cover using the difference in format, the title can be extracted therefrom. The name of the presenter can be extracted by comparing text information indexed (index information added) to the document image from which the title is extracted with a dictionary of first names (first name and last name). The processing flow will be described below.

Step1.
The CPU selects the document image displayed first from the document group classified as the same presentation.
Step 2.
The CPU extracts all the character string rectangles separated for each line from the document image, measures their height, and selects the character string rectangle having the maximum height.
Step 3.
The CPU obtains position information (the coordinates of the upper left point and the lower right point) of the maximum character string rectangle in the document image.
Step 4.
The CPU determines whether the two coordinates acquired in Step 3 are in the area defined as the title area of the cover or in any other area.
Step 5.
If it is determined that the two coordinates are within the area defined as the title area of the cover, the CPU determines that the document image is the cover and proceeds to Step 6. If it is not determined to be a cover, the next displayed document is selected, and Step 2 to Step 5 are repeated. Even if all the document images are processed, if there is no document image determined to be a cover, identification information indicating that there is no title is generated and recorded in the temporary storage.

Step 6.
The CPU extracts characters within the maximum character string rectangle and temporarily stores them as titles.
Step 7.
The CPU collates all text information (extracted by the text information extraction unit 405) assigned to the document image from which the title has been extracted with the name dictionary, and extracts the collated character string. If there is no text information to collate with the name dictionary, identification information indicating that the presenter's name has not been extracted is generated and recorded in the temporary storage.
Step8.
The CPU checks whether or not characters other than “comma (,), dot (,), and (&)” follow the extracted surname or surname name string, and if so, announces It is determined that the name is not the presenter name, and information indicating that the presenter name has not been extracted is generated and temporarily recorded. On the other hand, if it does not follow, the name of the presenter is determined and the extracted character string is recorded in the temporary storage as the name of the presenter.

Next, the index file update process will be described.
The index file update processing unit 413 is an example of the assigning unit of the present invention, and is information (title information, no title, presenter name, presenter) recorded in the temporary storage in the title / presenter name extraction process described above. (No name) is assigned as attribute information to all documents classified as the same announcement. This can be done by adding a data entry field for attribute information about the title and the name of the presenter to the index file and writing the above information there.

Next, the display message creation unit 414 will be described.
The display message creation unit 414 uses the title or the name of the presenter recorded in the index file so that the thumbnail image group arranged and displayed for each presentation unit can know the presentation material about what subject by which person. Then, a message for clearly showing to the user is created and stored in the data storage unit 408. A display control unit (not shown) displays these messages in association with the thumbnail image group. FIG. 7 shows a display example. For the conventional technology for extracting the title and the name of the presenter, these methods pointed out at the beginning may be used.

(Second Embodiment)
Subsequently, a second embodiment of the present invention will be described.
The second embodiment is a method of detecting a change of announcement that is completely different from the first embodiment, and the point is meeting bibliographic information (conference name, meeting date and time, meeting date described in the agenda and minutes) By extracting the location, presenter, and presentation (presentation title)), the text information in the document image obtained during the presentation is collated to identify the “first slide” immediately after the presentation change, The title is given as attribute information to a group of slides regarded as the same presentation. Whereas conventional methods have attempted to extract presenters and abstracts using only a single information source called document images, here the accuracy of extraction has been improved by referring to the agenda and minutes. There is a feature.

FIG. 8 shows a basic configuration diagram of a display screen recording apparatus in the second embodiment.
1 is a PC. Reference numeral 3 denotes a display device. Reference numeral 5 denotes a display screen recording apparatus according to the second embodiment. The video signal (RGB signal) of the PC 1 for the presenter is sent to the display device 3 and the display screen recording device 5 using the signal distributor 2.

  The display device 3 images the RGB signal sent from the PC 1 and projects the image on a screen or the like. On the other hand, the RGB signal sent from the PC 1 is sent to the image forming unit 501 of the display screen recording device 5 to form an RGB image.

The image forming unit 501 forms images at a constant cycle (for example, every 1 second).
The change screen detection unit 502 detects whether there is a change in the two images by comparing with the image acquired immediately before each new image formation. Whether or not there is a change may be determined based on whether or not the amount of change reaches a preset number of pixels (for example, 5000 pixels). The image whose change is detected by the change screen detection unit 502 is subjected to the following series of processing as being stored in the database 509.

  First, time information (year / month / day / hour / minute / second) is given by the time stamp unit 503. Next, the thumbnail generation unit 504 generates a thumbnail (reduced image) to be displayed by the browser based on the RGB image formed by the image forming unit 501. In order to reduce the memory size, the format conversion unit 505 converts the format of the RGB image into a JPEG image. In the character recognition unit 506, text information is extracted from the JPEG image and segmented into, for example, words or morphemes. The layout information extraction unit 507 acquires the position coordinates (upper left coordinates and lower right coordinates) and character size information of each word and morpheme extracted by the character recognition unit 506 on the slide.

  The index file creation unit 508 creates index information in the form of associating the information acquired by the above-described series of processing with an image (ID), and stores it in the database 509 together with images and thumbnails. The Web server 510 distributes information in the database via a communication network in response to a client request.

  The agenda / minutes registration unit 511 is a user interface for registering the agenda and minutes in the database 509. The minutes are registered after the meeting (announcement), but in the case of an agenda, it may be before the announcement. This registration operation is performed at a different time from the slide image recording. As shown in FIG. 9, in the case of the present embodiment, a document is registered in the data storage unit with an agenda / meeting identifier using a menu screen. Although the display configuration is not described, the registered agenda or minutes are displayed on the same screen as the presentation document image and can be referred to.

  The slide classification information automatic assigning unit 512 uses the agenda and minutes registered by the agenda / minute registering unit 511 to classify information (presentation ID, meeting name, meeting, etc.) on an already captured image (slide). This is a user interface for a user to give an instruction to execute a process for automatically assigning a room, presentation title, conference bibliographic information such as the name of the presenter).

  In the case of the present embodiment, as shown in FIG. 9, an item “giving conference bibliographic information” is prepared in the menu item, and the user uses “manual input” “use agenda” “use minutes” One of three options can be selected. When “Use agenda” or “Use minutes” is selected, the agenda and minutes file list registered in the database 509 is displayed, and which of them is used to specify the classification process Request to do.

  The agenda / minutes selection unit 513 provides a user interface for that purpose, searches the database 509 for the agenda or minutes selected by the user, and sends it to the text information extraction unit 514.

The text information extraction unit 514 extracts text information from the agenda and minutes, and then divides the text information into words or morphemes.
The name / title extraction unit 515 extracts a name / title pair from the text information extracted by the text information extraction unit 514.
The layout information extraction unit 516 extracts the extracted name / title character size and position information. The flow of these processes will be described later.

  The presentation segmentation processing unit 517 detects the title slide (cover slide) from the captured slide, and executes a process for separating the slide in association with the change of the presenter. The slide including the name and title It consists of three parts: an extraction unit 5171, a head slide identification unit 5172, and a last slide identification unit 5173.

A slide extraction unit 5171 including a name / title extracts a slide including a name / title pair described in the agenda or minutes from the name / title extraction unit 515 from the database 509.
The first slide identification unit 5172 determines whether or not the slide detected by the slide detection unit 5171 including the name and title is the first slide displayed first by the presenter, and identifies the first slide.
The last slide identification unit 5173 identifies the last slide based on the time information on which the first slide identified by the first slide identification unit 5172 is displayed. The flow of these processes will be described in detail later.

  The index file update unit 518 updates the index information of each slide based on the information obtained by the above processing. More specifically, the index file update processing unit 518 performs the following index file addition processing. (1) Information such as a meeting name and a meeting place is added to all slides associated with the agenda. (2) Based on the time information (presentation start time) of each “first slide” extracted by the first slide extraction process, the recorded slide images are separated for each presentation (the last slide of each presentation is A specific presentation identifier is given to a slide image belonging to each presentation. (3) The name of the presenter and the title of the presentation (agenda) are added to the index file on the slide given the same presentation identifier. (4) A head slide identifier is given to each “head slide”.

Here, the flow of the extraction processing of the slide classification information (conference bibliographic information) including the name / title pair described above will be described with reference to FIG.
FIG. 10 schematically shows the flow of processing for extracting conference bibliographic information using an agenda.
Briefly, according to the layout analysis result (relative position of the character rectangle frame) made for the agenda, select a bibliographic information extraction template that approximates it and extract character information, and for the extracted characters Using the identification / judgment rule, the type of bibliographic information is determined whether or not it is bibliographic information to be extracted.

  The presentation segmentation processing unit 517 detects the slide (document image) including the title and presenter name pair extracted by the conference bibliographic information extraction process for the above-described agenda, and thereby the first slide (cover slide) of each presentation. Is identified. That is, as shown in FIG. 11, since the text information of each slide has already been extracted by OCR and stored as an index file, the information of these index files and the title extracted by the conference bibliographic information extraction process for the agenda And the presenter name pair can be matched to extract the first slide.

The flow of processing for identifying the first slide will be described as follows.
Step1.
The CPU detects all pairs of the agenda (Tn) and name (Nn) from the conference bibliographic information extracted from the agenda (in the example of FIG. 11, “{electronic paper, Takashi Sasamura}, {Hikari Interconnection, Harada Yoshio} ... {Next-generation environmental technology, Akira Oyama} "counts the total number of pairs.
Step2.
The CPU sequentially checks each pair of agenda (Tn) and name (Nn) and index information of each slide image, and counts the number of matched agendas and names (slide 1 = agenda 0, Name 0, Slide 2 = Agenda 1, Name 1 ...)

Step3.
(A) In Step 2, if there is a slide image in which the number of agenda / name pairs matches the total number of pairs counted in Step 1 (or if a predetermined threshold is exceeded even if they do not completely match) ), Consider it as an “Agenda” slide, give it an “Agenda Slide” identifier, and remove it from later analysis.
(B) At Step 2, the CPU includes either a slide determined to contain only one agenda / name pair (slide of agenda = 1, name = 1) or an agenda / name pair. The slide determined to have been selected (slide with agenda = 0, name = 1, or slide with agenda = 1, name = 0) is extracted as a “first slide” candidate, and the process proceeds to step 4.

Step4.
If there is only one slide image extracted in the process (B), the CPU determines that it is a “first slide”. When two or more slides are extracted in the process of STEP 3 (B), the time information is compared, and the slide with the earliest presentation time is determined as the “first slide”.

  If the above processing is used, as shown in FIG. 12, the description content of the agenda and minutes can be displayed in association with the description content of the slide group via the same text information. This is convenient because you can quickly display slides that correspond to the description contents of the agenda and minutes, and you can clearly indicate the correspondence with the highlight function. The above description does not describe the acquisition and playback of the presenter's explanation audio performed during slide display, but the method described here also applies to a display screen recording / playback device having a voice recording / playback function. Naturally applicable.

  The preferred embodiments of the present invention have been described in detail above. However, the present invention is not limited to the specific embodiments according to the present invention, and various modifications can be made within the scope of the gist of the present invention described in the claims. Deformation / change is possible.

  As described above, according to the present invention, it is possible to classify a document into units of presenters with high accuracy regardless of the document format, and the industrial applicability is high.

1 is a system configuration diagram of a document image acquisition apparatus according to a first embodiment of the present invention. It is a figure explaining the division | segmentation of the document image in a presenter unit. It is an internal block diagram of the document image acquisition apparatus which concerns on 1st Embodiment of this invention. It is a figure which shows the structural example of an index file. It is a display example of a document image. It is an internal block diagram of the modification of 1st Embodiment. It is a figure which shows the thumbnail group by which arrangement | sequence display identified for every presentation unit was carried out. It is an internal block diagram of the document image acquisition apparatus which concerns on 2nd Embodiment of this invention. It is a figure which shows registration menus, such as an agenda. It is a figure which shows roughly the flow of the process for extracting conference bibliographic information using an agenda. It is a figure explaining a keyword matching process. It is a figure which shows the example which matches and displays the description content of an agenda or the minutes via the text information same as the description content of a slide group.

Explanation of symbols

1: PC
2: Signal distributor 3: Display device 4, 5: Display screen recording device 401: VGA signal ON / OFF detection unit 402: Time recording unit 403: Image sampling unit 404: Changed image extraction unit 405: Text information extraction unit 406: Thumbnail generation unit 407: Index file creation unit 408: Data storage unit 409: Network I / F unit 410: Presentation change determination unit 411: Presentation change identification image generation unit 412: Title / presenter name extraction unit 413: Index file update processing Unit 414: Display message creation unit 501: Image formation unit 502: Changed image detection unit 503: Time stamp unit 504: Thumbnail generation unit 505: Format conversion unit 506: Character recognition unit 507: Layout information extraction unit 508: Index file creation unit 509: Database 510: Web server 5 1: Agenda / minutes registration unit 512: Slide classification information automatic assignment instruction unit 513: Agenda / minutes selection unit 514: Text information extraction unit 515: Name / title pair extraction unit 516: Layout information extraction unit 517: Presentation segmentation processing Unit 5171: Slide extraction unit including name / title 5172: First slide identification unit 5173: Last slide identification unit 518: Index file update unit

Claims (9)

Acquisition means for acquiring a document image based on a display output signal of a document whose display contents change according to the progress of the presenter's explanation;
Detecting means for detecting the presence or absence of the display output signal;
A determination unit that determines a delimiter of a document image for each presenter with respect to the document image acquired by the acquisition unit based on a detection result of the detection unit;
A document image acquisition apparatus comprising:   The document image acquisition apparatus according to claim 1, further comprising a classification unit that classifies the document image acquired by the acquisition unit based on a determination result of the determination unit.   3. The document image acquisition apparatus according to claim 2, wherein the classification unit inserts an image indicating a delimiter for each presenter between the document images between the document images.   3. The document image acquisition apparatus according to claim 2, wherein the classifying unit identifies a partition between a document image group in a presenter unit and a document image group in another presenter unit. Extracting means for extracting attribute information in units of presenters from the document image acquired by the acquiring means;
An assigning means for assigning the attribute information extracted by the extracting means as index information to the document image group of the classification result of the classifying means;
The document image acquisition apparatus according to claim 2, further comprising: Acquisition means for acquiring a document image based on a display output signal of a document whose display contents change according to the progress of the presenter's explanation;
An input means for inputting at least one of meeting holding information and minutes;
First extraction means for extracting attribute information for each presenter including at least one of the presenter name and the title name from the input content of the input means;
A determination unit that determines a delimiter of a document image for each presenter with respect to the document image acquired by the acquisition unit based on the extraction result of the first extraction unit;
A document image acquisition apparatus comprising:   The document image acquisition apparatus according to claim 6, further comprising a classification unit that classifies the document image acquired by the acquisition unit based on a determination result of the determination unit. Second extracting means for extracting attribute information for each presenter including at least one of the presenter name and the title name from the document image obtained by the obtaining means;
An assigning means for assigning the attribute information extracted by the extracting means as index information to the document image group of the classification result of the classifying means;
The document image acquisition apparatus according to claim 7, further comprising: The document image acquisition apparatus according to claim 1, wherein the document image is a slide image.

Images