JP2001318941A - Information processor and its method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To automatically execute document processing suited to an inputted document by identifying the sort of the document without specification of a document processing method of a user. SOLUTION: A document feature extraction part 22 extracts the features of a document from a document image and a document sort identification part 23 identifies the sort of the document by comparing the extracted document features with document sort feature information read out from a document feature database 24. On receiving the identification result of the document sort, a processing method suited to the document is read out from a processing selection database 26 and transmitted to a processing selection part 25 to execute prescribed document processing.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an information processing apparatus and an information processing method, and more particularly to a "memo", a "postcard",
The present invention relates to a system in which a document such as a "business card", a "receipt" or various other documents and materials is read by a scanner or the like, and the type of the document is automatically identified.

[0002]

2. Description of the Related Art For processing image information read by a scanner or a camera, see, for example, Japanese Patent Application Laid-Open No. 9-305.
No. 685 (hereinafter referred to as prior art 1),
-240901 (hereinafter referred to as prior art 2).

FIG. 11 shows an example shown in prior art 1. In FIG. 11, an information processing apparatus 1 includes an image display unit 2 that displays a document and image information, and a “receipt”, “receipt”, “postcard”, “business card”, and “photograph” presented by a user.
, A storage unit 4 for storing information from the scanner unit 3, a shape recognition unit 5 for recognizing the size and shape of the input paper or photograph, and a shape recognition unit 5 And an image classifying unit 6 that classifies the image information into a plurality of types according to size and shape based on the information from the image display unit and causes the image display unit 2 to classify and display the image information.

The image classifying unit 6 includes a pattern recognition unit 12
The image information from the scanner unit 3 or the storage unit 4 is collated with the image information from the image type storage unit 13 based on the information on the shape and image pattern of the image from. , And are classified and displayed on the image display unit 2.

Further, the image classification section 6 automatically sorts and transmits data as image information to the application program 14. The application selected by the application program 14 is, for example, "address book" 14a for "postcard" or "business card", "household book" 14b for "receipt" or "receipt", and "album" 14c for "photo". In this way, it specifies a compatible application that suits real life. In FIG. 11, reference numeral 7 denotes a clock unit, and reference numeral 15 denotes an application starting unit.

The prior art 2 discloses a document filing apparatus and a method for automatically classifying items of a document correctly even when layout changes are large and items are omitted.

Although the above-mentioned prior art 1 shows a technical idea similar to that of the present invention, it shows that data stored in a document feature database serving as a reference for identifying a document type is updated. Absent.

[0008] In the prior art 2, "postcard" or "business card" is [address book], "receipt", "receipt" is related to [household book], and "photo" is album. There is no technical idea to execute the specification of the corresponding application suitable for life and to automatically classify documents.

[0009]

SUMMARY OF THE INVENTION The present invention solves such a conventional problem. By automatically identifying the type of an input document, a user can specify a processing method without specifying a processing method. It is an object of the present invention to provide an information processing apparatus and an information processing method capable of executing information processing suitable for the document.

[0010]

According to a first aspect of the present invention, there is provided a document image input / output unit for recognizing a document and outputting image information of the document, and extracting a feature of the document from the image information of the document. A document feature extraction unit, a document feature database storing document feature information of registered document types, document feature information stored in the document feature database, and document feature information extracted by the document feature extraction unit. A document feature registration unit in which features of the document are registered based on the document type; a unit for newly registering or updating document feature information in the document feature database by the document feature registration unit; and a document feature extraction unit. A document type identification unit for comparing the document characteristics extracted in the above with document characteristic information stored in the document characteristic database to identify the document type; An information processing apparatus comprising: a processing selection database that stores a corresponding processing method; and a processing selection unit that responds to information on the processing method stored in the processing selection database and a document type output from the document type identification unit. is there.

The invention described in claim 1 will be explained mainly with reference to FIGS. That is, a document image input / output unit 21 for recognizing the document 20 and outputting information of the document image, a document feature extraction unit 22 for extracting document characteristics from the output document image, and a document of the registered document type A document feature database 24 for storing feature information; a document feature information stored in the document feature database; a document feature extracted by the document feature extraction unit 22; and a document feature information stored in the document feature database 24. A document type identification unit 23 for collating and identifying the document type; and a process selection database 26 for storing a processing method corresponding to the document type
And a processing selection unit 25 for extracting the document processing information stored in the processing selection database 26.

In response to the instruction, for example, when the document is recognized as "postcard", the processing unit 2 for executing "add address book" as document processing is provided at the subsequent stage of the processing selection unit 25.
7 to 29 are provided. With this configuration, even if the user does not know in advance the contents of the document processing and the operation method, just input the document. For example, if the document type is “memo”, “postcard”, or “receipt”, "Paste notepad", "Add address book", "Register household account"
The document processing is automatically executed.

According to a second aspect of the present invention, there is provided a document type recognizing section for collating a document characteristic with document characteristic information stored in a document characteristic database to identify a document type. 2. The information processing apparatus according to claim 1, further comprising a user document type selection unit for inquiring the user about the type of the document.

The invention according to claim 2 will be described with reference to FIGS. In particular, the document type identification unit 23 shown in FIG.
Includes a document type recognition unit 31 and a user document type selection unit 32.

Thus, even if the identification accuracy of the document type identification unit 23 is initially low, the user is inquired of the document type candidate through the user document type selection unit 32, so that the document type selected by the user is selected. At the same time, the accuracy of subsequent document type identification is increased.

According to a third aspect of the present invention, there is provided a character recognition unit for recognizing characters in a document image output from a document image input / output unit, and a document of a registered document type. A keyword keyword database that stores the keyword information included in the keyword information; a character recognition result output from the character recognition unit; and a keyword document feature extraction unit that extracts keyword document features from the keyword information stored in the document keyword database. A document type recognizing unit for comparing the document features extracted by the document feature extracting unit and the keyword document features extracted by the keyword document feature extracting unit with document feature information stored in a document feature database to identify a document type. An information processing apparatus according to claim 1.

The invention according to claim 3 will be explained with reference to FIGS. 1, 2 and 5. In particular, in FIG. 5, a document type identification unit 23 includes a character recognition unit 41 that recognizes characters of a document image output from the document image input / output unit 21 and a document that stores keyword information included in a document of the registered document type. A keyword database 43; a character recognition result output from the character recognition unit;
A keyword document feature extraction unit 42 for extracting keyword document features from the keyword information stored in the document information storage unit 24. The document features extracted by the document feature extraction unit 42 are compared with the document feature information stored in the document feature database 24, and the document is compared. A document type recognition unit 44 for identifying the type.

With the above configuration, if the identification result of the document selected by the user and the document recognized by the document type recognition unit 44 are incompatible, the document feature database 24 is updated. The accuracy of the document type recognition is further improved.

According to a fourth aspect of the present invention, in the image input / output unit, a document image reading unit that reads a document as an image, and a document detection unit that detects that a document is input to the document image reading unit. An information processing apparatus according to any one of claims 1 to 3, including a unit.

The invention according to claim 4 will be described with reference to FIGS. 1, 2 and 6. Particularly, as shown in FIG. 6, a document image reading unit 51 and a document detecting unit 52 are provided in the document image input / output unit 21 or in addition thereto.

According to the above, when a user takes a document into the document image reading section 51, the document detecting section 52 detects the state and immediately executes information processing automatically.

According to a fifth aspect of the present invention, a step of outputting a document image with a document as an input, a step of extracting document characteristics from the document image, and storing document characteristic information of a registered document type. Performing new registration and update registration of the document type and document characteristic information by a user's operation input, and identifying the document type by comparing the document characteristic information with the extracted document characteristics, Storing information on the processing method corresponding to the document type; selecting document processing based on the stored information on the processing method and the registered document type information; and selecting the document processing This is an information processing method including a step of executing document processing based on the information.

The invention according to claim 5 will be described with reference to FIGS. According to this information processing method, even if the user does not know in advance the contents and operation method of the document processing, a process of automatically organizing and classifying some document types only by inputting a document is performed. .

According to a sixth aspect of the present invention, the step of outputting a document image with a document as an input further includes the step of collating document characteristic information with the extracted document characteristic to identify a document type and a document process. The information processing method according to claim 5, wherein image information is provided to at least one of the selecting steps.

The invention described in claim 6 will be explained with reference to FIG. In particular, from the document image input / output unit 21 where the step of outputting the document image is executed, the document type identification unit 23 where the step of identifying the document type is executed and the processing selection unit 25 where the step of selecting the document processing are executed Is provided with a document image.

According to this information processing method, the identification of the document type and the accuracy of selecting the document processing can be enhanced.

According to a seventh aspect of the present invention, in the step of collating the document characteristic information with the extracted document characteristic and identifying the document type, the user determines the type of the registered document and the evaluation value of the document. 6. The information processing method according to claim 5, wherein the communication is performed with a unit that refers to the certainty factor of the document and a keyword appearing in the document.

The invention according to claim 7 will be described with reference to FIGS. In particular, in FIG. 3, the user (not shown) inputs the document type recognition unit 3 through the user document type selection unit 32.
1 and confirms the result of identifying the document type, and selects a predetermined document type based on the result of the identification.

According to this information processing method, in the initial state, even if the accuracy of identifying the document type is low, the user is inquired of a candidate (selection) of the document type. Can be set to the document type. The document feature database 24 and the document keyword database 43 are updated by the user's selection and instruction of the document type, and the accuracy of the evaluation value,
Increases confidence.

According to an eighth aspect of the present invention, in the step of comparing the document feature information with the extracted document feature and identifying the document type, the character in the document image output from the document image input / output unit is replaced with a character. A character recognition step for recognizing, a step of extracting a keyword suitable for the document type from a document keyword database registered in advance based on the character recognition step, and a step of extracting information and a document feature database from the step of extracting the keyword. 6. The information processing method according to claim 5, further comprising: a document type recognition step of comparing the stored document characteristic information with the stored document characteristic information to identify a document type.

This information processing method is shown in FIGS.
Explained by In particular, in FIG. 5, the step of extracting a keyword suitable for the document type from the document keyword database 43 and the information from the step of extracting the keyword and the document feature information stored in the document feature database 24 are used for document type recognition. A document type recognizing step of collating by the unit 44 to identify the document type.

According to the above information processing method, when the identification result of the document selected by the user and the identification of the document recognized by the document type recognizing unit 44 do not match, the document characteristic database 2 is used.
4 is updated, so that the accuracy of document type recognition can be further improved.

[0033]

Embodiments of the present invention will be described below in detail with reference to the accompanying drawings.

(Embodiment 1) Hereinafter, Embodiment 1 will be described with reference to FIGS. In the present invention,
Documents 20 such as “memo”, “postcard”, “business card”, and “receipt” are to be processed.

In the document 20, in addition to these, "various minutes", "photographs", "books", etc. which are generally used in daily life and are well-known or between specific organizations or individuals Includes documents and documents exchanged at

The above “memo” may deviate from the concept of a document, but in the present invention, for convenience,
The above items are collectively referred to as "documents".

The document 20 is read by a document image input / output unit 21 using a scanner or a camera, and a document image is output. That is, the document image input / output unit 21 receives the document 20 and performs a step of outputting a document image of the document.

In FIG. 1, the document image generated by the document image input / output unit 21 is provided not only to the document feature extraction unit 22 but also to both a document type identification unit 23 and a processing selection unit 25 described later. However, these are not indispensable components, and the information processing effect of the present invention can be achieved only by providing them to the document feature extraction unit 22.

The document feature extraction unit 22 extracts the features of the document image of the document 20 captured by the document image input / output unit 21. Here, the characteristics of the document image include the size of the document paper, the density of black pixels, the area of a circumscribed rectangle of connected black pixels, the distribution and position of black pixels, and the number and position of lines. Thus, the document feature extracting unit 22 performs a step of extracting features of the document image.

The "circumscribed rectangular area of connected black pixels" refers to an area surrounded by a circumscribed rectangle so that the black pixels connected in the image are included in one set and are included.

The document type identification unit 23 compares the document feature information stored in the document feature database 24 with the document features extracted by the document feature extraction unit 22, and determines whether the document input to the document image input / output unit 21 is valid. The type of the document is identified as "memo", "postcard", "business card", "receipt" or other document.

The document type identification section 23 can fetch a document image input or output to the document image input / output section 21 and perform document identification based on the document image. Thereby, the identification accuracy of the document type is further enhanced.

That is, in the document type identifying section 23, the document information from the document feature database 24 is
(Or second) reference information, and the document image input / output unit 21
The step of identifying the document type may be performed using the document image information from the document as the second (or first) reference information. However, as described above, it is not an essential component that the document information is directly fetched from the document image input / output unit 21 into the document type identification unit 23.

In the document characteristic database 24, information on the type of document and the document characteristics of the document of the type is registered in advance, and such information is provided by the document characteristic database 24. For example, when the document is a business card or a receipt, as shown in Table 1, the document characteristic database includes (a) the length of the document, (b) the width of the document, (c) the number of lines, (c) the density of black pixels, Are registered, and their standard deviations are also registered and stored.

[0045]

[Table 1]

The document feature database 24 can register a new document and update existing information. Referring to FIG. 2, a document feature registration unit 30 which is a feature of the present embodiment is shown. The document type registered by the user and the document features extracted by the document feature extraction unit 22 are stored in a document feature database 24. , Or the document characteristic information for the document type is updated and registered.

As described above, by additionally registering new information and data in the document feature database 24 and registering the updated information, an appropriate document type can be identified in subsequent information processing.

The processing selection section 25 is a processing selection database 2
6 and the document type identification unit 2
In response to the document type output from the document processing unit 3, the document image is sent to the document processing units 27 to 29.

In the processing selection section 25, the document information from the processing selection database 26 is stored in the first (or second)
The document type identification step can be executed using the document image information from the document image input / output unit 21 as the second (or first) reference information. However, capturing the document image from the document image input / output unit 21 into the processing selection unit 25 is not an essential component.

That is, the two providing routes of providing the document image from the document image input / output unit 21 to the document type identifying unit 23 and the processing selecting unit 25 are not essential, but may depend on the function and performance of the document feature extracting unit 22. Does not require two provision routes, or one of the provision routes is sufficient.

The document processing units 27 to 29 include a plurality of processing units as shown. As mentioned above,
Classify documents according to multiple documents, such as [address book] for "postcard" or "business card", [household account] for "receipt" or "receipt", and album for "photo".
This is for organizing. Therefore, at least the number of document processes is prepared.

FIG. 7 shows "document type" and "processing method" applied to each document. As shown in FIG. 7, for example, when the document 20 is recognized as a "memo", the image is sent to the document processing unit 27, and the document processing unit performs a document process of "pasting a memo pad".

When the type of the document is “postcard” or “business card”, for example, a document process of “add address book” is selected and instructed, and the document image is sent to the document processing unit 28.

If the recognized document is “receipt” or “receipt”, for example, a processing method of “register household account book” is selected and instructed, and the document image is sent to the document processing unit 29. It is.

As described above, by performing document processing by identifying the document type, even if the user does not know in advance the content of the document processing and the operation method, the document processing is automatically performed only by inputting the document. Is performed.

(Embodiment 2) Next, Embodiment 2 will be described. The main components are the same as those in the first embodiment shown in FIGS.
This is illustrated by FIG.

Documents 20 such as "memo", "postcard", "business card", and "receipt" are read by a scanner or a camera (not shown), processed by a document image input / output unit 21, and output a document image. The document feature extraction unit 22 is a document image input / output unit 21
Extract the features of the document image extracted from. The document type identification unit 23 compares the document characteristic information stored in the document characteristic database 24 with the document characteristics extracted by the document characteristic extraction unit 22, and identifies the document type. The process selection database 26 stores information on a document type and a document processing method. The processing selection unit 25 selects a processing method from the information on the processing method stored in the processing selection database 26 and the document type output from the document type identification unit 23, and converts the document image into one of the document processing units 27 to 29. To send to.

FIG. 3 shows a document type identification unit 2 according to this embodiment.
3 is a specific configuration diagram of FIG. The document type recognizing unit 31 collates the document feature information stored in the document feature database 24 with the document features extracted by the document feature extracting unit 22 to recognize the document type. It sends to the user document type selection section 32 whether the recognition result is appropriate or not, inquires the user, and waits for the user's selection and instruction. That is, in the present embodiment, the document type identification unit 23 includes a means for communicating with the user.

The document processing is performed in the information processing apparatus having the above configuration as follows. Various documents 20 such as “memo”, “postcard”, “business card”, and “receipt” are processed by a document image input / output unit 21 using a scanner or a camera to obtain a document image.

The document feature extraction unit 22 extracts document feature information, which is a feature of the document, from the document image output from the document image input / output unit 21. The document characteristic information includes, for example, the size of a document sheet, the density of black pixels, the area of a rectangle circumscribing connected black pixels, the distribution and position of black pixels, and the number and position of lines.

In the document characteristic database 24, document types and document characteristic information of the documents of the types are registered in advance. The document type recognizing unit 31 collates the document features extracted by the document feature extracting unit 22 with information on the document features registered in the document feature database 24, recognizes the document type, and determines a document type candidate and a document type. An evaluation value representing the accuracy of type identification is obtained. FIG. 8 shows an example of a document type candidate and an evaluation value. The evaluation value is not set in advance, but is calculated by the document type recognition unit 31. The evaluation value is a so-called relevance ratio that indicates which document the input and identified document has been recognized, and indicates that the larger (higher) the numerical value is, the higher the probability of being the document is.

The evaluation value (a) shown in FIG. 8 is an evaluation result (evaluation value) in which when a document is recognized by the document type recognition unit 31, the document is recognized as a "business card".
, 100, and the evaluation values recognized as “postcard”, “memo”, and “receipt” are 50, 30, and 20, respectively. In this case, the evaluation value of the “business card” with the highest evaluation value is 100, and the evaluation value of the “postcard” with the second highest evaluation is 5
Since the evaluation value is 0 and there is a large difference between these evaluation values, it can be convinced that the document is a “business card”. Therefore, in this case, it is not necessary to entrust the user with the selection and instruction of the document type.

On the other hand, the results shown in the evaluation value (b) of FIG. 8 indicate that the evaluation values of “business card”, “postcard”, “memo”, and “receipt” are 40, 20, 5, and 10, respectively. ing. In this case, the highest evaluation value of "business card" is 40, and the second highest evaluation value is "postcard" of 20,
Although there is a difference of 20 between the two evaluation values, the evaluation value of the highest “business card” is as low as 40. Therefore,
In this case, the probability that the document is a “business card” is low,
The selection and instruction of the document type are entrusted to the user.

The results shown in the evaluation value (c) of FIG. 8 indicate that the evaluation values of “business card”, “postcard”, “memo”, and “receipt” are 75, 80, 10, and 30, respectively. ing. In this case, the highest evaluation value of “postcard” is 80, and the second highest evaluation value is “business card” of 75.
And the evaluation value of both is high, but the difference between them is only 5, so it is difficult to identify whether the document is a “postcard” or a “business card”. Therefore, in such a case, the selection and instruction of the document type are entrusted to the user.

FIG. 9 shows an example of a menu that allows the user to select a document type. The user is asked, for example, "what kind of document", and at that time, for example, "memo", "business card", "postcard", "receipt"
Are displayed, and the user is asked to select a menu. The document type selected by the user is returned to the document type recognition unit 31, and the document type recognition unit 31
Outputs its document type.

In the process selecting section 25, the document type identifying section 23
, And selects a processing method from the process selection database 26 with reference to the process selection database 26, and transfers the document image input from the document image input / output unit 21 to the selected document processing units 27 to 29. Send out.

As described above, when the accuracy of the document type identification is low, the user is inquired of the selection of the document type candidate, so that even if the accuracy of the document type identification is low, correct information processing suitable for the document type can be performed. It is possible to do.

(Embodiment 3) Next, Embodiment 3 will be described. FIG. 4 shows the configuration of the entire information processing apparatus according to this embodiment. The configuration shown in FIG. 4 is substantially the same as the configuration obtained by combining FIG. 1 with FIG. 1 and FIG. 2 and FIG.

As described in the first embodiment, the document image input / output unit 21 inputs the document 20 such as “memo”, “postcard”, “business card”, “receipt” with a scanner or a camera, and outputs the document image. Is output. The document feature extraction unit 22 extracts a document feature from a document image. Document feature database 2
Reference numeral 4 stores document characteristic information of the registered document type. The document type recognizing unit 31 forms a part of the document type recognizing unit 23, and recognizes the document type by comparing the document characteristic information stored in the document characteristic database 24 with the document characteristics extracted by the document characteristic extracting unit 22. I do. Whether the recognition of the document type is appropriate or inappropriate is sent to another user document type selection unit 32 constituting the document type identification unit 23, an inquiry is made to the user, and the judgment of the user is made. , Choices, and directions.

The document feature registration unit 30 includes the document feature extraction unit 2
The document feature extracted in step 2 and the document type output from the document type recognition unit 31 are input, and document feature information of the document type is added to the document feature database 24. The process selection database 26 stores information on document types and document processing methods. The processing selection unit 25 selects a processing method from the information on the processing method stored in the processing selection database 26 and the document type output from the document type identification unit 23, and sends a document image to the document processing units 27 to 29. .

The document processing is performed in the information processing apparatus having the above configuration as follows. Using a scanner or a camera, various types of documents 20, such as "memos", "postcards", "business cards", and "receipts", which differ in medium size, medium shape, character size, and character shape, are input using a scanner or camera. Processing is performed by the output unit 21 to obtain a document image.

The document feature extracting section 22 extracts a document feature which is a feature of the document from the document image. The characteristics of the document include, for example, the size of the document paper, the density of black pixels, the area of a rectangle circumscribing connected black pixels, the distribution and position of black pixels, and the number and position of lines.

In the document characteristic database 24, information on the type of document and the document characteristics of the document of that type is registered in advance. The document type recognizing unit 31 collates the document features extracted by the document feature extracting unit 22 with information on the document features registered in the document feature database 24, recognizes the document type, and determines a document type candidate and a document type. An evaluation value representing the accuracy of type identification is obtained.

FIG. 8 shows an example of a document type candidate and an evaluation value.
If the accuracy of the document type identification is low from the evaluation value, the document type candidate is sent to the user document type selection unit 32. The user document type selection unit 32 displays the document type candidates and allows the user to select a document type. The selected document type is returned to the document type recognition unit 31. The document type recognition unit 31 outputs the document type. When the document type selected by the user is different from the first candidate of the document type candidate, the document type is sent to the document feature registration unit 30.

The document feature registration unit 30 additionally registers the document type input by the user and the document features extracted by the document feature registration unit 30 in the document feature database 24, and updates the document feature information for the document type.

In the process selecting section 25, the document type identifying section 23
, And selects a processing method from the process selection database 26 with reference to the process selection database 26, and transfers the document image input from the document image input / output unit 21 to the selected document processing units 27 to 29. Send out.

As described above, when the document type selected by the user is different from the first candidate of the document type candidate recognized by the document type recognizing unit 31, the document feature database 24 is updated, so that the next and subsequent document types are updated. The accuracy of document type recognition is increased.

(Fourth Embodiment) Next, a fourth embodiment will be described. The overall configuration will be described with reference to FIG.

The document 20 is input by a document image input / output unit 21 via a scanner or a camera, and the document image input / output unit 21 outputs a document image. The document feature extraction unit 22 extracts features of the document image input from the document image input / output unit 21. The document type identification unit 23 includes a document feature database 24
Of the document image extracted by the document feature extraction unit 22 and identifies the document type. The processing selection unit 25 selects a processing method from the information on the processing method stored in the processing selection database 26 and the document type output from the document type identification unit 23, and converts the document image into one of the document processing units 27 to 29. To execute document processing.

FIG. 5 is a diagram showing a document type identification unit 2 according to the present embodiment.
3 is a specific configuration diagram of FIG. The character recognizing unit 41 recognizes characters written in the document image fetched from the document image input / output unit 21 and outputs text information. The document keyword database 43 stores keyword information included in the registered document types. Keyword document feature extraction unit 4
Reference numeral 2 compares keyword information stored in the document keyword database 43 with text information recognized by the character recognition unit 41, and outputs keyword document characteristics. The document type recognition unit 44 collates and identifies the document feature information stored in the document feature database 24, the document features extracted by the document feature extraction unit 22, and the document keyword features extracted by the keyword document feature extraction unit 42. Then, the identified document type is sent to the document type identification unit 23.

The document processing is performed in the information processing apparatus having the above configuration as follows. First, as shown in FIG. 1, various documents 20 such as “memo”, “postcard”, “business card”, and “receipt” are processed by a document image input / output unit 21 using a scanner or a camera, and a document image is processed. obtain.

The document feature extraction unit 22 extracts a document feature which is a feature of the document from the document image. As the document features, for example, the size of the document paper, the density of black pixels, the area of a rectangle circumscribing connected black pixels, the distribution and position of black pixels, the number and position of lines, and the like are extracted.

Next, as shown in FIG.
Recognizes the characters in the document from the document image,
Output text information. In the document keyword database 43, document keyword information such as a keyword and a type of a document presumed to include the keyword is registered in advance. Note that Embodiment 4 is Embodiment 1 to Embodiment 1.
It can be applied to all of the third embodiment.

FIG. 10 shows the document keyword database 43.
Shows an example of the data stored in. For example, in general, a telephone number and an address are often described in a "business card". Therefore, the keywords "TEL" and "prefecture" are associated with the document "business card" and registered in the document keyword database 43. Good. Here, for the sake of explanation, the degree of certainty for the keyword “TEL” is 70, whereas the degree of certainty for the keyword “prefecture” is 30. that is,
"Business card" is printed without the word "prefecture", and "To", "road", and "fu" are relatively common, and the keyword "prefecture" appears in "business card". It is shown assuming that the rate of doing is low. In recent years, an “e-mail address” is often written in a “business card”, so that characters or symbols frequently used in an “e-mail address” may be registered in association with a “business card”.

Since the character "prefecture" also appears in "postcard", it is sufficient to register them in association with each other. However, the more surely the postal code is entered, the lower the degree of certainty of the character "prefecture" will be, which indicates that the degree of certainty is relatively low as well. Although not shown in FIG. 10, a seven-digit postal code keyword may be registered in association with “postcard”.

In addition, since characters such as “total” and “subtotal” indicating the purchase amount and the transaction amount are often used in receipts and receipts, these keywords are linked to “receipts” and the document keyword database 43 is used. It is good to register in. Further, the symbol “￥” may be registered in association with the receipt.

It should be understood that the numerical value of the certainty factor shown in FIG. 10 is tentatively set by the applicant for the sake of explanation, and therefore lacks accuracy. In any case, the document type and the degree of certainty of the keyword that can be predicted to appear in the document are grasped and registered in the document keyword database 43.

The correspondence between the document type and the keyword is as follows.
The document type may be associated with the sum, product, and combination of a plurality of keywords. For example, considering that the “receipt” contains the total and subtotal characters and the number and date indicating the amount, ｛(total + subtotal + ￥) × (numbers) × (years + Heisei) ｝ May be associated with a “receipt” document.

Now, the "keyword" shown in FIG.
The “document type” and the “certainty” are stored in the document keyword database 43 configured in the document type identification unit 23 as described above.

By optimizing the registered keywords in the document keyword database 43, the accuracy of document type identification is increased, and the accuracy of the evaluation value of the document type candidate shown in FIG. 8 can be increased. To this end, it is effective to associate the keyword combination {(total + subtotal + ￥) × (number) × (number in the Christian era + Heisei)} with the document called “receipt” as described above. It is.
When the accuracy of the evaluation value shown in FIG. 8 is increased, the frequency of selecting and instructing the document type by the user shown in FIG. 9 is reduced, and the information processing is automatically performed.

The keyword document feature extracting unit 42 compares the document keyword information stored in the document keyword database 43 with the text information recognized by the character recognizing unit 41, and determines the type of the document according to the keyword included in the document. Then, a document keyword feature representing whether the document keyword corresponds to is extracted.

In the process selecting section 25, the document type identifying section 23
, And a processing method selected from the processing selection database 26, and the processing method is determined by the document processing units 27 to 29.
To send to.

As described above, by extracting keyword information from the text information obtained by performing character recognition and using the keyword information for extracting keyword document characteristics, more detailed document characteristics can be used. It is possible to increase the accuracy of type recognition.

Note that the third embodiment and the fourth embodiment may coexist. In this case, the accuracy of document type recognition is further improved.

(Fifth Embodiment) Next, a fifth embodiment will be described. The overall configuration will be described with reference to FIG. 1 as in the first embodiment.

The document image input / output unit 21 inputs the document 20 from a scanner or a camera and outputs a document image. The document feature extraction unit 22 extracts features of the document 20 from the document image. The document feature database 24 stores document feature information of registered document types. Document type identification unit 23
Compares the document feature information stored in the document feature database 24 with the document features extracted by the document feature extraction unit 22, and identifies the document type. The processing selection unit 25 selects a processing method from the information on the processing method stored in the processing selection database 26 and the document type output from the document type identification unit 23, and sends a document image to the document processing units 27 to 29. .

The features of this embodiment are shown in FIG.
That is, the document image reading section 51 outputs a document image by reading the document 20 from a scanner or a camera (not shown). When a document sheet is detected by the document image reading unit 51, the document detection unit 52 notifies the document image reading unit 51 that the document is in an input state.

In the information processing apparatus having the above configuration, document processing is performed as follows. "Memo", "postcard",
When the user inserts various documents 20, such as "business card" and "receipt", into the reading opening of the scanner or into the shooting range of the camera, the document detection unit 52 determines that the document is in the input state and enters the input state. Is notified to the document image reading unit 51. When receiving a notification that the document is in the input state from the document detection unit 52, the document image reading unit 51 reads the document from a scanner or a camera and obtains a document image.

The read document image is sent to the document feature extraction unit 22 shown in FIG. The document feature extraction unit 22 extracts a document feature that is a feature of the document from the document image. The document characteristics are indicated by, for example, the size of the document paper, the density of black pixels, the area of a rectangle circumscribing connected black pixels, the distribution and position of black pixels, the number and position of lines, and the like.

In the document feature database 24, information on the type of the document and the document characteristics owned by the document of the type is registered in advance. The document type identification unit 23 compares the document feature extracted by the document feature extraction unit 22 with the information of the document feature registered in the document feature database 24, and a document image in which the most suitable document type is input. Output as document type.

The process selection database 26 stores information on the type of document and the processing method for each document type. FIG. 7 shows an example of the process selection database 26. In addition. FIG. 7 has already been described in the first embodiment, and is the same in the present embodiment, so description thereof will be omitted.

In the process selecting section 25, the document type identifying section 23
, And a processing method is selected with reference to the processing selection database 26, and the document image input / output unit 21
The document image input from the selected document processing unit 27
To 29.

As described above, when the user inserts a document into the document image reading section 51, the information processing automatically proceeds. As a result, the user can perform the document processing without taking any other operation by merely loading the document into the document image input / output unit 21.

[0104]

As described above, according to the information processing apparatus and the information processing method of the present invention, the type of the input document is identified, and the document processing corresponding to the type of the document is automatically selected and executed. I do. Thus, even if the user does not know the contents of the document processing and the operation method in advance, the document processing can be performed only by inputting the document.

Further, according to the present invention, the document feature database can be updated by inputting a new type of document to be identified. As a result, a new type of document or a type of document that has been erroneously identified can be correctly identified.

Further, according to the present invention, when the accuracy of the document type identification is low, the user is inquired about the selection of the document type candidate, so that even if the accuracy of the document type identification is low,
Correct document processing suitable for the type of document can be performed.

According to the present invention, when the accuracy of document type identification is low, the user is inquired about the document type, and the document type selected by the user is different from the first candidate of the document type candidate recognized by the document type recognition unit. Sometimes, the document feature database is updated. As a result, the accuracy of the next and subsequent document type recognition is improved.

Further, according to the present invention, keyword information is extracted from text information obtained by performing character recognition on a document image, and is used as one of the document features, thereby using more detailed document features. Therefore, the accuracy of document type recognition can be increased.

In addition, according to the present invention, when a user takes in a document into a document image reading section, information processing is automatically executed. Thus, the user can perform information processing only by inserting a document into the input unit without performing other operations.

[Brief description of the drawings]

FIG. 1 is a configuration diagram showing an entire information processing apparatus according to a first embodiment of the present invention;

FIG. 2 is a configuration diagram showing a main part of the information processing apparatus according to the first embodiment of the present invention;

FIG. 3 is a configuration diagram of a document type identification unit in the processing device according to the second embodiment of the present invention;

FIG. 4 is a configuration diagram of an entire information processing apparatus according to a third embodiment of the present invention;

FIG. 5 is a configuration diagram of a document type identification unit in the information processing apparatus according to the fourth embodiment of the present invention.

FIG. 6 is a configuration diagram of a document image input / output unit in the information processing apparatus according to the fifth embodiment of the present invention.

FIG. 7 is a diagram showing a process selection database according to the first embodiment of the present invention;

FIG. 8 is a document type candidate diagram according to the third embodiment of the present invention.

FIG. 9 is a display diagram of a user document type selection unit according to the third embodiment of the present invention.

FIG. 10 is a document keyword database diagram according to the fourth embodiment of the present invention.

FIG. 11 is a configuration diagram of a conventional information processing apparatus.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 Information processing apparatus 2 Image display part 3 Scanner part 4 Storage part 5 Shape recognition part 6 Image classification part 12 Pattern recognition part 14 Application program 14a Address book 14b Housekeeping book 14c Album 15 Application program 20 Document 21 Document image input / output part 22 Document Feature extraction unit 23 Document type identification unit 24 Document feature database 25 Process selection unit 26 Process selection database 30 Document feature registration unit 31 Document type recognition unit 32 User document type selection unit 41 Character recognition unit 42 Keyword document feature extraction unit 43 Document keyword database 44 Document type recognition unit 51 Document image reading unit 52 Document detection unit

Continuing on the front page (72) Kazushima Kayashima, 1006 Kazuma Kadoma, Kadoma, Osaka Prefecture Inside Matsushita Electric Industrial Co., Ltd. (72) Keiji Ogawa 4-5-2, Takaidahondori, Higashi Osaka City, Osaka Matsushita Refrigerator Co., Ltd. F term (reference) 5B050 AA10 BA16 CA07 DA06 EA07 EA08 GA08 5B075 ND07 NK02 NK06 NR12 PP04 PP30

Claims (8)

[Claims] 1. A document image input / output unit for recognizing a document and outputting image information of the document, a document feature extraction unit for extracting a document feature from the image information of the document, and a document feature of a registered document type A document feature database for storing information;
A document feature registration unit in which features of the document are registered based on the document feature information stored in the document feature database, the document feature information extracted by the document feature extraction unit, and the document type; Means for newly registering or updating document feature information in the document feature database by the unit, and comparing the document feature information extracted by the document feature extraction unit with the document feature information stored in the document feature database to obtain the document. A document type identification unit for identifying a type, a process selection database for storing a processing method corresponding to the document type, information on the processing method stored in the process selection database, and a document type output from the document type identification unit An information processing apparatus, comprising: 2. A document type recognizing unit for comparing a document feature extracted from a document feature extracting unit with document feature information stored in a document feature database to identify a document type; 2. The information processing apparatus according to claim 1, further comprising a user document type selection unit for inquiring a user of a type. 3. A character recognition unit for recognizing characters of a document image output from a document image input / output unit,
Extracting a keyword document feature from a document keyword database storing keyword information included in a document of a registered document type, a character recognition result output from the character recognition unit, and keyword information stored in the document keyword database. A key word document feature extracting unit, and comparing the document features extracted by the document feature extracting unit and the keyword document features extracted by the keyword document feature extracting unit with the document feature information stored in the document feature database to identify a document type. The information processing apparatus according to claim 1, further comprising a document type recognizing unit that performs the processing. 4. The image input / output unit includes a document image reading unit that reads a document as an image, and a document detection unit that detects that a document has been input to the document image reading unit. Item 4. The information processing device according to item 1, 2, or 3. 5. A step of outputting a document image with a document as an input, a step of extracting document characteristics from the document image, a step of storing document characteristic information of a registered document type, and A step of newly registering and updating the document type and document characteristic information; a step of comparing the document characteristic information with the extracted document characteristics to identify a document type; and a processing method corresponding to the document type. Storing information, selecting document processing based on the stored processing method information and the registered document type information, and executing the document processing based on the selecting the document processing. An information processing method, comprising: 6. A step of outputting a document image with a document as input, supplying image information to a step of extracting document characteristics from the document image, collating document characteristic information with the extracted document characteristics, and performing document type 6. The information processing method according to claim 5, wherein the image information is provided to at least one of the step of identifying the document processing and the step of selecting the document processing. 7. In the step of comparing the document feature information with the extracted document feature to identify the document type, the user may select an already registered document type and an evaluation value indicating the relevance of the document, or 6. The information processing method according to claim 5, further comprising: communicating with a means for referring to a certainty factor indicating an appearance rate of a keyword appearing in the document. 8. A character recognition step of recognizing a character in a document image output from a document image input / output unit, wherein the step of comparing the document characteristic information and the extracted document characteristic to identify a document type; Extracting a keyword suitable for the document type from a document keyword database registered in advance based on the recognition step; and collating information from the step of extracting the keyword with document feature information stored in the document feature database. 6. A method according to claim 5, further comprising the step of recognizing a document type.