JPS62165269A - Business card file method - Google Patents

Abstract

PURPOSE:To enable an automatic classification and a registration process by discriminating the partition of each character string in a new business card, if the card has the same structure, based on the partition of a character string content, and registering the content of a prescribed character string on a table for a retrieval. CONSTITUTION:The retrieval for a department name is performed by inputting the department name from a keyboard 3, finding a record having the department name from a business card administration table 30, and displaying a business card picture on a display device 5. When plural business cards in the same department are registered, they are displayed by n-number (n>=1) of pictures in order on one screen corresponding to an indication from the keyboard 3 or a PD4. It is possible, even when n>=2, to display the whole plane of each business card, and when necessary, it is outputted to a picture data output device 9, then a hard copy being generated. The retrieval for a name is performed by inputting a name data from the keyboard 3, generating a font pattern, extracting a feature quantity for each character, and retrieving the said business card performing a feature matching with a name part feature quantity registered at the business card administration table 30. In this way, a filing system can be configured automatically, and an efficient registration can be realized.

Description

[Detailed description of the invention] [Industrial application field] The present invention relates to a business card file system.

[Technology of invention]

When managing business cards using a computer, the easiest method using conventional technology is to number the actual business cards, put them in a sorting box, and then use that number, affiliation, name, etc.
This is a method of managing addresses and other information in a computer database. Alternatively, a business card may be input as image data into a document image file, and this may be searched.

As a method for automating the input of names, etc., there is a known method, for example, as described in Japanese Patent Laid-Open No. 60-69788, in which information on the front side of a business card is input using a barcode printed on the back side. ing. In addition, as a method of using the structure information of one business card, for example, Information Processing Society of Japan Tube 22
As seen in [Structural storage of business card images and search using feature parameters] reported in Proceedings of the Japan National Conference, 3D-6, pp. 643-644, only business card images are input and stored. A method is known in which a character string in the name part is extracted using knowledge about the structure of a business card, and a search is performed using characteristic parameters of each character.

[Problem that the invention seeks to solve]

Among these conventional techniques, the method using a business card sorting box and database uses actual business cards.

There are problems in terms of space and management, the fact that the management information and the entity are separate, and the work of inputting names of affiliations and other information into the database. The method using a document image file also has the same problem of input work such as two names.

The barcode method solves the data entry problem, but it also requires processing of business cards with double-sided printing (Japanese on the front and English on the back), processing of existing business cards without barcodes, or standardization for all business cards. Practical problems remain, such as the problem of unifying barcodes.

In the method of extracting the character string of the name part and performing a search using the characteristic parameters of the character, the structural information of the business card has conventionally been applied only to the extraction of the character string of one name part.
It is not used for data storage or retrieval. Furthermore, conventionally, all business card data is uniformly stored without being classified, and there is a problem in that the time required for searching increases as the number of stored data increases.

An object of the present invention is to provide a business card file system that can automatically classify and store business cards by simply inputting them as image data without adding special information to them.

[Means for solving problems]

The above purpose is to store the character strings and non-text areas of each input business card either as is or compressed, and to store the structural characteristic information of each character string, such as the position and size of each area, and the first and last names. The number of constituent characters and the characteristic parameters of each character are registered in the data file, and when inputting business card data, area extraction processing is performed on the input business card image to separate character string areas and non-text areas such as marks. After extracting, refer to the registered data above to determine which information category the extracted character string corresponds to, such as affiliation, location, name, etc. Department etc.
This is achieved by determining whether there are any similarities in the structure of business cards.

The character string region and non-character region can be extracted using a method that utilizes marginal distribution and connected components of black pixels, which are known in the field of document image processing, for example, Journal of the Institute of Electronics and Communication Engineers, Vol. J66.
D.No.1. A method such as ``Paper component extraction method j that does not depend on format specification information'' on pages 111 to 118 can be applied.

The first method to identify the information classification of each character string is to match the character string structure of registered business cards, for which the nature of each character string is already known, with the character string structure of input business cards. If there is a registered business card, the content of the character string is identified by determining the correspondence between the character strings. Matching of string structures.

For example, the character string area and the non-character area are each represented as a rectangle, and the internal area of each rectangle is represented as #1 n, and the other area is represented as re O, etc., and the cross-correlation of the image is calculated from the degree of pattern matching. Just ask. If there are no registered business cards with a similar structure, use an input device such as a keyboard,
Enter classification information for each character string.

The second method for identifying each character string is to locate the character string in the center of the business card, the size of its constituent characters is equal to or larger than the characters in other areas, and the number of characters included is 2 to 8. Assume that the string of characters is the name part. Next, when the business card is arranged vertically, if there is a string of characters above the name part that is smaller than the characters that make up the name part, this is determined to be a normal posture.
If the lowercase character string is on the lower side, it is determined that the character string is in the opposite direction, and a 180 degree rotation process is performed. If the lowercase string is neither above nor below the name section, compare the width or font size of the strings located on the left and right sides of the name section. If the right side is larger, it is normal, and if the left side is larger, it is backwards. The input image is rotated 180 degrees based on the judgment. If the left and right character strings are equal, correct the input image so that the one with the largest number of character strings is on the left,
After that, the strings on the right side of the name section are the affiliation section, the strings on the left are the address section, and if there are strings above, the title section,
Other data areas are determined to be mark areas.

[Effect]

When a business card is input as image data, each character string area and non-character area are extracted, and the content category of each character string is identified. In this case, it is checked whether or not there is any business card data with a similar structure among the registered business card data, and the mark portion is checked by pattern matching. If there are business cards with a similar structure, they are classified as business cards belonging to the same organization.

be registered. If there is no similar structure among the registered data, the content category can be specified using an input device such as a keyboard. In this way, business cards of the same affiliation that are input thereafter can be automatically classified and registered simply by performing an image input operation.

In addition, by storing characteristic parameters for each character in the surname and name part, during the search process, business cards for each classified department are displayed, and the operator can specify the name by converting kana to kanji or inputting handwritten characters. To enable business card search using the first and last name as a key, based on the characteristics of the character pattern of the first and last name.

〔Example〕

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

FIG. 1 is a hardware configuration diagram of a business card file system according to the present invention, including an image input device 7 for inputting business cards as image data, and an image input device 7 for temporarily storing input images for processing such as structure extraction. Memory 8. A file device 6 for storing yK images and secondary information accompanying them, a CPU for processing image data and controlling the operation of the entire system, and a main memory for storing programs and data necessary for the operation of the CPU. 2. Keyboard 3 for inputting data and control commands, Pointing device (hereinafter abbreviated as PD) 4, Display 5
, an image output device 9. Note that in order to speed up image processing, a dedicated image processing device may be provided in addition to the CPU.

FIG. 2 shows the control procedure for business card registration processing.

The business card is input into the image memory 8 from the image input device 7 (
After step 100), character strings are extracted and it is identified to which category each character string corresponds, such as affiliation, name, title, etc. (step 110).

Here, the business card 14 is attached to the department 15, as shown in FIG.
, name section 16, title section 17, address section 18, mark section 19
It shall consist of each area.

Regarding the name part 16, after cutting out individual characters, the feature amount of each character is determined. In order to classify business cards belonging to the same affiliation, the file 6 has a classification table 20 for storing the name of the affiliation and the structure information 2 of the business card 2, as shown in FIG. 4, and the format of the representative business card. Structure matching is performed to find out which business card 6 structure of which affiliation the input business card structure is similar to is already stored in the classification table 20 (step 120). If necessary, the representative business card of the affiliation determined to be similar may be displayed on the display 10 together with the input business card, and the user may confirm it using the keyboard 13 or PDII. If there is no business card belonging to the same affiliation, the affiliation name is entered from the keyboard 13 and registration processing is performed, the affiliation name and structure information are registered in the 1-class table 20, and the ID is written using this business card as a representative business card (step 190). If there is a business card belonging to the same organization, the name part of the business card belonging to that organization is matched using the characteristic parameters, and the presence or absence of a registered business card with the same name is checked (step 13).
0,140). If not, perform the registration process, and if there is,
The business card is displayed and instructions for the next process are input (steps 150, 160').

If data replacement is instructed, delete the corresponding registered business card (step 180), u-recording process 200
Do this. In the case of a registration instruction, the registration process is performed as is. In the case of a termination instruction, the process terminates without performing the registration process.

In the registration process 200, the original image data of the business card is accumulated for each part in the original image file area on the file device 6, and the address, the position and size of each part, the number of characters in the name part, and the feature amount of each character are stored. is written into the file device 6 in the form of a business card management table 30 shown in FIG. 5 along with a business card ID that uniquely indicates the business card.

The business card search process is performed as follows.

Affiliation name search: Enter the affiliation name from the keyboard 3, and search for the business card management table 3.
A record having the above-mentioned affiliation name is found from 0, and its business card image is displayed on the display 5. If multiple business cards are registered for the same affiliation, the number n (n≧1) will be displayed on one screen according to instructions from the keyboard 3 or PD 4.
Display one after another. Even in the case of n≧2, it is possible to display each business card in its entirety, and if necessary, output it to the image data output device 9 to create a hard copy. In addition, instead of entering the affiliation name from the keyboard 3, 1 classification table 2
The representative business cards of 0 may be displayed entirely or partially, for example, only the affiliation name, and the above-described process may be performed on the business card data of the affiliation name selected from among them. Alternatively, a list of mark portions of business cards registered in the classification table 20 may be displayed, and the above-described processing may be performed on business card data having a designated mark.

Name search: Enter name data from the keyboard 3 using, for example, kana-kanji conversion, generate a font pattern, extract features for each character, and search the business card management table 3.
The corresponding business card is searched by performing feature matching with the name part feature amount registered in 0.

Note that the name may be input using the PD 4 or by inputting handwritten characters from the image input device 1.

Structural information search: Structural information is extracted from the business card input from the image input device 6, and business cards with similar structures are searched for by structure matching. In this case, using PD4, display 5
Specify the position and size of each part graphically above,
You may also search for business cards that have a similar structure.

Note that in the above embodiment, the business card is searched without performing character recognition, but after recognizing the characters in each part printed on the business card, the data is stored in the file device 6 and this is searched. According to this embodiment, when classifying input business card images, it is not necessary to perform structural matching on all registered business cards, and the affiliation data registered in the classification table 20 can be used for classification. By using this, the time required for classification processing can be significantly reduced. Further, when registering a business card, if the affiliation is determined by structural matching, there is an advantage that it can be checked in a short time whether the same business card has already been registered by name feature matching against registered business cards belonging to the same affiliation.

FIG. 6 is a flowchart 1 showing a modification of the present invention. Steps 200-230 are steps 100 in FIG.
Corresponds to ~125. If there is no character string structure that is the same as the input image in the registered data, refer to the display contents and have the operator input classification information for each character string and the affiliation name that will serve as a search key from the keyboard 3 ( Steps 240.250). If there are any with the same structure, the classification of each character string is estimated according to the registered W format, the classification information and affiliation name are displayed, and any errors are corrected (steps 260 and 270). Next, in step 280) of cutting out the characters in the name part, the feature amount of each character is determined, and the data of the registered business cards of the same organization is targeted.
The name portion is matched (step 290). The following processing (steps 300 to 340) is the same as in the previous embodiment.

〔Effect of the invention〕

As is clear from the above embodiments, according to the present invention, registration and search are performed by focusing on the unique structure of business cards, such as nine names of affiliations. If it is registered, it can be automatically classified and registered by affiliation, and only if the name φ11 of the same affiliation is not registered, by entering additional information such as affiliation basis, Business cards can be automatically classified. In this way, the file system is automatically constructed as it is used, allowing efficient registration.

[Brief explanation of drawings]

Figure 1 is a system diagram showing an embodiment of the business card file system of the present invention, and Figure 2 is a diagram showing one example of business card data registration processing.
Flowchart showing the embodiment, Fig. 3 is an explanatory diagram of structural information of business cards, Fig. 4 is a diagram showing the structure of the classification table, Fig. 5 is a diagram showing the structure of the business card management table, and Fig. 6 is a diagram showing the structure of business card data. It is a flowchart which shows another example of registration processing. Ginin sacrifice, 3rd eye

Claims (1)

[Claims] 1. In a business card file system that stores business cards as image data, information indicating the structural characteristics and content classification of each character string for registered business cards is stored, and information on business cards to be newly registered is stored. Extract each character string from the image data of A business card file method characterized in that the classification of each character string of a business card is determined, and the contents of a predetermined character string are registered in a table for search. 2. If a business card with the same structure as a new business card to be registered is not found in the existing registered data, separately input the character string classification information of the business card and register the information in the search table based on it. The business card file method described in item 1, which is characterized by: 3. Identify the character string corresponding to the name from the position of the character string extracted from the image data and the size of the characters contained in it, and identify the content classification of other character strings from the relative positional relationship with the name character string. 2. The business card file method according to claim 1, wherein the information is registered in the search table based on the judgment.