JP2007272615A - Character reader - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To allow work while making an operator performing correction work or confirmation of reading not recognize contents of information when reading the information of a slit or the like including personal information by use of an OCR device. <P>SOLUTION: A display control part 3 or 7 replaces display order of image information at random and displays it, and the operator cannot recognize the image information only as fragmentary information. <P>COPYRIGHT: (C)2008,JPO&INPIT

Description

  The present invention relates to a handwritten character recognition device (hereinafter abbreviated as an OCR device), and more particularly, to a technology for supporting confidentiality at the time of correction input of characters that could not be automatically recognized by the OCR device, that is, reject characters. It is.

2. Description of the Related Art Conventionally, in financial institutions and administrative institutions, there are many cases where necessary items are described in paper slips and forms for the purpose of ensuring the convenience of users and the reliability of paperwork.
However, even in these organizations, it is urgent to improve the efficiency of paperwork and the advancement of information processing. Therefore, there are many examples in which information described in these slips and forms is encoded by an OCR device and the encoded information is processed by a computer during the paperwork process.
However, there is a possibility that the OCR device cannot recognize characters written on a paper slip or form. Such a defect is remarkable when the described character is a so-called handwritten character. When a case where the character cannot be recognized as described above occurs, the OCR apparatus sets the unrecognizable character as a target for special processing as a reject character.

In the case that is often seen at present, as the processing of reject character in the OCR device, the image of the reject character captured by the scanner part in the OCR device is displayed on the screen, and the operator visually interprets the image and converts it into the reject character. A method of inputting a corresponding character code into the OCR device through an input unit such as a keyboard is employed.
Japanese Patent Application Laid-Open No. 07-021304

However, in the above method, information such as a slip must be shown to the operator. This is because in order for the OCR device to recognize the rejected character, it must rely on interpretation based on the visual observation of the operator.
In addition, depending on the content of the information, there is a case where the reading is performed without relying on the OCR device from the beginning and depending only on the visual observation of the operator.
However, in the information handled by the financial institutions and administrative institutions as described above, there are many pieces of information that are originally not desired to be disclosed to third parties including the operator for the purpose of confidentiality.
An example of such information that requires confidentiality is personal information, for example. Specifically, there are a name, address, telephone number, and, in a financial institution, an account number, a transaction amount, and the like.

In order to solve the above problems, in the first invention, means for randomly changing the display order of characters automatically recognized by the OCR device;
Means for displaying automatically recognized characters in the order of replacement;
Means for accepting correction input by an operator;
Means for rearranging the corrected input information in the order before random replacement and means for combining the rearranged information with the automatically recognized information are provided.

In the second invention,
Means for randomly extracting and synthesizing image information generated by reading the description content on a form by a scanner device from a plurality of forms in item units;
Means for displaying the extracted information;
Means for receiving input by an operator;
Means for associating the corrected input information with image information is provided.

By providing each of the above means, the operator can recognize the reject character and input the correction although the information can be recognized only in a fragmentary manner.
Some information that requires confidentiality, such as phone numbers and account numbers, is meaningless if the order of the numbers is changed.
In addition, there is information such as a name that has the property of increasing the importance of confidentiality when combined with other information.
For these pieces of information, it is possible to make it difficult to recognize the contents of the information by changing the order of letters and numbers constituting the information or arbitrarily changing the combination with other information.
On the other hand, as long as it is limited to the task of recognizing characters and numbers alone for correction input, there are many cases where even one character at a time can be read visually.

  When the operator visually recognizes and inputs characters that cannot be automatically recognized, and when the operator confirms whether automatic recognition is correct or not, the order of image display for the operator to visually check is random for each character or slip item Thus, the operator can recognize the information to be visually checked only to the extent necessary for correction or confirmation.

Hereinafter, an example of the form that the applicant considers best in carrying out the present invention will be described. FIG. 1 shows an outline of the configuration of an OCR apparatus to which the present invention is applied.
In FIG. 1, reference numeral 1 denotes a scanner unit. The scanner unit 1 recognizes information such as characters and numbers written on a slip to be read, and converts this information into image data composed of a bitmap, as in a scanner used in a conventional OCR apparatus.

  Reference numeral 2 denotes an automatic recognition unit. The automatic recognition unit 2 analyzes the image data generated by the scanner unit 1 through conversion, specifies a code corresponding to characters or numbers written on a slip or the like, and generates a character code string.

When there is information in the image data that cannot identify the corresponding code, this information is recognized as an unreadable reject character.
Further, the automatic recognition unit 2 recognizes which of the character strings on the slip is information corresponding to each item for each item such as name and address described in the slip.
For this recognition, there may be a method of recognizing image data recognized by the scanner unit in the description frame as information corresponding to the description frame based on the position of the description frame pre-printed on the slip. .

  Reference numeral 3 denotes a display control unit. The display control unit 3 causes the display unit 4 to display the image data and the automatic recognition result in association with each other, which will be described later.

  Reference numeral 4 denotes a display unit. The display unit 4 displays the image data and the automatic recognition result in association with each other based on the control from the display control unit 3.

Reference numeral 5 denotes an image data storage unit.
The image data storage unit 5, the recognition result storage unit 6, decomposes the image data generated by the scanner unit 1 for each character, and then gives a character ID, name, address, account number given to each character The item ID assigned to each character string such as the item ID, the slip ID assigned to each slip, and the like are stored in association with each character. ,

Reference numeral 6 denotes a recognition result storage unit.
The recognition result storage unit 6 is a character ID, name, address, account number, etc. given to each character of a character string generated as a result of character recognition by the automatic recognition unit 2 from the image data generated by the scanner unit 1. The item ID given for each character string of item 1 and the slip ID given for each slip are stored in association with each character. ,

  Reference numeral 7 denotes a display control unit. Similar to the display control unit 3 described above, the display control unit 7 causes the display unit 8 to display image data and an automatic recognition result in association with each other, which will be described later.

  Reference numeral 8 denotes a display unit. Similar to the above-described display control unit 4, the display unit 8 displays the image data and the automatic recognition result in association with each other based on the control from the display control unit 7.

  Reference numeral 9 denotes a correction input storage unit. The correction input storage unit 9 stores the content input by the operator in the correction input unit 10 described below in association with an ID described later, and transfers the stored content to the recognition result storage unit 6.

  Reference numeral 10 denotes a correction input unit. The correction input unit 10 receives a correction input by the operator and transfers the received correction input to the correction input storage unit 9.

The operation of the OCR apparatus of the present invention whose configuration has been described so far will be described below with reference to the flowcharts of FIG.
First, automatic character recognition will be described with reference to the flowchart of FIG.

  When a slip (not shown) to be read is set in the scanner unit 1, the scanner unit 1 scans the slip in step 201 to generate image data of the slip. At this time, an entry frame or the like preprinted on the slip is printed by so-called dropout color that is not recognized by the scanner unit 1 and is not included in the image data. For this reason, the image data includes only an image of character and number information such as name, address, and account number entered by the customer on the slip. At this stage, the generated image data is one image corresponding to the entire slip for one slip.

  The generated image data is sent to the image data storage unit 5. Here, in step 202, a slip ID is given to the image data for one piece of slip in the image data storage unit 5.

  Next, in step 203, the slip image is analyzed in the image data storage unit 5, and the image data generated by reading the information described at a predetermined position on the slip designated in advance is divided and extracted in units of items. The In this extraction, image data corresponding to a range corresponding to the inside of a predetermined entry frame corresponding to each item on the slip is extracted as image data corresponding to the item.

  Next, in step 204, an item ID is given to the image data for each divided item in the image data storage unit 5.

  Next, in step 205, the image data for each divided item in the image data storage unit 5 is sent to the automatic recognition unit 2 shown in FIG. In the automatic recognition unit 2, the transmitted image data is further divided into character units. The processes after the segmentation are performed as a part of the known character recognition process.

  In step 206, the automatic recognition unit 2 assigns a character ID to the character unit image data segmented in step 205. The image data separated from the character ID is sent again to the image data storage unit 5 and stored in association with each other.

  Next, in step 207, the automatic recognition unit 2 performs character recognition based on the character unit image data segmented in step 205. Further, in step 208, the characters recognized by the automatic recognition unit 2 are collected for each item, and a character string for each item is created. Thereby, for example, all the characters recognized in the address portion are integrated into one character string and become meaningful as address information.

  The character string created in this way is sent to the recognition result storage unit 6 of FIG. 1. At the same time, the character ID and item ID of the image data used as the basis of recognition are transferred from the automatic recognition unit 2 to the recognition result storage unit 6. The ID is sent to the recognition result storage unit 6, and in steps 209 and 210, these IDs are given to the character string of the recognition result. At this time, the same ID is assigned to both the character ID and the item ID to the image data that is the basis of recognition and the character or character string recognized from the image data.

By this ID assigning process, it is possible to refer to the image data before recognition and the characters after recognition or the r time sequence in association with each other. That is, by extracting image data having the same ID and characters or character strings, it is possible to compare the data before character recognition with the data after character recognition.
With respect to the information of the portion where the automatic recognition is performed, a confirmation correction operation for confirming whether the automatic recognition is correctly performed is performed. The confirmation correction process will be described with reference to the flowchart of FIG.

  First, in the display control unit 3 shown in FIG. 1, a slip ID is designated in step 301, an item ID is designated in step 302, and one item to be displayed for confirmation correction is designated.

  Next, in step 303, the display control unit 3 generates a pseudo random number, designates a character ID using the pseudo random number, and among the designated items, for one character of the character ID designated by the pseudo random number, image data and Both characters of the automatic recognition result are displayed on the screen of the display unit. Here, the image data and the character of the automatic recognition result have the same character ID, and both of them have the relationship that the image data recognized by the automatic recognition unit 2 is the character of the automatic recognition result. is there.

In step 305 and step 306, the display control unit 3 sends the image data and the characters of the automatic recognition result to the display unit 4 to display them on the screen of the display unit 4.
The operator who is in charge of the confirmation and correction work visually confirms this display on the screen of the display unit 4 and confirms whether or not the automatic recognition is correctly performed for each character. As a result, if a misrecognition is found, the operator corrects the automatic recognition result as appropriate. However, since this correction process is well known to those skilled in the art, a description thereof will be omitted. The series of processes shown in FIG. 3 is executed for all items that are the targets of automatic recognition. After that, the corrected automatic recognition result is stored in the recognition result storage unit 6.

  For items that are not subject to automatic recognition other than the above confirmation correction work, the operator performs correction input work, the image is visually checked by the operator, the operator determines the contents, and the information corresponding to the contents described in the slip is Enter. Hereinafter, this correction input operation will be described.

FIG. 4 shows a flowchart in the correction input operation.
First, in step 401, the display control unit 7 specifies a slip ID. Here, a plurality of IDs are specified at the same time. As described later, by mixing information extracted from a plurality of slips, the relationship between a plurality of items on the same slip, such as names and account numbers, will be described later. This is to make it difficult to guess from the screen display.

Next, in step 402, the display control unit 7 specifies an item ID.
Next, in step 403, the display control unit 7 generates a pseudo random number.
Next, in step 404, the display control unit 7 specifies the slip IDs based on the generated pseudo-random numbers and rearranges the slip IDs.

Next, in step 405, the display control unit 7 first sets the image data of the item corresponding to the item ID specified in step 402 out of the image data of the first slip in accordance with the order of the slip IDs rearranged in step 404. Is called from the image storage unit 5 and stored.
This operation is repeated for all item IDs. When this process ends, the display control unit 7 stores image data of one slip. However, in this state, there is no relationship between the image data of a plurality of items. For example, in the image data, the name information and the address information are simply read from another slip at the same time according to a pseudo-random number. What is the personal information between the name and address? The relationship is not guaranteed.

Next, in step 406, the display control unit 7 displays these image data on the screen of the display unit 8.
Looking at the display content of the image data, in step 407, the operator who is in charge of the correction input operation inputs characters or numbers corresponding to the image data from the character input unit 10.

When input is performed, in step 408, the display control unit 7 reads out the slip ID and the item ID with respect to the image data displayed on the screen of the display unit 8, and inputs the slip ID and the item ID. It is stored in the input information storage unit 9 in association with the written character or number data.
After that, the character or number data stored in the input information storage unit 9 is sent to the recognition result storage unit 6 together with the slip ID and the item ID, where the automatic recognition result confirmed by the confirmation correction work and the correction The correction input result recognized and input by the input operation is integrated.

  The present invention is applicable to an OCR device, and is particularly suitable for an OCR device that reads personal information that requires confidentiality of information.

It is explanatory drawing which shows the outline | summary of a system configuration. It is a flowchart which shows the process of an automatic recognition process. It is a flowchart which shows the process of a correction | amendment input process. It is a flowchart which shows the process of a confirmation correction process.

Explanation of symbols

1 Scanner Unit 2 Automatic Recognition Unit 3 Display Control Unit 4 Correction Input Unit 5 Correction Unit 6 Recognition Result Storage Unit 7 Image Data Storage Unit

Claims (2)

Means for randomly changing the display order of characters automatically recognized by the OCR device;
Means for displaying automatically recognized characters in the order of replacement;
Means for accepting correction input by an operator;
A character reading apparatus, comprising: means for rearranging corrected input information in an order prior to random replacement; and means for combining the rearranged information with automatically recognized information. Means for randomly extracting and synthesizing image information generated by reading the description content on a form by a scanner device from a plurality of forms in item units;
Means for displaying the extracted information;
Means for receiving input by an operator;
A character reading apparatus, characterized in that means for associating corrected input information with image information is provided.

Images