JPH04293185A - Filing device - Google Patents

Abstract

PURPOSE:To offer the filing device whose operability is good, and also, to make the storage capacity small. CONSTITUTION:By a scanner 1, a document is read in as an image, and a character is recognized by an OCR recognizing part 6. Subsequently, in the case a recognized character is recognized by higher accuracy than the degree of conviction set by a condition setting part 7, it is converted to a code and filed in a filing part 9. On the other hand, as for a character being lower than the degree of conviction, it is not converted to a code but filed as image information. In the case of filing it, each separate management information is added to the code information and the image information by a file management information adding part 10. The character subjected to filing as an image to a filing part 9 is read out of the added management information, and converted into code information by on-line or a batch processing.

Description

[Detailed description of the invention]

0001

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a filing device, and more particularly to a filing device that performs input using OCR, which optically recognizes characters.

0002

[Background Art] In order to efficiently manage various types of information,
2. Description of the Related Art Filing has become increasingly common using information processing devices such as computers. In such filing systems, O
Some use CR (optical character recognition device). In other words, typed or handwritten characters,
This allows efficient input by directly inputting information into an information processing device such as a word processor or computer without requiring the operator to operate the keyboard. This OCR scans printed characters, recognizes each character from the amount of reflected light, and converts it into a predetermined code. On the other hand, as a method of filing character and graphic information into an information processing apparatus, inputting the document by reading the document as image information in the same way as graphics is also practiced.

0003

[Problem to be Solved by the Invention] When characters are recognized and input using OCR, no matter how good the OCR device is, it is impossible to achieve a recognition rate of 100. Therefore, the recognized characters are displayed on a display device, The operator finds characters that are misrecognized and corrects them one by one. Therefore, it takes time to correct misrecognized characters, which poses a problem in terms of operability. Furthermore, when filing documents as images, a large capacity storage device is required. Therefore, the first object of the present invention is to provide a filing device with good operability, and the second object is to reduce the storage capacity.

0004

[Means for Solving the Problems] The invention according to claim 1 includes: image reading means for reading a document as an image;
Character recognition means for recognizing the characters in the image read by the image reading means, and confidence determining means for determining whether the characters recognized by the character recognition means have been correctly recognized with a confidence higher than a predetermined confidence. and a filing means for filing characters judged to have a higher certainty than a predetermined certainty by the certainty judging means as a code and filing the characters judged to have a low recognition rate as images. . In the invention as set forth in claim 2, the filing device as set forth in claim 1 is provided with a certainty specifying means for specifying the predetermined certainty determined by the certainty determining means as a percentage. In the invention according to claim 3, display means displays characters filed as images in the filing means at the time of filing;
The filing device according to claim 1 or 2 is provided with a conversion means for converting the characters of the image displayed on the display means into codes. The invention according to claim 4 further comprises a display means for collectively displaying a plurality of characters filed as an image in a filing means, and a conversion means for converting the characters of the image displayed on the display means into a code. The filing device described in No. 2 is equipped with the filing device.

[0005]

[Operation] The present invention files images of characters read by an image reading means, characters that are judged to be accurately recognized as codes, and files characters that may be misrecognized as images instead of codes. File as. Then, if necessary, the characters filed as images are code-converted by a code conversion means by online or batch processing.

[0006]

DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the filing apparatus of the present invention will be described in detail below with reference to FIGS. 1 to 5. FIG. 1 shows the configuration of a filing device. Reference numeral 1 denotes a scanner for reading documents such as characters and images, and the documents read by this scanner 1 are stored in an image memory 2. A document reading unit 4 located in the control unit 3 controls the reading of documents by the scanner 1. Reference numeral 5 denotes a text determination unit that determines whether the document read by the scanner 1 is a text portion consisting of character information or an image portion consisting of image information. Reference numeral 6 denotes an OCR recognition unit that recognizes characters in the text portion, and is equipped with Japanese kanji recognition and English text recognition functions.

Reference numeral 7 denotes a condition setting section in which various conditions are set, such as setting the reliability of character recognition in the text portion and setting a mode for correcting misidentification. 8 is the confidence check part, O
A condition setting unit 7 sets the reliability of characters recognized by the CR recognition unit 6.
It is checked whether the confidence level is higher than the confidence level set in . Reference numeral 9 denotes a filing unit that files information that has been encoded through character recognition, character information that is not encoded and is in an image state, and an image portion, together with file management information that distinguishes each of them. Reference numeral 10 denotes a file management information adding section that adds file management information to each piece of information stored in the filing section 9. 11 is a word processing unit equipped with document creation functions including text editing functions, 12 is a utility, 13 is a text and image display unit, 14 is a batch misidentification correction unit, 15 is a font conversion unit, 16 is a garbage collection unit, and 17 is an OCR unit. This is the activity report publishing department.

The document filing operation performed by the filing device configured as described above will be explained with reference to FIG. 2. First, the operator uses the condition setting section 7 to set the confidence level for OCR recognition of a text portion included in a document. Then, the document set in the scanner 1 is read into the image memory 2 under the control of the document reading section 4. Input the image stored in image memory 2 (step 1)
, the text determining unit 5 determines whether it is a text portion or an image portion (step 2). If it is an image part (Step 2; N),
The file management information adding unit 10 creates a management header for filing as an image (step 3), and files it together with the management header in the filing unit 9 (
Step 4) End the process.

On the other hand, if the input image is a text part (step 2; Y), it is supplied to the OCR recognition unit 6 and each character is OCR recognized (step 5), and the confidence level of the recognized character is determined by the condition setting unit. It is determined whether the confidence level is greater than the confidence level set in step 7 (step 6). If it is larger than the set confidence level (Step 6; Y), after creating a management header for filing as text (Step 7), the code information of the recognized character and the management header are filed in the filing section 9. (Step 4), the process ends. If the confidence level of OCR recognition is lower than the set confidence level in step 6 (N), create management information for filing as an image (step 3),
The recognized characters are filed as image information together with management information without code conversion (step 4), and the process ends.

FIG. 3 shows the concept of the file format when filing in the filing section. The file header created by the file management information addition unit 10 has a pointer to the document header, and the document header has the following information. In other words, pointers to images in a document, pointers to text in a document, information about the types of fonts appearing in a document and where each font is located in the document, and information about the structure of the document. This includes information such as writing, horizontal writing, paragraphs, number of lines, number of images, document size, document printing quality, and information regarding garbage collection. Among the characters in the text portion, management information shown in FIG. 4 is added to misrecognized candidate characters whose reliability recognized by the OCR recognition unit 6 is lower than the reliability set by the condition setting unit 7. Here, 0xFFFF indicates that the next value is a pointer to a group of misrecognized candidate character images. In this way, misrecognized candidate characters are conceptually filed and managed as a separate group of images using a specific management header.

Next, the operation of correcting characters recognized by the OCR recognition section 6 will be explained. This device has two methods for correcting misidentification.
There is one. There are an online correction method and a batch mode correction method, and which mode is used is set by the operator in the condition setting section 7. The online correction method is a method that has been used conventionally, and when recognizing the text part using OCR,
This method prompts the user to correct the error each time a misidentification occurs. The batch mode correction method means that when multiple documents are filed using the above method, the characters with low confidence that were filed as images are manually corrected by the user operator.
It converts it into code.

FIG. 5 shows the flow of batch mode correction operation. First, a document filed in the filing section 9 is read from the word processing section 11 (step 1), and a misrecognition display format is selected according to the settings of the condition setting section 7 (step 2), as well as whether an OCR report is necessary or not. (Step 3) If an OCR report is necessary, (Y) the OCR activity report issuing unit 17 issues the OCR report. After issuing the report, or
If the CR report is not required (step 3; N), the batch misrecognition correction unit 14 converts it into a corresponding code according to the instructions from the word processor 11 (step 5).

[0014] After correcting the misrecognition, if the image is unnecessary, the image is separated from the document by manipulating the management information. When the number of separated images reaches a certain amount (step 6; Y), the garbage collection unit 16 organizes the storage area of the external storage device (step 7), and the process ends. When reading a document in step 1 of the above correction operations, the characters stored as images are attached with the header shown in Figure 4, so the word processor
1 can be read out separately from coded characters and image parts. Note that it is also possible to display all text and display characters with low confidence in high brightness, invert them, display them in different colors, or display them with various attributes added. Regarding these, the display state in which the operator corrects the misrecognized characters depends on the settings of the condition setting section 7.

[0015] Furthermore, the user can perform work by referring to the report issued by the OCR activity report issuing section 17. The report will show you which parts of the document (
The number of pages in the report, the percentage of misrecognitions, the time required for OCR recognition, etc. are shown. The operator also sets in the condition setting unit 7 whether this report is actually output on paper or displayed in a window on the display screen.

[0016]

As described above, according to the present invention, even if OCR recognition and filing is performed without human intervention, incorrect characters will not be filed. Furthermore, the storage area of the external storage device can be saved because OCR recognition is performed, the code is once written, and then filed. Furthermore, by allowing the user to specify the degree of certainty, the text portion can be forcibly converted into a code. Specifying to file documents of low importance is effective when storage space is insufficient. By correcting misrecognition in batch mode, it becomes possible to perform efficient correction.

[Brief explanation of drawings]

FIG. 1 is a block diagram showing the configuration of an embodiment of a filing device of the present invention.

FIG. 2 is a flowchart showing a document filing operation by the filing device.

FIG. 3 is a conceptual diagram showing a format for filing in a filing section.

FIG. 4 is a diagram showing the structure of management information for characters stored as images.

FIG. 5 is a flowchart illustrating the operation of batch mode correction.

[Explanation of symbols]

1 Scanner 5 Text judgment section 6 OCR recognition section 7 Condition setting section 8 Confidence check section 9 Filing department 10 File management information addition section

Claims (4)

[Claims] Claim 1: an image reading means for reading a document as an image; a character recognition means for recognizing characters in the image read by the image reading means; a reliability determining means for determining whether or not the characters have been correctly recognized based on the degree of recognition; A filing device comprising a filing means for filing characters as images. 2. The filing device according to claim 1, further comprising certainty specifying means for specifying the predetermined certainty determined by the certainty judging means as a percentage. 3. The apparatus according to claim 1 or 2, further comprising display means for displaying characters filed as images in the filing means at the time of filing, and conversion means for converting the characters of the image displayed on the display means into codes. filing device. 4. Claim 1 or 2, further comprising display means for collectively displaying a plurality of characters filed as images on the filing means, and conversion means for converting the characters of the image displayed on the display means into codes. Filing device as described.