Text information recognition device, method, and program

Abstract

To provide a text information recognition device, a method, and a program capable of immediately identifying misrecognition of text information.SOLUTION: A device includes: a text information recognition part 21 that accepts read document data of a document containing text information and recognizes the text information contained in the document data; and a table generation part 22 that generates a table that in which the recognized text information is associated with predefined correct interpretation text information that may be included in the document.SELECTED DRAWING: Figure 1

Description

Technical Field

[0001] The present invention relates to a text information recognition device, a method, and a program for recognizing text information included in read manuscript data.

Background Art

[0002] Conventionally, a manuscript including handwritten text information has been optically read to obtain manuscript data, and the text information included in the manuscript data has been automatically recognized.

[0003] For example, attempts have been made to perform automatic scoring by reading an answer written by hand, such as a calculation print of addition or subtraction, to obtain manuscript data and automatically recognizing the text information of the answer included in the manuscript data.

[0004] However, since it is handwritten text information, misrecognition may occur when automatically recognized, and as a result, incorrect scoring may occur. Therefore, finally, it is necessary for a person to confirm the misrecognition.

[0005] For example, Patent Document 1 proposes a method of visually checking the recognition result of handwritten characters and, when there is misrecognition, specifying the character type and re-recognizing.

Prior Art Documents

Patent Documents

[0006]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0007] However, in Patent Document 1, when the user visually inspects the recognition result, the misrecognized characters are simply displayed as they are, making it difficult to immediately identify which characters were misrecognized. For example, in Patent Document 1, if the handwritten text information "This product costs 8320 yen" is recognized and "3" is misrecognized as "ro" and "0" (zero) is misrecognized as "o" (oh), the misrecognized result "This product costs 8ro20 yen" is simply displayed. Therefore, it is difficult to immediately identify the misrecognized characters among a mixture of numbers, hiragana, and alphabet characters.

[0008] In view of the above circumstances, the present invention aims to provide a text information recognition device, method, and program that can immediately identify misrecognitions of text information. [Means for solving the problem]

[0009] The text information recognition device of the present invention comprises a text information recognition unit that receives document data obtained by reading a document containing text information and recognizes the text information contained in the document data, and a table generation unit that generates a table that associates the recognized text information with pre-set correct text information that may be contained in the document. [Effects of the Invention]

[0010] According to the text information recognition device of the present invention, a table is generated that associates the recognized text information with pre-set correct text information that may be included in the document, so that misrecognition of text information can be immediately identified. [Brief explanation of the drawing]

[0011] [Figure 1] Block diagram showing the schematic configuration of a text recognition system using one embodiment of the text information recognition device of the present invention. [Figure 2] This diagram shows an example of a table that maps recognized text information to pre-configured correct text information. [Figure 3] A diagram showing an example of a table in case of misrecognition. [Figure 4] This diagram shows an example of a screen displaying manuscript data and a table simultaneously. [Figure 5] A diagram showing an example of a correction screen. [Figure 6] A flowchart illustrating the processing flow of a text recognition system using one embodiment of the text information recognition device of the present invention. [Modes for carrying out the invention]

[0012] Hereinafter, a text information recognition system using one embodiment of the text information recognition device of the present invention will be described in detail with reference to the drawings. Figure 1 is a block diagram showing the schematic configuration of the text information recognition system 1 of this embodiment.

[0013] As shown in Figure 1, the text information recognition system 1 of this embodiment comprises a document reading device 10, a text information recognition device 20, a display device 30, and an input device 40.

[0014] The document reader 10 photoelectrically reads a document containing text information and outputs document data. Any document containing text information such as letters, numbers, and symbols is acceptable as a document, but examples include documents with handwritten numbers such as multiplication worksheets or calculation worksheets, documents with handwritten English such as English test printouts, and documents with handwritten lists of email addresses including "@".

[0015] The text information recognition device 20 comprises a text information recognition unit 21, a table generation unit 22, and a display control unit 23.

[0016] The text information recognition unit 21 receives the manuscript data output from the manuscript reading device 10 and recognizes the text information included in the manuscript data. The text information is, as described above, characters, numbers, symbols, etc. The characters include all characters of all languages regardless of the language. The text information recognition unit 21 recognizes the text information using, for example, OCR (Optical Character Recognition) technology. However, as a method for recognizing text information, not limited to this, known methods can be used. For example, the text information may be recognized using a convolutional neural network (Convolutional Neural Network).

[0017] The table generation unit 22 generates a table in which the text information recognized by the text information recognition unit 21 is associated with the preset correct text information. The preset correct text information is the text information that may be included in the manuscript read by the manuscript reading device 10. For example, when the manuscript is the above-mentioned multiplication print or addition print, etc., the correct text information is the numbers from 0 to 9 written by hand in the answer column. Also, when the manuscript is an English test print, the correct text information is the alphabet from a to z. Also, when the manuscript is a list of email addresses, the correct text information is the alphabet, numbers, symbols, etc. that may be included in the email address.

[0018] The table generation unit 22 may, for example, store only the correct text information of numbers, that is, only one type of correct text information, or may store a plurality of types of correct text information.

[0019] When the table generation unit 22 stores a plurality of types of correct text information, for example, it may receive the types of text information of the manuscript such as characters, numbers, or a combination of characters, numbers, and symbols, and select the preset correct text information according to the received types of text information.

[0020] Further, instead of accepting the type of text information, it may be possible to accept the type of manuscript such as addition calculation printing, subtraction printing, and English test printing, and select the preset correct text information according to the type of the manuscript. In this case, it is assumed that a correct text information reference table associating the type of manuscript with the correct text information is preset.

[0021] As described above, by accepting the type of text information or the type of manuscript and enabling the selection of the corresponding correct text information, the variations of the manuscript can be expanded and the convenience can be improved.

[0022] Next, the table generated by the table generation unit 22 will be described. FIG. 2 is a diagram showing an example of the table generated by the table generation unit 22. The table shown in FIG. 2 is a table when the correct text information is 0 to 9. In the table shown in FIG. 2, the numbers 0 to 9, which are the correct text information, are arranged as indexes in the top row. And below each index of 0 to 9, the text information recognized by the text information recognition unit 21 is arranged in a column. Specifically, the table generation unit 22 arranges each number recognized by the text information recognition unit 21 below the index to which the number belongs based on the recognition information. FIG. 2 is a diagram showing an example of the table in the case of no misrecognition.

[0023] Here, as mentioned above, when the text information recognition unit 21 recognizes handwritten characters, it may misrecognize them depending on how they are written. That is, even if the original data is "2", the text information recognition unit 21 may misrecognize it as "1". In such a case of misrecognition, the table generation unit 22 accepts "1" as the recognition information for the number "2" in the original data, so the number "2" will be placed under the index of "1". Figure 3 is a diagram showing an example of a table in the case of the misrecognition described above. By generating a table in this way, it is possible to immediately understand that the number "2" placed under the index of "1" is a misrecognized number.

[0024] Note that the tables in Figures 2 and 3 were generated using 0-9 as the index because the correct text information was numerical. However, if the correct text information is in English, a table will be generated using the alphabet (a-z) as the index.

[0025] The display control unit 23 displays the table generated by the table generation unit 22 on the display device 30. The display control unit 23 also displays the table generated by the table generation unit 22 and the original data simultaneously on the display device 30. Figure 4 is a diagram showing an example of a screen displaying the table generated by the table generation unit 22 and the original data simultaneously. In Figure 4, an example is shown where the original data is data obtained by reading a 100-square calculation printout. By displaying the table in this way, misrecognition can be immediately identified by visually inspecting the table. Furthermore, by displaying the table and the original data simultaneously, they can be compared, and the locations of misrecognition in the original data can be identified. Note that when recognizing the text information contained in the original data, it is assumed that the system is pre-set to recognize only the numbers written in the answer fields.

[0026] Furthermore, as shown in Figure 4, if the source data is data obtained by scanning a 100-square calculation worksheet, the scoring results may also be displayed. Specifically, the recognition information of the numbers in the answer fields included in the source data may be compared with pre-set answer numbers, and correct answers may be displayed by circling the numbers.

[0027] Furthermore, if specific text information is specified in either the original data or the table shown in Figure 4, the display control unit 23 will display the portion of the specific text information in the other document where that specific text information is not specified, in a way that allows for identification.

[0028] Specifically, for example, in the table shown in Figure 4, if a misrecognized "2" is included in the column with index "1", the user uses the input device 40 to specify the misrecognized "2". At this time, the display control unit 23 displays the "2" specified by the user and the "2" in the original data corresponding to that specified "2" in a distinguishable manner. As for how to display "2" in a distinguishable manner, as shown in Figure 4, a dotted rectangular frame surrounding "2" may be displayed, the color of "2" may be displayed differently from the color of other numbers, or "2" may be displayed blinking, and any display method that makes "2" distinguishable is acceptable.

[0029] In this way, by making the "2" in the original data corresponding to the "2" in the table specified by the user identifiable, the correspondence between misrecognized numbers can be immediately grasped.

[0030] Furthermore, while the above explanation described an example where, when specific text information ("2") is specified in the table, the corresponding text information ("2") is displayed in the original data in an identifiable manner, the opposite may also be done: when specific text information ("2") is specified in the original data, the corresponding text information ("2") is displayed in the table in an identifiable manner.

[0031] Furthermore, the display control unit 23 associates specific text information in the table with specific text information in the original document data based on the recognition information from the text information recognition unit 21.

[0032] Furthermore, in the original data shown in Figure 4, when the misrecognized "2" is displayed in an identifiable state, the display control unit 23 displays a correction screen to correct the identification information of the misrecognized "2" when that misrecognized "2" is selected by clicking or tapping.

[0033] Figure 5 shows an example of the correction screen. In the correction screen shown in Figure 5, the recognition information of the text information before correction ("1") is displayed, and the recognition information of the text information after correction ("2") is displayed for input. As shown in Figure 5, a numeric keypad with numbers from 0 to 9 is displayed on the correction screen, and when the user selects a number on the numeric keypad, that selected number is displayed in the display field for the corrected number.

[0034] Then, if the "Finish" button on the correction screen shown in Figure 5 is selected, the recognition information of the corrected number displayed in the display field for the corrected number is determined as the corrected number recognition information, and the misrecognized number recognition information in the original data is replaced with the corrected number recognition information. By displaying the correction screen in this way, the misrecognized number recognition information can be corrected immediately.

[0035] Furthermore, the method for correcting misrecognized numbers is not limited to correction using the correction screen described above. For example, in the table shown in Figure 4, the misrecognized number "2" placed in the column with the index "1" can be dragged and dropped into the column with the index "2" to correct the recognition information of the misrecognized number "2" from "1" to "2".

[0036] As mentioned above, after the numerical recognition information in the manuscript data is corrected, the numerical recognition information is compared with the correct numerical answers in the 100-square calculation, and the 100-square calculation is automatically scored.

[0037] The text information recognition device 20 is composed of a computer and includes a CPU (Central Processing Unit), semiconductor memory such as ROM (Read Only Memory) and RAM (Random Access Memory), storage such as a hard disk, and a communication interface. A text information recognition program is installed in the semiconductor memory or storage of the text information recognition device 20, and when this text information recognition program is executed by the CPU, the text information recognition unit 21, table generation unit 22, and display control unit 23 of the text information recognition device 20 function.

[0038] In this embodiment, the functions of the text information recognition unit 21, the table generation unit 22, and the display control unit 23 are implemented by a text information recognition program. However, the embodiment is not limited to this, and some or all of the functions or controls may be implemented by hardware such as an ASIC (Application Specific Integrated Circuit), FPGA (Field-Programmable Gate Array), or other electrical circuits.

[0039] The display device 30 has a display such as a liquid crystal display and displays the original data and tables generated by the table generation unit 22 as described above.

[0040] The input device 40 includes, for example, a keyboard or mouse, and accepts various setting inputs from the user.

[0041] Furthermore, the text information recognition device 20, display device 30, and input device 40 may be implemented using a tablet terminal. In addition, the text information recognition system 1, including the document reading device 10, may be implemented using a tablet terminal with a camera function, and the document containing text information may be read using the camera function.

[0042] Next, the processing flow of the text information recognition system 1 of this embodiment will be explained with reference to the flowchart shown in Figure 6.

[0043] First, the document is read by the document reader 10 and document data is acquired (S10). The document data is output from the document reader 10 to the text information recognition device 20, and the text information recognition unit 21 receives the input document data and recognizes the text information contained in the document data (S12).

[0044] Next, the table generation unit 22 generates the above-mentioned table based on the text information recognized by the text information recognition unit 21 (S14).

[0045] Next, the display control unit 23 simultaneously displays the table generated by the table generation unit 22 and the original document data on the display device 30 (S16).

[0046] The user then visually checks the table displayed on the display device 30 to confirm whether any misrecognition has occurred in the text information recognition unit 21 (S18). Specifically, the user checks whether any misrecognition has occurred by checking whether or not different text information is included under each index.

[0047] Then, if it is confirmed in S18 that a misrecognition has occurred (S18, YES), the user can display a correction screen by specifying the misrecognized text information and correct it to the correct recognition information on that screen (S20). The corrected recognition information is reflected in the text information of the original data (S22). The process from S18 to S22 is repeated until there is no more misrecognized text information.

[0048] Then, if it is confirmed that there is no misrecognized text information in S18 (S18, NO), the process is terminated.

[0049] It should be noted that the present invention is not limited to the embodiments described above, and the components can be modified and implemented in practice without departing from the spirit of the invention. Furthermore, various inventions can be formed by appropriately combining the multiple components disclosed in the embodiments described above. For example, all the components shown in the embodiments may be combined as appropriate. It goes without saying that various modifications and applications are possible without departing from the spirit of the invention.

[0050] The present invention relates to a text information recognition device, and further disclosures are made as follows. (Note)

[0051] The text information recognition device of the present invention may include a display control unit that displays the above-mentioned table.

[0052] Furthermore, in the text information recognition device of the present invention, the display control unit can simultaneously display the scanned document data and the table.

[0053] Furthermore, in the text information recognition device of the present invention, if specific text information is specified in either the read document data or the table, the display control unit can display the portion of the specific text information in the other document where the specific text information is not specified in the other document.

[0054] Furthermore, in the text information recognition device of the present invention, the display control unit can display a correction screen for specific text information when specific text information is specified.

[0055] Furthermore, in the text information recognition device of the present invention, the table generation unit can receive the type of text information or the type of document, and generate the above-mentioned table based on the correct text information corresponding to the type of text information or document received.

[0056] The text information recognition method of the present invention receives document data obtained by reading a document containing text information, recognizes the text information contained in the document data, and generates a table that associates the recognized text information with pre-set correct text information that may be contained in the document.

[0057] The text information recognition program of the present invention causes a computer to perform the following steps: receiving document data obtained by reading a document containing text information; recognizing the text information contained in the document data; and generating a table that associates the recognized text information with pre-set correct text information that may be contained in the document. [Explanation of Symbols]

[0058] 1. Text Information Recognition System 10. Document scanning device 20 Text Information Recognition Device 21 Text Information Recognition Unit 22 Table generation section 23 Display Control Unit 30 Display device 40 Input devices

Claims

1. A text information recognition unit that receives document data obtained by reading a document containing handwritten text information selected from a plurality of pre-set correct text information, and recognizes the handwritten text information contained in the document data, A table generation unit generates a table by assigning the read data of the text information to each of the multiple correct text information based on the recognition results of the handwritten text information, A text information recognition device comprising a display control unit that displays the aforementioned table.

2. The text information recognition device according to Claim 1, wherein the table generation unit generates the table by arranging the read data of the text information side by side for each of the plurality of correct text information to form a column for each of the correct text information.

3. The text information recognition device according to claim 2, wherein the display control unit simultaneously displays the scanned document data and the table.

4. The text information recognition device according to claim 3, wherein if the display control unit specifies specific text information in either the scanned document data or the table, it displays the portion of the specific text information in the other where the specific text information is not specified in an identifiable manner.

5. The text information recognition device according to claim 4, wherein the display control unit displays a correction screen for the specified text information when the specified text information is specified.

6. The text information recognition device according to any one of claims 1 to 5, wherein the table generation unit receives the type of text information or the type of manuscript, and generates the table based on the correct text information corresponding to the received type of text information or the type of manuscript.

7. A document data is received which includes handwritten text information selected from a set of multiple correct text information in advance, The system recognizes the handwritten text information contained in the manuscript data. Based on the recognition results of the handwritten text information, the reading data of the text information is assigned to each of the multiple correct text information items to generate a table. A text information recognition method for displaying the generated table.

8. A step of receiving document data obtained by reading a document containing handwritten text information selected from a plurality of pre-set correct text information, The steps include: recognizing handwritten text information contained in the manuscript data; Based on the recognition results of the handwritten text information, the step of assigning the read data of the text information to each of the multiple correct text information items and generating a table, A text information recognition program that causes a computer to perform the steps of displaying the generated table.

Images