JP2016212812A - Information processor and information processing program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide an information processor for reducing man-hours in the case of manually inputting character recognition for a character recognition object without increasing an error ratio of a data input in comparison with a case of manual inputting.SOLUTION: Classification means of the information processor classifies a character recognition object to any of three kinds, extraction means extracts a character recognition result of the character recognition object when the classification means classifies the character recognition object into a first kind, first control means extracts the character recognition result of the character recognition object to perform control so as to manually input the character recognition object when the classification means classifies it into a second kind, and second control means performs control so as to cause a plurality of persons to manually input the character recognition object when the classification means classifies it into a third kind.SELECTED DRAWING: Figure 1

Description

  The present invention relates to an information processing apparatus and an information processing program.

  In Patent Document 1, when automatically recognizing image data of a credit card enrollment application form, an object is to enable reduction of input processing time and cost by an operator that complements this character recognition. The department captures a large number of completed credit card application forms into a computer using a scanner, and the character recognition part recognizes the characters in the credit card application form according to a predetermined reading format and responds to the credit card application form. Recognize data is entered in the specified field of the data file, and the data input unit inputs either the keyboard or voice in the incomplete entry field where no character is entered due to an error in character recognition. It is disclosed that complementary data is input based on the above.

  Patent document 2 relates to a key input editing method and editing apparatus for correcting and editing misrecognized characters, and aims to improve the efficiency of correcting misrecognized characters. Stored in an image data file, read the image data in line / field / column units from the image data file according to the definition information in the form definition information file, etc., and recognized the characters by the character recognition unit. The edit processing unit displays the recognized character of the same character type or the same character code and its image data in correspondence with each other on the display unit in line / field / column units, and corrects misrecognized characters from the keyboard. Then, it is disclosed to correct the recognized characters stored in the database.

  Patent Document 3 aims to detect an input error of character data, optically reads character information by an image input unit, inputs the character information from a key input unit, and reads the character information read by the image input unit. The data recognized by the recognition unit and the key-input data are compared by the character verification unit. When the comparison result indicates a match, the data is output as correct data. Check that the correction is necessary on the display unit, the data input by the key is output as correct data after this notification, and the error is detected by comparing the image input result with the key input result. Even if logical judgment by digit etc. is impossible, the error can be detected, and the key puncher can process faster than the key input twice. Can be reduced is disclosed.

JP 2005-056099 A JP-A-11-007492 Japanese Patent Laid-Open No. 06-274679

  The present invention relates to an information processing apparatus capable of reducing man-hours when inputting manually and without increasing the error rate of data input, compared with a case where character recognition is performed on a character recognition target and inputting manually. The purpose is to provide an information processing program.

The gist of the present invention for achieving the object lies in the inventions of the following items.
According to the first aspect of the present invention, there is provided a classifying unit for classifying the character recognition target into one of three types, and an extraction for extracting a character recognition result of the character recognition target when the character recognition target is classified into the first type by the classifying unit. And a first control means for extracting the character recognition result of the character recognition target and controlling the character recognition target to be manually input when the character is classified into the second type by the classification means; An information processing apparatus comprising: a second control unit configured to control a character recognition target to be manually input by a plurality of people when the classification unit classifies the third type.

  According to a second aspect of the present invention, a character recognition result extracted by the control of the first control means and an input result input by hand, or a plurality of input results input by the control of the second control means. The information processing apparatus according to claim 1, further comprising integration means for integrating.

  The invention according to claim 3 is characterized in that the integration means controls to manually select or input when a character recognition result and an input result are different, or when a plurality of input results are different. The information processing apparatus according to Item 2.

  The invention according to claim 4 is characterized in that the classification means performs classification by comparing the recognition accuracy of the character recognition result of the character recognition target with a plurality of predetermined thresholds. It is an information processing apparatus as described in any one.

  According to a fifth aspect of the present invention, when the computer classifies the character recognition target into one of the three types, and the classification unit classifies the character recognition target as the first type, the character recognition result of the character recognition target is Extracting means for extracting and first control for extracting the character recognition result of the character recognition target and controlling the character recognition target to be manually input when the character is classified into the second type by the classification means And an information processing program for functioning as second control means for controlling the character recognition target to be manually input by a plurality of persons when classified by the classification means into a third type.

  According to the information processing apparatus of the first aspect, compared with the case of character recognition and manual input with respect to the character recognition target, man-hours for manual input are reduced without increasing the data input error rate. be able to.

  According to the information processing apparatus of the second aspect, the two result data can be integrated.

  According to the information processing apparatus of the third aspect, when the character recognition result and the input result are different, or when a plurality of input results are different, it is possible to control to make manual selection or input.

  According to the information processing apparatus of the fourth aspect, classification can be performed by comparing the recognition accuracy of the character recognition result of the character recognition target with a plurality of predetermined threshold values.

  According to the information processing program of claim 5, compared with the case of character recognition and manual input with respect to the character recognition target, the man-hour when inputting manually is reduced without increasing the error rate of data input. be able to.

It is a conceptual module block diagram about the structural example of 1st Embodiment. It is explanatory drawing which shows the system configuration example using this Embodiment. It is a flowchart which shows the process example by 1st Embodiment. It is explanatory drawing which shows the process example by this Embodiment. It is explanatory drawing which shows the process example by this Embodiment. It is a conceptual module block diagram about the structural example of 2nd Embodiment. It is a flowchart which shows the process example by 2nd Embodiment. It is explanatory drawing which shows the process example in which this Embodiment is utilized. It is explanatory drawing which shows an example of a matching process. It is explanatory drawing which shows an example of a matching process. It is explanatory drawing which shows an example of a matching process. It is explanatory drawing which shows an example of a matching process. It is a block diagram which shows the hardware structural example of the computer which implement | achieves this Embodiment.

First, before describing the present embodiment, the premise or an information processing apparatus using the present embodiment will be described with reference to FIGS. This description is intended to facilitate understanding of the present embodiment.
FIG. 8 is an explanatory diagram showing a processing example in which this exemplary embodiment is used. There is a task of inputting data in a form in which various numbers such as “address”, “name”, and “product number” are entered. For example, as shown in the example of FIG. 8, there is a form image 810 having a name field 812, an address field 814, and a product number field 816, and in the name field 812, address field 814, and product number field 816, Characters are written by hand.
Generally, looking at the form image 810 and using the information processing apparatus 800, an operator (user) inputs manually (key input), so data input costs are a problem.
In order to eliminate erroneous input, as shown in the example of FIG. 9, in the case of conventional manual input, a plurality of people (Mr. A, Mr. B) see the same form image 810 and manually input device (Mr. A) 920A and manual input device (Mr. B) 920B perform key input (double input) in parallel. In the matching processing module 940, the results of input performed by each of the manual input device (Mr. A) 920A and the manual input device (Mr. B) 920B are matched. If the two results are the same (“when the result is the same” 946), the data of the input result is adopted as it is (“determines that the key input result matches” 948). When the two results are different (“if the results are different” 942), it is determined that the key input result is incorrect (“the key input result is determined to be incorrect” 944). If it is wrong, another process (such as re-input) is performed.
In Patent Document 3 and the like listed as the background art, as shown in the example of FIG. The same form image 810 that the user A sees and inputs the key is input to the character recognizer 1030. The character recognizer 1030 recognizes characters in the input form image 810 (specifically, characters written in the name column 812, the address column 814, and the product number column 816), and outputs the recognition result. . The matching processing module 1040 matches the input result of Mr. A by the manual input device (Mr. A) 1020 and the recognition result of the character recognizer 1030, and if the two results are the same ("if the result is the same" 1046 ), The data of the input result (recognition result) is employed as it is (“determine that the key input result and the recognition result match” 1048). If the two results are different (“if the results are different” 1042), it is determined that the key input result or the character recognition result is incorrect (“determines that the key input result or the recognition result is incorrect” 1044). In this case, another process (such as re-input) is performed.
As described above, it is a conventional technique to perform double input of the character recognizer 1030 and manual input (manual input device (Mr. A) 1020).

In the prior art shown in the example of FIG. 10, when the recognition rate of the character recognizer 1030 is poor, the quality of the final data input is deteriorated by being dragged by the bad recognition rate of the character recognizer 1030. There is a case.
This will be described with reference to examples shown in FIGS. The example shown in FIG. 11 corresponds to the example shown in FIG. 9, and the example shown in FIG. 12 corresponds to the example shown in FIG.
11 and 12, when the two results are different (“if the results are different” 1142 and 1242), the integration processing modules 1160 and 1260 indicate “Mr. C” (manual input device (Mr. C) 1150 and 1250). ) Is used.
Here, let r be the human error rate.
As in the example shown in FIG. 11, when two people are input, the probability of error is 1− (1−r) ² . Since the probability E that the final data is wrong is also a case where Mr. C is wrong, E = r [1- (1-r) ² ]. If the human error rate is r = 0.01, the data error rate in the case of FIG. 11 is 1.99 × 10 ⁻⁴ .
Next, let R be the error rate of the character recognizer 1030. The final error rate is E = r [1- (1-r) (1-R)].
If R = 0.01, the results shown in FIGS. 11 and 12 are the same. However, when the error rate of the character recognizer 1030 is large, for example, when R = 0.1, the final error rate is E = 1.09 × 10 ⁻³ , which increases the error rate by about 1 digit. End up.

When inputting data, if a character recognizer is used, there is a problem that the data error rate increases.
In order to avoid this problem, it is sufficient to manually perform double input without using a character recognizer.
However, in this case, man-hours for two people are always required.

Hereinafter, examples of various preferred embodiments for realizing the present invention will be described with reference to the drawings.
FIG. 1 is a conceptual module configuration diagram of a configuration example according to the first embodiment.
The module generally refers to components such as software (computer program) and hardware that can be logically separated. Therefore, the module in the present embodiment indicates not only a module in a computer program but also a module in a hardware configuration. Therefore, the present embodiment is a computer program for causing these modules to function (a program for causing a computer to execute each procedure, a program for causing a computer to function as each means, and a function for each computer. This also serves as an explanation of the program and system and method for realizing the above. However, for the sake of explanation, the words “store”, “store”, and equivalents thereof are used. However, when the embodiment is a computer program, these words are stored in a storage device or stored in memory. This means that control is performed so as to be stored in the apparatus. Modules may correspond to functions one-to-one, but in mounting, one module may be configured by one program, or a plurality of modules may be configured by one program, and conversely, one module May be composed of a plurality of programs. The plurality of modules may be executed by one computer, or one module may be executed by a plurality of computers in a distributed or parallel environment. Note that one module may include other modules. Hereinafter, “connection” is used not only for physical connection but also for logical connection (data exchange, instruction, reference relationship between data, etc.). “Predetermined” means that the process is determined before the target process, and not only before the process according to this embodiment starts but also after the process according to this embodiment starts. In addition, if it is before the target processing, it is used in accordance with the situation / state at that time or with the intention to be decided according to the situation / state up to that point. When there are a plurality of “predetermined values”, they may be different values, or two or more values (of course, including all values) may be the same. In addition, the description having the meaning of “do B when it is A” is used in the meaning of “determine whether or not it is A and do B when it is judged as A”. However, the case where it is not necessary to determine whether or not A is excluded.
In addition, the system or device is configured by connecting a plurality of computers, hardware, devices, and the like by communication means such as a network (including one-to-one correspondence communication connection), etc., and one computer, hardware, device. The case where it implement | achieves by etc. is also included. “Apparatus” and “system” are used as synonymous terms. Of course, the “system” does not include a social “mechanism” (social system) that is an artificial arrangement.
In addition, when performing a plurality of processes in each module or in each module, the target information is read from the storage device for each process, and the processing result is written to the storage device after performing the processing. is there. Therefore, description of reading from the storage device before processing and writing to the storage device after processing may be omitted. Here, the storage device may include a hard disk, a RAM (Random Access Memory), an external storage medium, a storage device via a communication line, a register in a CPU (Central Processing Unit), and the like.

Hereinafter, a character image will be mainly exemplified and described as a character recognition target. However, it is not necessary to limit to a character image. For example, it may be an online character configured by stroke information. Moreover, not only a handwritten character but a printed character etc. may be sufficient.
The information processing apparatus 100 according to the present embodiment uses the text data indicating the characters included in the character image 108 as output data 152. As shown in the example of FIG. , A character string classification module 120, a recognition result selection module 130, a manual input module 140, and a result integration module 150.

The character recognition module 110 is connected to the character string classification module 120 and the recognition result selection module 130, and includes a character image 108 (specifically, a character image (A mountain B male) 108A, a character image (via C River D) 108B. And the like, and the recognition accuracy 112 is passed to the character string classification module 120 and the recognition result 116 is passed to the recognition result selection module 130. The character recognition module 110 recognizes characters in the character image 108. An existing character recognition technique that outputs the recognition result 116 that is text data and the recognition accuracy 112 of the recognition result 116 may be used. Here, it is assumed that the higher the value of the recognition accuracy 112 is, the higher the probability of being the recognition result 116 is.
The character string classification module 120 is connected to the character recognition module 110, the recognition result selection module 130, the manual input module 140, and the result integration module 150, receives the threshold value 118, receives the recognition accuracy 112 from the character recognition module 110, and recognizes it. The classification result 122 is passed to the result selection module 130, the manual input module 140, and the result integration module 150. The character string classification module 120 classifies the character image 108 that is a character recognition target into one of three types.
Further, the character string classification module 120 may perform the classification by comparing the recognition accuracy 112 of the character recognition result of the character image 108 to be recognized with a plurality of predetermined threshold values 118. Specifically, Th1 and Th2 (Th1 is a threshold higher than Th2) are used as the threshold value 118, and any one of the first type, the second type, and the third type is output as the classification result 122. May be. When the recognition accuracy 112 is higher than Th1 (that is, when the recognition result 116 by the character recognition module 110 may be adopted), it is classified into the first type, and the recognition accuracy 112 is equal to or lower than Th1 and higher than Th2. In the case (that is, not “when the recognition result 116 by the character recognition module 110 may be adopted” but not “when the recognition result 116 by the character recognition module 110 should not be adopted”), the second When the recognition accuracy 112 is equal to or less than Th2 (that is, when the recognition result 116 by the character recognition module 110 should not be adopted), it is classified into the third type.

The recognition result selection module 130 is connected to the character recognition module 110, the character string classification module 120, and the result integration module 150. The recognition result selection module 130 receives the recognition result 116 from the character recognition module 110 and the classification result 122 from the character string classification module 120. The recognition result 132 is passed to the result integration module 150. The recognition result selection module 130 extracts the recognition result 116 by the character recognition module 110 for the character image 108 to be recognized when it is classified into the first type by the character string classification module 120. The recognition result selection module 130 extracts the recognition result 116 by the character recognition module 110 for the character image 108 to be recognized when the character string classification module 120 classifies the second type. That is, when the classification result 122 is the first type or the second type, the recognition result 116 is passed to the result integration module 150 as the recognition result 132.
The manual input module 140 is connected to the character string classification module 120 and the result integration module 150, receives the character image 108, receives the classification result 122 from the character string classification module 120, and receives the manual input result 142 in the result integration module 150. hand over. The manual input module 140 performs control so that the character image 108 that is a character recognition target is manually input when it is classified into the second type by the character string classification module 120. Further, the manual input module 140 performs control so that the character image 108 to be recognized is input manually by a plurality of people when the character string classification module 120 classifies the third type. That is, when the classification result 122 is the second type or the third type, the manual input result 142 (one input result for the second type or a plurality of input results for the third type) is used. The result is passed to the result integration module 150. Hereinafter, the case of two persons is shown as an example of a plurality of persons, but three or more persons may be used.

The result integration module 150 is connected to the character string classification module 120, the recognition result selection module 130, and the manual input module 140. The result integration module 150 receives the classification result 122 from the character string classification module 120, the recognition result 132 from the recognition result selection module 130, A manual input result 142 is received from the manual input module 140 and output data 152 is output. The result integration module 150 outputs the character recognition result extracted by the recognition result selection module 130 as output data 152 when it is classified into the second type by the character string classification module 120. The result integration module 150 integrates the character recognition result extracted by the recognition result selection module 130 and the input result input manually by the human input module 140 when classified by the character string classification module 120 into the second type. To do. Alternatively, the result integration module 150 integrates a plurality of input results input under the control of the manual input module 140 when the result is classified into the third type by the character string classification module 120.
When the recognition result 132 and the manual input result 142 are different, or when the plurality of manual input results 142 are different, the result integration module 150 selects manually (selection of either the recognition result 132 or the manual input result 142, a plurality of It is also possible to perform control such that one of the manual input results 142 is selected) or manual input is performed. The user here may be a user who has made an input with the manual input module 140, but preferably a user other than the user who has made an input with the manual input module 140.

An example of operation will be described using the example shown in FIG.
A form image is input as a character image 108. For example, an image of the name column in the form is input. Although it is not necessary to cut out in particular, this example shows an example in which the name field is cut out and input as a character image (A mountain B male) 108A, a character image (via C River D) 108B, etc. ing.
The character image 108 is recognized by the character recognition module 110, and the recognition accuracy 112 and the recognition result 116 are output.
The recognition accuracy 112 is input to the character string classification module 120, and is classified into three types of character strings (1) to (3) as described later in the example of FIG. The character string classification module 120 uses two threshold values 118.
The recognition result selection module 130 selects the recognition result 116 in the case of (1) or (2).
In the case of (2), the manual input module 140 performs data display and data reception so as to input manual data for one person. In the case of (3), data display and data reception are performed so that manual data input for two persons is performed.
In the result integration module 150, the recognition result 132 and the result of the manual input module 140 (manual input result 142) are integrated according to the classification result 122 to perform final processing. As an example of the final processing, as will be described later with reference to FIGS. 11 and 12, two results are matched, and if the results are different, processing such as manual input is performed. The processing result by the result integration module 150 becomes output data 152. The main output data 152 is data input content (text data corresponding to the character image 108).
A character string refers to a sequence of one or more characters. Therefore, one character is also called a character string.

The threshold value may not be two of Th1 and Th2. Either one is acceptable.
If only Th1,
(1) Data input is performed only by the character recognition module 110.
(2) The character recognition module 110 and a person perform double input.
Can be selected. In this case, the recognition result selection module 130 is not necessary.
If only Th2,
(2) The character recognition module 110 and a person perform double input.
(3) A person double-inputs.
Can be selected.
Alternatively, by setting Th1 and Th2 to the same value,
(1) Data input is performed only by the character recognition module 110.
(3) A person double-inputs.
Can be selected.

FIG. 2 is an explanatory diagram showing a system configuration example using the present embodiment.
The information processing apparatus 100, the image reading apparatus 210, the input terminal 220A, the input terminal 220B, and the input terminal 220C are connected via a communication line 290, respectively. The communication line 290 may be wireless, wired, or a combination thereof, and may be, for example, the Internet or an intranet as a communication infrastructure.
The input terminal 220 is operated by the user 222. For example, the image reading apparatus 210 reads a form image 810 and transmits it to the information processing apparatus 100. Under the control of the information processing apparatus 100, in the case of the second type, the character image read by the image reading apparatus 210 is transmitted to the input terminal 220A, and the input terminal 220A sends the input result by the user 222A to the information processing apparatus 100. Send. Also, in the case of the third type, the character image read by the image reading device 210 is transmitted to the input terminal 220A and the input terminal 220B by the control by the information processing apparatus 100, and the input result by the user 222A is input to the input terminal. 220A transmits to the information processing apparatus 100, and the input terminal 220B transmits the input result by the user 222B to the information processing apparatus 100. If the recognition result 116 by the character recognition module 110 is different from the input result by the user 222A, or if the input result by the user 222A and the input result by the user 222B are different, the image is read by the input terminal 220C by the image reading device 210. The input character 220C is transmitted to the information processing apparatus 100 by the input terminal 220C.
The information processing apparatus 100 outputs the recognition result 116 by the character recognition module 110 in the case of the first type, and the input result by the user 222A and the recognition result 116 by the character recognition module 110 in the case of the second type. In the case of the third type, a result obtained by integrating the input result by the user 222A and the input result by the user 222B is output. The integration result here is a value (input result or recognition result) when both are the same, and an input result by the user 222C when they are different.

FIG. 3 is a flowchart illustrating a processing example according to the first exemplary embodiment.
In step S302, the character recognition module 110 receives the character image 108.
In step S304, the character recognition module 110 recognizes the character image 108.
In step S306, the character string classification module 120 determines the recognition accuracy (K) 112. If “K> Th1”, the process proceeds to step S308. If “Th2 <K ≦ Th1”, the process proceeds to step S312. If “K ≦ Th2”, the process proceeds to step S318.
In step S308, the recognition result selection module 130 selects the recognition result 116.
In step S <b> 310, the result integration module 150 sets the recognition result 132 as the output data 152.

In step S312, the recognition result selection module 130 selects the recognition result 116.
In step S314, the manual input module 140 prompts input of human data for one person.
In step S316, the manual input module 140 receives a manual input result.
In step S318, the manual input module 140 prompts input of manual data for two people.
In step S320, the manual input module 140 receives a manual input result for two people.
In step S322, a matching process is performed. If “same”, the process proceeds to step S324, and if “different”, the process proceeds to step S326.
In step S324, the result integration module 150 sets the manual input result 142 as the output data 152.
In step S326, the result integration module 150 performs integration processing.

FIG. 4 is an explanatory diagram showing a processing example according to the present embodiment. 4 and 5 show an outline of a processing example according to this embodiment.
The information processing apparatus 100 includes a character recognizer 410 (corresponding to the character recognition module 110), a character recognizer 420 (corresponding to the character recognition module 110), a manual input device (Mr. A) 430, a manual input device (Mr. B) 440, It is connected to a manual input device (Mr. C) 450 and receives a character image 108.
The character image 108 for which data is input is classified into the following three types by the information processing apparatus 100.
(1) Data is input only with a character recognizer.
(2) Double input with character recognizer and person.
(3) A person double-inputs.
In the prior art, since only (2) was mentioned above, the problem was that the overall error rate was high. In the present embodiment, in the case of the character image 108 with a poor recognition rate of the character recognizer 420, the above (3) is used to prevent the error rate from deteriorating.
Furthermore, in the case of (3), since two people are required, the total man-hour increases. In order to avoid this, in the case of an input image with a high character recognition rate, manual input is not performed, and only the character recognizer 410 performs processing.
As described above, by classifying into three types of processing, data input accuracy and man-hour reduction can be realized at the same time.
In the example shown in FIG. 4, the operators of the manual input device (Mr. A) 430, the manual input device (Mr. B) 440, and the manual input device (Mr. C) 450 are different, but the manual input device (Mr. A) is different. The operator of 430 and the manual input device (Mr. B) 440 may be the same person. Or the operator of the manual input device (Mr. A) 430 and the manual input device (Mr. C) 450 may be the same person.

FIG. 5 is an explanatory diagram showing an outline of a processing example according to the present embodiment.
As the character image 108, a character image (A mountain B male) 108A, a character image (C river D via) 108B, a character image (E field F child) 108C, and a character image (G Tani Hiro) 108D are targeted.
The character recognizer 510 receives a character image (A mountain B male) 108A, a character image (via C River D) 108B, and a character image (E field F child) 108C.
The manual input device 520 accepts a character image (via C River D) 108B, a character image (Eda Foko) 108C, and a character image (Hiro Gani) 108D.
The manual input device 530 accepts the character image (G Taniro) 108D.
That is, the information processing apparatus 100 classifies the input character string into the following three types.
(1) When inputting data using only a character recognizer.
(2) When performing double input with a character recognizer and a person.
(3) When double input is performed by multiple people.
In the case of (1) and (2), the character image 108 is sent to the character recognizer 510.
In the case of (2) and (3) above, the character image 108 is sent so that a person can input it. In the case of (3), the character image 108 is sent so that a plurality of people can input data.
For example, if the character image (A mountain B male) 108A corresponds to “(1) Input data only with character recognizer”, the character recognizer 510 performs recognition processing.
For example, when the character image (via C River D) 108B and the character image (E field F child) 108C correspond to “(2) Double input with a character recognizer and a person”, the character recognizer 510 performs recognition processing. And manual input is performed by the manual input device 520.
For example, when the character image (H. Gani) 108D corresponds to “(3) Performing double input by person”, manual input is performed by the manual input device 520 and the manual input device 530.

Classification by the character string classification module 120 is performed using the recognition accuracy 112. The recognition accuracy 112 is a certainty factor for the recognition result 116 that is a processing result by the character recognition module 110.
Here, the recognition accuracy is K. Two threshold values Th1 and Th2 are prepared.
At this time, it is classified as follows.
(1) When K> Th1: Data input is performed only with a character recognizer.
(2) When Th2 <K ≦ Th1: Double input is performed by a character recognizer and a person.
(3) When K ≦ Th2: Double input is performed by a plurality of persons.
In addition, the position of the equal sign in the above formula may be anywhere. For example, the following may be used.
(1) When K ≧ Th1: Data input is performed only with the character recognizer.
(2) When Th2 ≦ K <Th1: Double input is performed by a character recognizer and a person.
(3) When K <Th2: A person performs double input.

As a method for calculating the recognition accuracy, various conventional examples may be used as described below. For example, JP-A-5-040853, JP-A-5-020500, JP-A-5-290169, JP-A-8-101880, JP-A-2011-113125 (using an output value as recognition accuracy), JP-A-2013-0669132 (output value Used as recognition accuracy).
Some of these use recognition accuracy for each character. As a method for converting the recognition accuracy for each character into the recognition accuracy of the character string, an appropriate one may be selected from various methods as described below.
• The maximum character recognition accuracy in the character string is used as the character string recognition accuracy.
• The minimum character recognition accuracy in the character string is used as the character string recognition accuracy.
-The average (mode, median, etc.) character recognition accuracy in the character string is used as the character string recognition accuracy.

FIG. 6 is a conceptual module configuration diagram of a configuration example according to the second embodiment.
The information processing apparatus 600 includes a character string classification module 610, a character recognition selection module 620, a character recognition module 630, a manual input module 640, and a result integration module 650.
The character string classification module 610 is connected to the character recognition selection module 620, the manual input module 640, and the result integration module 650. The character string classification module 610 receives the character image 608 and receives the character recognition selection module 620, the manual input module 640, and the result integration module 650. The classification result 612 is passed. The character string classification module 610 has a function equivalent to that of the character string classification module 120. However, classification may be performed using recognition accuracy based on character recognition, or classification may be performed using information other than recognition accuracy. For example, character recognition processing by the character recognition module 630 may be performed and the recognition accuracy may be used, or character recognition processing other than the character recognition module 630 may be performed and the recognition accuracy may be used. In addition, it is determined whether or not the character image 608 is an image suitable for character recognition (“not suitable”, “not suitable”, “not suitable”, “not suitable”) May be classified into three categories).
The character recognition selection module 620 is connected to the character string classification module 610 and the character recognition module 630, receives the classification result 612 from the character string classification module 610, and passes the character recognition selection result 622 to the character recognition module 630. The character recognition selection module 620 causes the character recognition module 630 to perform processing when classified into the first type and the second type.
The character recognition module 630 is connected to the character recognition selection module 620 and the result integration module 650, receives the character image 608, receives the character recognition selection result 622 from the character recognition selection module 620, and receives the recognition result 632 in the result integration module 650. give. The character recognition module 630 performs processing equivalent to that of the character recognition module 110. However, the character recognition process here does not necessarily output the recognition accuracy.
The manual input module 640 is connected to the character string classification module 610 and the result integration module 650, receives the character image 608, receives the classification result 612 from the character string classification module 610, and receives the manual input result 642 to the result integration module 650. hand over. The manual input module 640 performs the same processing as the manual input module 140.
The result integration module 650 is connected to the character string classification module 610, the character recognition module 630, and the manual input module 640, and the classification result 612 from the character string classification module 610 and the recognition result 632 from the character recognition module 630 are input manually. A manual input result 642 is received from the module 640 and output data 652 is output. The result integration module 650 performs the same processing as the result integration module 150.

An example of operation will be described using the example shown in FIG.
In the first embodiment, the character string classification is performed using the recognition accuracy of the character recognition result, but it is not always necessary to adopt this method. Character string classification may be performed by another method.
A form image is input as a character image 608 (same as in the first embodiment).
The character string classification module 610 performs classification into three types of character strings (1) to (3) shown in the example of FIG. For example, a character recognition unit may be included in the character string classification module 610, and a method of performing classification using recognition accuracy may be used, or a method of not performing character recognition may be used. For example, if character recognition feature extraction is performed and the feature is in a first partial space (a space suitable for recognition as a character) predetermined in the feature space, the feature is classified into the first type. If it is in a predetermined third partial space (a space that cannot be recognized as a character), it is classified into the third type, and in other cases (“space suitable for recognition as a character”, “character May be classified as the second type).
In the case of (1) or (2), the character recognition selection module 620 selects to perform character recognition. The character recognition module 630 performs character recognition processing using the character recognition selection result 622 by the character recognition selection module 620.
In the case of (2), the manual input module 640 performs data display and data reception so as to input manual data for one person. In the case of (3), data display and data reception are performed so that manual data input for two persons is performed.
In the result integration module 650, the recognition result 632 and the manual input result 642 by the manual input module 640 are integrated according to the classification result 612 to perform final processing. As an example of the final processing, as shown in the examples of FIGS. 11 and 12, two results are matched, and if the results are different, processing such as manual input is performed. The processing result by the result integration module 650 becomes output data 652. The main output data 652 is data input content (text data corresponding to the character image 608).

FIG. 7 is a flowchart illustrating a processing example according to the second exemplary embodiment.
In step S702, the character string classification module 610 receives a character image 608.
In step S704, the character string classification module 610 classifies the character image 608.
In step S706, the character string classification module 610 determines the classification result 612. If “pattern A”, the process proceeds to step S708. If “pattern B”, the process proceeds to step S712. If “pattern C”, the process proceeds to step S712. Proceed to S720.
In step S708, the character recognition selection module 620 selects to perform character recognition.
In step S710, the character recognition module 630 performs character recognition.
In step S712, the character recognition selection module 620 selects to perform character recognition.
In step S714, the character recognition module 630 performs character recognition.
In step S716, the manual input module 640 prompts input of manual data for one person.
In step S718, the manual input module 640 receives a manual input result.
In step S720, the manual input module 640 prompts input of manual data for two people.
In step S722, the manual input module 640 receives the manual input results for two people.
In step S724, a matching process is performed. If “same”, the process proceeds to step S726, and if “different”, the process proceeds to step S728.
In step S726, the result integration module 650 sets the manual input result 642 as output data 652.
In step S728, the result integration module 650 performs integration processing.

Next, an effect example according to the present embodiment will be described.
When E = r [1- (1-r) ² ] and the human error rate is r = 0.01, the data error rate in the case of FIG. 11 is 1.99 × 10 ⁻⁴ . This case is taken as an example to show the effect.
For example, when the threshold value Th1 and the threshold value Th2 are set so that the error rate of the character recognizer is 1.99 × 10 ⁻⁴ , even when the character recognizer is used, when two people double-input Data can be input with the same error rate.
Further, if the threshold value Th1 is controlled so that the error rate of the character recognizer is 1.99 × 10 ⁻⁴ , the number of man-hours is not required in the case of recognition accuracy equal to or higher than the threshold value Th1, and therefore the man-hour is reduced. Is also possible.
Further, specific effects will be described by taking the following case as an example.
-Ratio of character string greater than or equal to threshold Th1: 40% (error rate of character recognizer in this case: 0.0001)
-Ratio of character string not less than threshold Th2 and less than Th1: 20% (error rate of character recognizer in this case: 0.01)
-Ratio of character string less than threshold Th2: 40% (error rate of character recognizer in this case: 0.1)
Consider the case.
(1) In the case of the prior art (Japanese Patent Laid-Open No. 6-274679),
Since one person always inputs per character string, the man-hour in this case is 1.
The overall error rate is E = 4.96 × 10 ⁻⁴ (see below).
The average error rate of the character recognizer is
R = 0.0001 × 0.4 + 0.01 × 0.2 + 0.1 × 0.4≈0.04
Therefore, from E = r [1- (1-r) (1-R)], the overall error rate E is E = 4.96 × 10 ⁻⁴ .
(2) In the case of this embodiment,
The number of man-hours per character string is 1 × 0.2 + 2 × 0.4 = 1.0, which is the same as the conventional technology.
In this embodiment,
Error rate when threshold Th1 or more: 0.0001
Error rate when threshold value is Th2 or more and less than Th1: 1.99 × 10 ⁻⁴
Error rate when threshold sleep is less than Th2 (same as human double input error rate): 1.99 × 10 ⁻⁴
Considering the proportion of each, the overall error rate is
0.0001 × 0.4 + 1.99 × 10 −4 × 0.6 = 1.6 × 10 −4
As described above, by using this embodiment, the error rate can be reduced to about ３ even though the man-hour is the same.

  A hardware configuration example of the information processing apparatus according to the present embodiment will be described with reference to FIG. The configuration illustrated in FIG. 13 is configured by, for example, a personal computer (PC) or the like, and illustrates a hardware configuration example including a data reading unit 1317 such as a scanner and a data output unit 1318 such as a printer.

  A CPU (Central Processing Unit) 1301 includes various modules described in the above-described embodiments, that is, a character recognition module 110, a character string classification module 120, a recognition result selection module 130, a manual input module 140, a result integration module 150, It is a control unit that executes processing according to a computer program describing an execution sequence of each module such as a character string classification module 610, a character recognition selection module 620, a character recognition module 630, a manual input module 640, and a result integration module 650. .

  A ROM (Read Only Memory) 1302 stores programs used by the CPU 1301, calculation parameters, and the like. A RAM (Random Access Memory) 1303 stores programs used in the execution of the CPU 1301, parameters that change as appropriate during the execution, and the like. These are connected to each other by a host bus 1304 including a CPU bus or the like.

  The host bus 1304 is connected via a bridge 1305 to an external bus 1306 such as a PCI (Peripheral Component Interconnect / Interface) bus.

  A keyboard 1308 and a pointing device 1309 such as a mouse are input devices operated by an operator. The display 1310 includes a liquid crystal display device or a CRT (Cathode Ray Tube), and displays various types of information as text or image information.

  An HDD (Hard Disk Drive) 1311 includes a hard disk (may be a flash memory or the like), drives the hard disk, and records or reproduces a program executed by the CPU 1301 and information. The hard disk stores a character image 108, recognition accuracy 112, recognition result 116, classification result 122, recognition result 132, manual input result 142, output data 152, and the like. Further, various other data, various computer programs, and the like are stored.

  The drive 1312 reads data or a program recorded on a removable recording medium 1313 such as a magnetic disk, an optical disk, a magneto-optical disk, or a semiconductor memory, and reads the data or program into an interface 1307 and an external bus 1306. , The bridge 1305, and the RAM 1303 connected via the host bus 1304. The removable recording medium 1313 can also be used as a data recording area similar to a hard disk.

  The connection port 1314 is a port for connecting the external connection device 1315 and has a connection unit such as USB and IEEE1394. The connection port 1314 is connected to the CPU 1301 and the like via the interface 1307, the external bus 1306, the bridge 1305, the host bus 1304, and the like. A communication unit 1316 is connected to a communication line and executes data communication processing with the outside. The data reading unit 1317 is, for example, a scanner, and executes document reading processing. The data output unit 1318 is, for example, a printer, and executes document data output processing.

  Note that the hardware configuration of the information processing apparatus illustrated in FIG. 13 illustrates one configuration example, and the present embodiment is not limited to the configuration illustrated in FIG. 13, and the modules described in the present embodiment are executed. Any configuration is possible. For example, some modules may be configured with dedicated hardware (for example, Application Specific Integrated Circuit (ASIC), etc.), and some modules are in an external system and connected via a communication line Alternatively, a plurality of systems shown in FIG. 13 may be connected to each other via communication lines so as to cooperate with each other. In particular, in addition to personal computers, portable information communication devices (including mobile phones, smartphones, mobile devices, wearable computers, etc.), information appliances, robots, copiers, facsimiles, scanners, printers, multifunction devices (scanners, printers, It may be incorporated in an image processing apparatus having any two or more functions such as a copying machine and a facsimile.

Note that the above-described various embodiments may be combined (for example, adding or replacing a module in one embodiment in another embodiment), and processing contents of each module The technique described in the background art may be employed.
Further, in the description of the above-described embodiment, “more than”, “less than”, “higher”, and “lower (less than)” in the comparison with a predetermined value contradict the combination. As long as the above does not occur, “higher”, “lower (less than)”, “more than”, and “less than” may be used.

In the above-described embodiment, an example in which “image” is input and character recognition is performed on the image is shown. However, character recognition is not necessarily limited to “image”. Character recognition may be performed using stroke order information (stroke information) or the like. In that case, when inputting manually, the stroke order data may be imaged so that it can be read by humans.
In the above-described embodiment, human double input is used. However, the number is not double (two) but may be any number as long as there are a plurality.
Moreover, although it was set as the double input of a person (one person) and a character recognizer, it is good also as input of a person (one or more persons) and a character recognizer. For example, two people may be input as a character recognizer.
The person may see the character recognition result and input (or correct the result of the character recognizer).
In the present embodiment, the character recognizer only needs to be seen from the outside as one logical character recognizer. That is, it does not preclude the recognition device from integrating the outputs of a plurality of character recognition devices to produce one recognition result.

The program described above may be provided by being stored in a recording medium, or the program may be provided by communication means. In that case, for example, the above-described program may be regarded as an invention of a “computer-readable recording medium recording the program”.
The “computer-readable recording medium on which a program is recorded” refers to a computer-readable recording medium on which a program is recorded, which is used for program installation, execution, program distribution, and the like.
The recording medium is, for example, a digital versatile disc (DVD), which is a standard established by the DVD Forum, such as “DVD-R, DVD-RW, DVD-RAM,” and DVD + RW. Standard “DVD + R, DVD + RW, etc.”, compact disc (CD), read-only memory (CD-ROM), CD recordable (CD-R), CD rewritable (CD-RW), Blu-ray disc ( Blu-ray (registered trademark) Disc), magneto-optical disk (MO), flexible disk (FD), magnetic tape, hard disk, read-only memory (ROM), electrically erasable and rewritable read-only memory (EEPROM (registered trademark)) )), Flash memory, Random access memory (RAM) SD (Secure Digital) memory card and the like.
The program or a part of the program may be recorded on the recording medium for storage or distribution. Also, by communication, for example, a local area network (LAN), a metropolitan area network (MAN), a wide area network (WAN), a wired network used for the Internet, an intranet, an extranet, or a wireless communication It may be transmitted using a transmission medium such as a network or a combination of these, or may be carried on a carrier wave.
Furthermore, the program may be a part of another program, or may be recorded on a recording medium together with a separate program. Moreover, it may be divided and recorded on a plurality of recording media. Further, it may be recorded in any manner as long as it can be restored, such as compression or encryption.

DESCRIPTION OF SYMBOLS 100 ... Information processing apparatus 108 ... Character image 110 ... Character recognition module 112 ... Recognition accuracy 116 ... Recognition result 118 ... Threshold value 120 ... Character string classification module 122 ... Classification result 130 ... Recognition result selection module 132 ... Recognition result 140 ... Manual input module 142 ... manual input result 150 ... result integration module 152 ... output data 210 ... image reading device 220 ... input terminal 222 ... user 290 ... communication line 600 ... information processing device 608 ... character image 610 ... character string classification module 612 ... classification result 620 ... Character recognition selection module 622 ... Character recognition selection result 630 ... Character recognition module 632 ... Recognition result 640 ... Manual input module 642 ... Manual input result 650 ... Result integration module 652 ... Output data

Claims (5)

A classifying means for classifying the character recognition target into one of three types;
Extracting means for extracting the character recognition result of the character recognition target when the classification means classifies the first type;
A first control unit that extracts a character recognition result of the character recognition target when the classification unit classifies the second type and controls the character recognition target to be manually input;
An information processing apparatus comprising: a second control unit configured to control the character recognition target to be manually input by a plurality of people when the classification unit classifies the third type. An integration unit that integrates the character recognition result extracted by the control of the first control unit and the input result input manually or the plurality of input results input by the control of the second control unit; The information processing apparatus according to claim 1, wherein: 3. The information processing according to claim 2, wherein the integration unit performs control so that selection or input is performed manually when a character recognition result and an input result are different, or when a plurality of input results are different. apparatus. The said classification | category means classify | categorizes by comparing the recognition accuracy of the character recognition result of the said character recognition object with several predetermined threshold values. Information processing device. Computer
A classifying means for classifying the character recognition target into one of three types;
Extracting means for extracting the character recognition result of the character recognition target when the classification means classifies the first type;
A first control unit that extracts a character recognition result of the character recognition target when the classification unit classifies the second type and controls the character recognition target to be manually input;
An information processing program for functioning as second control means for controlling the character recognition target to be manually input by a plurality of persons when classified by the classification means into a third type.

Images