WO2019026147A1 - Information processing device, information processing system, control method, and control program - Google Patents

Abstract

The present invention makes it possible to more accurately detect an item value for a desired item from a form image that indicates an item name and an item value for each of a plurality of items. This information processing device processes a form image indicating an item name and an item value for each of a plurality of items, and comprises: a storage unit which stores a rule dictionary associating a feature quantity of each of a plurality of sample form images with one of a plurality of rule lists that each specify a plurality of item names and a priority assigned to each item name; an acquisition unit which acquires a form image; a feature quantity calculation unit which calculates a feature quantity from the acquired form image; a rule list extraction unit which extracts the rule list associated by the rule dictionary with the calculated feature quantity; a detection unit which sequentially determines whether or not each item name specified in the extracted rule list is included in the form image, in the priority order specified in the rule list, and detects the item value associated with the first item name that was determined to be included in the form image; and an output unit which outputs the detected item value.

Description

INFORMATION PROCESSING APPARATUS, INFORMATION PROCESSING SYSTEM, CONTROL METHOD, AND CONTROL PROGRAM

The present disclosure relates to an information processing apparatus, an information processing system, a control method, and a control program, and more particularly to an information processing apparatus, an information processing system, and a control method for processing a form image in which item names and item values are represented for each of a plurality of items And the control program.

At a life insurance company, a person in charge manually generates a form such as a medical receipt sent from a customer at the time of insurance payment. Since a life insurance company sends a huge number of forms from customers, the work burden on the person in charge is large, and there is a growing demand for assistance with the data conversion of forms.

A media processing apparatus is disclosed which recognizes information on the basis of an image read from a medium in which information is written in an arbitrary format (see Patent Document 1). The medium processing apparatus extracts features of the layout of the image from the data of the read image by referring to an analysis dictionary capable of learning and updating, and identifies the position where information to be recognized exists from the features. , Update the characteristics of the layout by learning.

Also, a frame and a character string of the image information of the input form are detected, and a frame having a high possibility that the character string is an item name is detected as an item name area candidate, and the character string included in the item name area candidate is A form recognition apparatus for determining whether or not an item name is disclosed is disclosed (see Patent Document 2).

Further, a form processing apparatus for performing form processing on image data acquired from a form document has been disclosed (see Patent Document 3). The form processing apparatus classifies each image data into a plurality of image groups based on the type of ruled line data extracted from each image data, and based on form template data assigned to each image group, Perform form processing on image data.

There is also disclosed a form recognition apparatus for recognizing a character string existing in an arbitrary table structure on a form (see Patent Document 4). The form recognition device recognizes a character string existing on a form image obtained by digitizing a form, and extracts a headline wording, which is a predetermined character string, from among the recognized character strings. Then, the form recognition device determines the table structure existing on the form image based on the arrangement of the extracted headline wording on the form image, and uses the determination result to determine the correspondence between the headline wording and the data. Identify

WO 97/05561 JP, 2009-93305, A JP, 2014-89575, A JP, 2010-3155, A

In an information processing apparatus that processes a form image in which item names and item values are represented for each of a plurality of items, it is desired that item values of desired items can be detected more accurately.

The object of the information processing apparatus, the information processing system, the control method and the control program is that the item value of the desired item can be detected more accurately from the form image in which the item name and the item value are represented for each of a plurality of items. And to.

An information processing apparatus according to an aspect of the present invention is an information processing apparatus that processes a form image in which an item name and an item value are represented for each of a plurality of items, and each of the feature amounts for each of a plurality of sample form images A storage unit storing a rule dictionary in which a plurality of item names and each of a plurality of rule lists in which priorities of the item names are defined are associated, an acquisition unit acquiring a form image, and an acquired form image The feature quantity calculation unit that calculates the feature quantity from the rule list, the rule list extraction unit that extracts the rule list associated with the calculated feature quantity in the rule dictionary, and the priority order defined in the extracted rule list A detection unit that sequentially determines whether each item name specified in the rule list is included in the form image, and detects an item value corresponding to the item name determined to be first included; And an output unit that outputs the item value.

An information processing system according to an aspect of the present invention is an information processing system including a generating device, and an information processing device that processes a form image in which an item name and an item value are represented for each of a plurality of items. A storage unit storing a plurality of sample form images, a plurality of item names, and a plurality of rule lists in which priorities of the item names are defined; and characteristics of the sample form images conforming to each of the plurality of rule lists A rule dictionary generation unit that generates a rule dictionary that associates the amount and each rule list; and a transmission unit that transmits the rule dictionary to the information processing apparatus, and the information processing apparatus receives the rule dictionary from the generation apparatus A receiving unit, an acquiring unit for acquiring a form image, a feature amount calculating unit for calculating a feature amount from the acquired form image, and a feature amount calculated in the received rule dictionary. It is sequentially judged whether or not each item name specified in the rule list is included in the form image in the order of the rule list extraction unit that extracts the rule list and the priority order specified in the extracted rule list. It has a detection part which detects the item value corresponding to the item name judged to be first contained, and the output part which outputs the detected item value.

A control method according to an aspect of the present invention is a control method of an information processing apparatus that has a storage unit and an output unit and processes a form image in which an item name and an item value are represented for each of a plurality of items. Storing, in the storage unit, a rule dictionary in which each of the feature amounts for each of a plurality of sample document images is associated with each of a plurality of item names and a plurality of rule lists in which the priorities of the item names are defined; , Form image is acquired, feature amount is calculated from the acquired form image, and a rule dictionary is used to extract a rule list associated with the calculated feature amount, and the priority order specified in the extracted rule list is extracted In order, it is sequentially determined whether each item name specified in the rule list is included in the form image, and the item value corresponding to the item name determined to be included first is detected, and the detected item value Output to the output unit Including the Rukoto.

A control program according to an aspect of the present invention is a control program of an information processing apparatus that has a storage unit and an output unit and processes a form image in which an item name and an item value are represented for each of a plurality of items. Storing, in the storage unit, a rule dictionary in which each of the feature amounts for each of a plurality of sample document images is associated with each of a plurality of item names and a plurality of rule lists in which the priorities of the item names are defined; , Form image is acquired, feature amount is calculated from the acquired form image, and a rule dictionary is used to extract a rule list associated with the calculated feature amount, and the priority order specified in the extracted rule list is extracted In order, it is sequentially determined whether each item name specified in the rule list is included in the form image, and the item value corresponding to the item name determined to be included first is detected, and the detected item value The To perform the outputting the force unit to the information processing apparatus.

According to the present embodiment, the information processing apparatus, the information processing system, the control method, and the control program can accurately calculate the item value of the desired item from the form image in which the item name and the item value are represented for each of the plurality of items. It becomes possible to detect.

The objects and advantages of the invention will be realized and obtained by means of the elements and combinations particularly pointed out in the claims. Both the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the invention as claimed.

FIG. 1 shows an example of a schematic configuration of an information processing system 10 according to an embodiment. FIG. 2 is a diagram showing a schematic configuration of a first storage device 110 and a first CPU 120. It is a figure which shows schematic structure of 2nd memory | storage device 210 and 2nd CPU220. FIG. 6 is a view showing an example of a plurality of sample document images 400, 401, 402. It is a figure which shows an example of several rule list | wrist 500, 501,502. FIG. 6 is a diagram showing an example of a rule dictionary 600. It is a flow chart which shows an example of operation of generation processing of a rule dictionary. It is a figure for demonstrating the feature-value regarding the shape of a ruled line. It is a figure for demonstrating the feature-value regarding a logo. It is a figure for demonstrating the feature-value regarding a character recognition result. It is a figure which shows an example of the rule dictionary produced | generated by a production | generation process. It is a figure which shows an example of the rule dictionary produced | generated by a production | generation process. It is a figure which shows an example of the rule dictionary produced | generated by a production | generation process. It is a flowchart which shows the example of operation | movement of the detection process of an item value. It is a figure which shows an example of a recognition result screen. It is a flowchart which shows the example of operation | movement of an update process. It is a figure which shows schematic structure of the 1st processing circuit 320 of another information processing apparatus. It is a figure which shows schematic structure of the 2nd processing circuit 420 of another production | generation apparatus.

Hereinafter, an information processing apparatus according to an aspect of the present disclosure will be described with reference to the drawings. However, it should be noted that the technical scope of the present disclosure is not limited to those embodiments, but extends to the inventions described in the claims and the equivalents thereof.

FIG. 1 is a diagram showing an example of a schematic configuration of an information processing system 10 according to the embodiment.

The information processing system 10 includes an information processing device 100 and a generation device 200.

The information processing apparatus 100 and the generating apparatus 200 each have a wired or wireless communication function, are connected to the network 300, and mutually communicate with each other via the network 300.

The information processing apparatus 100 is an information processing apparatus such as a personal computer (PC) or a notebook PC, and is used by a worker who is the user to process a form image in which an item name and an item value are represented for each of a plurality of items. Do. The information processing apparatus 100 may be a portable apparatus such as a tablet PC, a multi-function mobile phone (so-called smart phone), or a mobile information terminal. The information processing device 100 includes a first communication device 101, an input device 102, a display device 103, a first storage device 110, and a first central processing unit (CPU) 120. Hereinafter, each part of the information processing apparatus 100 will be described in detail.

The first communication apparatus 101 has a wired communication interface circuit such as Transmission Control Protocol / Internet Protocol (TCP / IP). The first communication device 101 communicates with the network 300 in accordance with a communication method such as Ethernet (registered trademark). The first communication device 101 supplies the data received from the generation device 200 via the network 300 to the first CPU 120, and transmits the data supplied from the first CPU 120 to the generation device 200 via the network 300. The first communication device 101 may be any device as long as it can communicate with an external device. For example, the first communication apparatus 101 may communicate with the generation apparatus 200 via an access point (not shown) according to a wireless local area network (LAN) communication scheme. Alternatively, the first communication apparatus 101 may communicate with the generation apparatus 200 via a base station apparatus (not shown) according to the mobile phone communication scheme.

The input device 102 is an example of an input unit, and includes an input device such as a touch panel type input device, a keyboard, a mouse, and the like, and an interface circuit that acquires a signal from the input device. The input device 102 receives input data input by the user, and outputs a signal corresponding to the received input data to the first CPU 120.

The display device 103 is an example of an output unit, and includes a display configured of liquid crystal, organic EL (Electro-Luminescence), and the like, and an interface circuit that outputs image data or various information to the display. The display device 103 is connected to the first CPU 120 and displays the image data output from the first CPU 120 on the display. The input device 102 and the display device 103 may be integrally configured using a touch panel display.

The first storage device 110 is an example of a storage unit. The first storage device 110 includes a memory device such as a random access memory (RAM) or a read only memory (ROM), a fixed disk device such as a hard disk, or a portable storage device such as a flexible disk or an optical disk. In addition, the first storage device 110 stores a computer program, a database, a table, and the like used for various processes of the information processing apparatus 100. The computer program may be installed from a computer-readable portable recording medium such as, for example, a compact disk read only memory (CD-ROM) or a digital versatile disk read only memory (DVD-ROM). The computer program is installed in the first storage device 110 using a known setup program or the like. Further, in the first storage device 110, a rule dictionary, a plurality of rule lists, and the like are stored as data. Details of the rule dictionary and the rule list will be described later.

The first CPU 120 operates based on a program stored in advance in the first storage device 110. The first CPU 120 may be a general purpose processor. Note that, instead of the first CPU 120, a digital signal processor (DSP), a large scale integration (LSI), or the like may be used. Also, in place of the CPU 160, an application specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or the like may be used.

The first CPU 120 is connected to the first communication device 101, the input device 102, the display device 103, and the first storage device 110, and controls these units. The first CPU 120 performs data transmission / reception control via the first communication device 101, input control of the input device 102, output control of the display device 103, control of the first storage device 110, and the like. Further, the first CPU 120 processes a form image acquired from the first communication device 101 or the like.

The generation device 200 is a server that generates a rule dictionary used by the information processing device 100. The rule dictionary is data used to process a form image, in particular to detect an item value of an item of interest from the form image. The form image is image data representing a form, and the form is a document in which an item name and an item value are represented for each of a plurality of items. In the present embodiment, forms of various layouts are used as forms. A form with various layouts is a form that is individually created by a plurality of different organizations, has different entry positions of item names for each form, and has a plurality of synonyms of item names. The item to be noted is an item that needs to be processed as electronic data, such as calculation or output. The generation device 200 includes a second communication device 201, a second storage device 210, and a second CPU 220. Hereinafter, each part of the generating device 200 will be described in detail.

The second communication device 201 has the same communication interface circuit as the first communication device 101, and communicates with the network 300 in accordance with the same communication method as the first communication device 101. The second communication device 201 supplies the data received from the information processing device 100 via the network 300 to the second CPU 220, and transmits the data supplied from the second CPU 220 to the information processing device 100 via the network 300.

The second storage device 210 is an example of a storage unit. The second storage device 210 includes a memory device similar to the first storage device, a fixed disk device, a storage device, and the like. In addition, the second storage device 210 stores a computer program, a database, a table, and the like used for various processes of the generation device 200. The computer program may be installed from a computer readable portable recording medium such as a CD-ROM, a DVD-ROM, etc. The computer program is installed in the second storage device 210 using a known setup program or the like. Further, in the second storage device 210, a plurality of sample form images, a rule dictionary, a plurality of rule lists, and the like are stored as data. Details of the sample form image, the rule dictionary, and the rule list will be described later.

The second CPU 220 operates based on a program stored in advance in the second storage device 210. The second CPU 220 may be a general purpose processor. Note that, in place of the second CPU 220, a DSP, an LSI, an ASIC, an FPGA, or the like may be used.

The second CPU 220 is connected to the second communication device 201 and the second storage device 210, and controls these units. The second CPU 220 performs data transmission / reception control via the second communication device 201, control of the second storage device 210, and the like. Also, the second CPU 220 generates a rule dictionary.

FIG. 2 is a diagram showing a schematic configuration of the first storage device 110 and the first CPU 120 of the information processing apparatus 100. As shown in FIG.

As shown in FIG. 2, in the first storage device 110, a reception program 111, an acquisition program 112, a feature quantity calculation program 113, a rule list extraction program 114, a detection program 115, an output control program 116, a reception program 117 and an update program Each program such as 118 is stored. Each of these programs is a functional module implemented by software operating on the processor. The first CPU 120 reads each program stored in the first storage device 110 and operates according to each read program. Thus, the first CPU 120 functions as the reception unit 121, the acquisition unit 122, the feature amount calculation unit 123, the rule list extraction unit 124, the detection unit 125, the output control unit 126, the reception unit 127, and the update unit 128.

FIG. 3 is a diagram showing a schematic configuration of the second storage device 210 and the second CPU 220 of the generation device 200.

As shown in FIG. 3, the second storage device 210 stores programs such as a rule list generation program 211, a rule dictionary generation program 212, and a transmission program 213. Each of these programs is a functional module implemented by software operating on the processor. The second CPU 220 reads each program stored in the second storage device 210 and operates according to each read program. Thus, the second CPU 220 functions as a rule list generation unit 221, a rule dictionary generation unit 222, and a transmission unit 223.

FIG. 4 is a view showing an example of a plurality of sample form images 400, 401, 402 stored in the second storage device 210 of the generation device 200. As shown in FIG.

Each sample form image 400, 401, 402 is a form image registered in advance to generate a rule dictionary. As shown in FIG. 4, in each of the sample form images 400, 401, 402, an item name corresponding to each item and an item value corresponding to each item name are displayed in association with each other for each of a plurality of items. There is. Further, in each of the sample document images 400, 401, and 402, the item value of the item of interest is associated and stored as the correct value.

In the example shown in FIG. 4, a medical receipt is used as a form. In this example, each expense concerning medical treatment is classified into a plurality of items, the expense nominal of each expense is described as an item name, and the amount of each expense is described as an item value. Also, in this example, “patient's burden” is specified as the item to be focused on. However, in forms with various layouts such as medical receipts, item names for the same item may differ from form to form. In the example shown in FIG. 4, for the item “patient's burden”, the item name “patient's burden” 403 is used in the sample form image 400, and the item name “partial burden” 404 is in the sample form image 401. The item name “total” 405 is used in the sample document image 402. When each sample document image 400, 401, 402 is registered, the registrant registers the correct value "15000" for the item "patient's burden" in each of the sample document images 400, 401, 402.

FIG. 5 is a diagram showing an example of the plurality of rule lists 500, 501, 502. As shown in FIG.

Each rule list 500, 501, 502 is generated by the generation device 200 and stored in the second storage device 210, and is also transmitted to the information processing device 100 and stored in the first storage device 110. As shown in FIG. 5, a rule list ID for identifying each rule list is assigned to each rule list 500, 501, 502. Also, in each rule list 500, 501, 502, a plurality of rules (information corresponding to each row in the table shown in FIG. 5) are defined in association with each other, and for each rule, the rule number, priority, item name And location information etc. are defined in association with each other. That is, one rule corresponds to one item name.

The rule number is a number for identifying each rule. The priority order is an order for defining the order in which the item names defined in each rule are inspected when detecting the item value of the item of interest from the form image using each rule list. In this example, the priority is assigned to each rule so that the priority becomes higher as the numerical value “1” has the highest priority and the numerical value becomes larger. The item name is a name to be inspected in each rule when detecting an item value of an item of interest from the form image using each rule list. The position information is information for defining the display position relationship between each item name and each item value corresponding to each item name in each sample form image. As position information, for example, relative positions of item values with respect to item names (for example, upper, lower, right, left, upper right, lower right, upper left, lower left) are defined. The position information is used to specify an item value corresponding to an item name defined in each rule when detecting an item value of an item of interest from the form image using each rule list.

FIG. 6 shows an example of the rule dictionary 600. As shown in FIG.

The rule dictionary 600 is generated by the generation device 200 and stored in the second storage device 210, and is also transmitted to the information processing device 100 and stored in the first storage device 110. As shown in FIG. 6, in the rule dictionary 600, each of a plurality of feature amounts is defined in association with each of a plurality of rule list IDs.

The feature amount is a feature amount in each sample form image. Details of the feature amount will be described later. In the rule dictionary 600, the feature amount of the sample form image that conforms to each rule list is associated with each rule list.

FIG. 7 is a flowchart showing an example of the operation of the generation process of the rule dictionary by the generation device 200.

Hereinafter, an example of the operation of the generation processing by the generation device 200 will be described with reference to the flowchart illustrated in FIG. 7. The flow of the operation described below is executed mainly by the second CPU 220 in cooperation with each element of the generation device 200 based on a program stored in advance in the second storage device 210.

First, the rule list generation unit 221 initializes the rule list, and the rule dictionary generation unit 222 initializes the rule dictionary (step S101). The rule list generation unit 221 extracts all item names that can correspond to the item of interest from all the sample form images stored in the second storage device 210. An item name that can correspond to the item of interest may be registered in advance. The rule list generation unit 221 generates a rule in which position information corresponding to all possible positions is associated with each extracted item name, and generates a rule list including all the generated rules as an initial rule. Generate as a list. In the initial rule list, rule numbers are assigned to each rule in an arbitrary order, and "undefined" is set as the priority. Further, the rule dictionary generation unit 222 generates a rule dictionary in which “indefinite” is set as the feature amount and the rule list ID as an initial rule dictionary.

Next, the rule list generation unit 221 selects one rule from the initial rule list (step S102).

Next, based on the selected rule, the rule list generation unit 221 detects item values from each of all sample form images that do not match any of the generated rule lists (step S103). When the process of step S103 is executed first, the rule list generation unit 221 detects item values from all sample form images stored in the second storage device 210. The rule list generation unit 221 detects an item name in the selected rule from each sample form image using a known optical character recognition (OCR) technology. Furthermore, the rule list generation unit 221 uses the well-known OCR technology to set the item values from the area of the predetermined size located in the display positional relationship defined in the positional information in the selected rule to the detected item name. To detect. Thus, the rule list generation unit 221 detects the item value corresponding to each item name based on the display position relationship defined in the rule list.

Next, the rule list generation unit 221 calculates a match rate indicating the rate at which each item value detected from each sample form image matches the correct value associated with each sample form image for which each item value is detected. (Step S104). The rule list generation unit 221 determines whether or not each item value detected from each sample document image matches the correct value associated with each sample document image for which each item value is detected, and detects the detected item value. The ratio of the item value that matches each correct value to the number is calculated as the matching rate.

Next, the rule list generation unit 221 determines whether the process is completed for all the rules included in the initial rule list (step S105). If there is a rule whose processing has not been completed yet, the rule list generation unit 221 returns the processing to step S102, and executes the processing of steps S102 to S104 on the unselected rule.

On the other hand, when the process is completed for all the rules included in the initial rule list, the rule list generation unit 221 assigns the priority of each rule so that the priority becomes higher in descending order of the match rate calculated in step S104. A rule list is generated (step S106). As described above, the rule list generation unit 221 is configured such that, in each sample form image that does not conform to any of the plurality of rule lists, the rule list is configured such that the item name with the higher proportion corresponding to the correct value has higher priority. Generate Further, the rule list generation unit 221 assigns a rule list ID to the generated rule list.

Next, the rule dictionary generation unit 222 selects one sample slip image from among sample slip images that do not conform to any of the generated rule lists (step S107).

Next, the rule dictionary generation unit 222 extracts item values from the selected sample document image based on the rule list generated in step S106 (step S108). The rule dictionary generation unit 222 includes, in the sample document image, each item name defined in the rule list in the order of priority defined in the rule list (in the order of high priority) using known OCR technology. It is sequentially determined whether it is possible or not. The rule dictionary generation unit 222 detects an item value from an area of a predetermined size located in the display position relationship defined in the position information associated with the item name, for the item name determined to be included first. Extract the detected item value as the item value corresponding to the item name.

Next, the rule dictionary generation unit 222 determines whether or not the extracted item value matches the correct value associated with the selected sample form image (step S109).

If the item value and the correct value do not match, the rule dictionary generation unit 222 determines that the rule list generated in step S106 and the sample document image selected in step S107 do not match, and the process proceeds to step S113. Do.

On the other hand, when the item value and the correct answer value match, the rule dictionary generation unit 222 determines that the rule list generated in step S106 matches the sample document image selected in step S107 (step S110).

Next, the rule dictionary generation unit 222 calculates the feature amount from the sample document image determined to conform to the rule list (step S111). The rule dictionary generation unit 222 calculates, as the feature quantities of the sample document image, a feature quantity relating to the shape of a ruled line, a feature quantity relating to a logo, or a feature quantity relating to a character recognition result.

FIG. 8 is a diagram for explaining the feature amount related to the shape of the ruled line.

As shown in FIG. 8, when the form image 800 includes the table 801, a plurality of apexes 803 or intersections 804 having different shapes are formed by the ruled lines 802 constituting the table 801. The rule dictionary generation unit 222 counts the number of vertices 803 and intersections 804 having each shape for each of the vertices 803 and intersections 804 having different shapes. The rule dictionary generation unit 222 calculates a histogram (feature vector) 810 having the number of vertexes 803 having the respective shapes and the number of intersections 804 as a feature of the form image 800. The rule dictionary generation unit 222 can identify each form of various layouts with high possibility of including a table with high accuracy by using the feature amount related to the shape of the ruled line.

FIG. 9 is a diagram for explaining the feature amount related to the logo.

As shown in FIG. 9, generally, the form image 900 includes various logo images, such as a pattern logo 901 or a character logo 902, representing an organization that issues the form. The rule dictionary generation unit 222 detects logos 901 and 902 having each shape and counts the number of logos 901 and 902 having each shape using a known pattern matching technology. For example, the generating device 200 stores in advance a correct image obtained by binarizing the images of the logos 901 and 902. The rule dictionary generation unit 222 sequentially cuts out the images while moving the cutout range in the form image 900, and the pixel coincidence rate between the binarized image obtained by binarizing the cutout image and the correct image is equal to or more than the threshold In this case, the extracted range is detected as a logo. The rule dictionary generation unit 222 calculates a histogram (feature vector) 910 having the number of logos 901 and 902 having each shape as elements as the feature amount of the form image 900. The rule dictionary generation unit 222 can identify each form individually created by each organization with high accuracy by using the feature amount related to the logo of each organization.

FIG. 10 is a diagram for explaining the feature amount related to the character recognition result.

As shown in FIG. 10, generally, the form image 1000 includes various character strings 1001 to 1005. The rule dictionary generation unit 222 detects character strings 1001 to 1005 from the form image 1000 using known OCR technology, and counts the number of the character strings 1001 to 1005. The rule dictionary generation unit 222 calculates a histogram (feature vector) 1010 having the number of character strings 1001 to 1005 as an element as a feature amount of the form image 1000. The rule dictionary generation unit 222 can identify each form of various layouts with higher accuracy by using the feature amount related to the character recognition result.

Note that the rule dictionary generation unit 222 may combine two or more of the feature amount related to the shape of the ruled line, the feature amount related to the logo, or the feature amount related to the character recognition result and use it as the feature amount of the sample document image . Thereby, the rule dictionary generation unit 222 can identify each form with higher accuracy. In addition, the generation device 200 may calculate in advance the feature amount of each sample document image and store the calculated feature amount in the second storage device 210. In that case, the rule dictionary generation unit 222 acquires the feature amount of each sample form image from the second storage device 210. As a result, the rule dictionary generation unit 222 can reduce the processing load for the generation process and generate the rule dictionary in a shorter time.

Next, the rule dictionary generation unit 222 stores (adds) the calculated feature amount in the rule dictionary in association with the rule list ID of the rule list generated in step S106 (step S112).

The rule dictionary generation unit 222 omits the update of the rule dictionary when the calculated feature amount matches the feature amount calculated in the past. The rule dictionary generation unit 222 may omit the update of the rule dictionary also when the calculated feature amount is similar to the feature amount calculated in the past. In that case, the rule dictionary generation unit 222 determines that the two feature quantities are similar when the calculated feature quantity and the similarity between the feature quantities calculated in the past are equal to or greater than a threshold. The similarity is, for example, the reciprocal of the Euclidean distance of each feature (feature vector), or a cosine similarity. In addition, when the calculated feature quantity matches or is similar to the feature quantity calculated in the past, the rule dictionary generation section 222 newly generates a rule list ID associated with the feature quantity, which is already stored in the rule dictionary. The rule list ID of the rule list may be updated (overwritten).

Alternatively, the rule dictionary generation unit 222 may allocate a feature amount ID to the calculated feature amount, and store the allocated feature amount ID in the rule dictionary in association with the rule list ID. For example, the rule dictionary generation unit 222 uses, as a feature quantity ID, a value (vector) obtained by rounding each element of the calculated feature quantity (feature vector) to a predetermined significant digit number. The rule dictionary generation unit 222 may use a value (vector) obtained by normalizing the calculated feature amount (feature vector) as a feature amount ID.

Next, the rule dictionary generation unit 222 determines whether or not the processing in steps S107 to S112 is completed for all sample form images that do not conform to any of the generated rule lists (step S113). If there is a sample form image that has not been processed yet, the rule dictionary generation unit 222 returns the process to step S107, and executes the processes of steps S107 to S112 on the unselected sample form image.

On the other hand, when the process is completed for each sample document image, the rule dictionary generation unit 222 determines whether all the sample document images match any of the generated rule lists (step S114). If there is a sample document image that does not match any of the generated rule lists, the rule dictionary generation unit 222 returns the process to step S102. In this case, the rule list generation unit 221 generates a new rule list based on the sample form image which does not match any of the generated rule lists by the processes of steps S102 to S106. On the other hand, the rule dictionary generation unit 222 updates the rule dictionary using the newly generated rule list by the processing of steps S107 to S112.

On the other hand, if all sample form images match any of the generated rule lists, the rule dictionary generation unit 222 ends the series of steps. The rule dictionary stored in the second storage device 210 at the end of the generation process becomes the rule dictionary generated by the rule dictionary generation unit 222.

As in step S106, the rule list generation unit 221 determines, based on the matching rate of the item value and the correct value in each sample form image not conforming to any of the generated rule lists, Determine the order of precedence in the rule list for performing a conformance check. As a result, the rule dictionary generation unit 222 can efficiently check whether each sample document image matches each rule list, making the rule dictionary simpler and reducing the generation time of the rule dictionary. It becomes.

When a sample document image is newly added after the rule dictionary is generated, the rule list generation unit 221 generates a new rule list based on the newly added sample document image. In such a case, the rule list generation unit 221 executes the processing of steps S102 to S106 with the newly added sample form image as a sample form image not conforming to any of the generated rule lists, and performs a new process. Generate a rule list. On the other hand, the rule dictionary generation unit 222 updates the rule dictionary using the newly generated rule list by the processing of steps S107 to S112. However, if the newly generated rule list is identical to any of the generated rule lists, the rule dictionary generation unit 222 updates the rule dictionary using the generated rule list.

Thus, the generating device 200 can update the rule dictionary by newly adding a sample form image to the rule dictionary generated once. In particular, when a sample document image that does not match any of the generated rule lists is newly added, each sample document image and the sample document image are based on the matching rate of the item value and the correct value in each of the newly added sample document images. The order of precedence in the rule list for performing a conformance check of As a result, the rule dictionary generation unit 222 can efficiently check whether each newly added sample form image and each rule list match, making the rule dictionary simpler and generating the rule dictionary. It is possible to reduce.

11 to 13 are diagrams showing an example of the rule dictionary generated by the generation process shown in FIG.

In FIGS. 11 to 13, generation processing is performed on nine sample form images 1 to 9 shown in FIG. First, as shown in FIG. 11, each item value corresponding to the item name "patient's burden" is detected from each sample document image 1 to 9 (step S103), and each detected item value and each sample document image The matching rate with the correct value associated with 1 to 9 is calculated (step S104). Similarly, a match rate is calculated for each item value corresponding to the item name "partial contribution" and each item value corresponding to the item name "total", and a rule list to which priority is assigned in descending order of the match rate 1101 is generated (step S106). In this example, in the rule list 1101, priorities are assigned in the order of patient burden, partial contribution, and total.

Next, it is determined whether or not each item name is included in the order in which the priority defined in the rule list 1101 is high (in order of patient burden, partial contribution, and total) with respect to the sample form image 1 The item value corresponding to the item name determined to be included first is extracted (step S108). Then, when the extracted item value matches the correct value value associated with the sample document image 1, the feature amount A of the sample document image 1 and the rule list ID of the rule list 1101 are associated and stored in the rule dictionary 1102. (Step S109). Similarly, when the item values extracted from the sample form images 2 to 9 match the correct values associated with the sample form images 2 to 9, the feature amounts B to I of the sample form images 2 to 9 and the rule list The rule list 1101 is stored in the rule dictionary 1102 in association with the rule list ID.

In this example, the item values extracted from the sample form images 1 to 6 match the correct values associated with the sample form images 1 to 6, and the feature amounts A to F of the sample form images 1 to 6 and the rule list The rule list 1101 is stored in the rule dictionary 1102 in association with the rule list ID. On the other hand, the item values extracted from sample form images 7 to 9 do not match the correct values associated with sample form images 7 to 9, and feature amounts G to I of sample form images 7 to 9 are still rules. It is not stored in the dictionary 1102.

Next, as shown in FIG. 12, the rule list 1201 is generated from the sample form images 7 to 9 that did not match the rule list 1101. In this example, in the rule list 1201, priorities are assigned in the order of partial contribution, patient contribution, and total.

Next, whether the item value extracted from the sample form images 7 to 9 matches the correct value value associated with the sample form images 7 to 9 using the rule list 1201 for the sample form images 7 to 9 It is determined whether or not the rule dictionary 1202 is updated. In this example, the item values extracted from the sample form images 7-8 match the correct values associated with the sample form images 7-8, and the feature amounts G to H of the sample form images 7 to 8 and the rule list The rule list ID 1201 is stored in the rule dictionary 1202 in association with the rule list ID. On the other hand, the item value extracted from the sample document image 9 does not match the correct value associated with the sample document image 9, and the feature amount I of the sample document image 9 is not stored in the rule dictionary 1202 yet.

Next, as shown in FIG. 13, a rule list 1301 is generated from the sample form image 9 which does not match the rule list 1201. In this example, priorities are assigned in the order of total, patient contribution, and partial contribution.

Next, it is determined whether or not the item value extracted from the sample form image 9 matches the correct value associated with the sample form image 9 using the rule list 1301 for the sample form image 9. The rule dictionary 1302 is updated. In this example, the item value extracted from the sample document image 9 matches the correct value associated with the sample document image 9, and the feature amount I of the sample document image 9 and the rule list ID of the rule list 1301 They are associated and stored in the rule dictionary 1302. Thus, the processing for all the sample form images 1 to 9 is completed, and the rule dictionary 1302 is completed.

FIG. 14 is a flowchart illustrating an example of the operation of the process of detecting an item value from a form image by the information processing apparatus 100.

Hereinafter, an example of the operation of the detection process by the information processing apparatus 100 will be described with reference to the flowchart illustrated in FIG. The flow of the operation described below is executed mainly by the first CPU 120 in cooperation with each element of the information processing apparatus 100 based on a program stored in advance in the first storage device 110.

First, the receiving unit 121 transmits an acquisition request for requesting acquisition of the rule dictionary and the rule list to the generating device 200 via the first communication device 101, and the rule from the generating device 200 via the first communication device 101. The dictionary and the rule list are received (step S201). When receiving the acquisition request from the information processing apparatus 100 via the second communication apparatus 201, the transmitting unit 223 of the generating apparatus 200 reads the rule dictionary and the rule list generated in advance from the second storage apparatus 210, and 2. Transmit to the information processing apparatus 100 via the communication apparatus 201. The receiving unit 121 stores the received rule dictionary and rule list in the first storage device 110. The information processing apparatus 100 may obtain the rule dictionary and the rule list in advance and store the rule dictionary and the rule list in the first storage device 110.

Next, the acquiring unit 122 acquires a form image to be detected as an item value from an image reading apparatus (not shown) via the first communication apparatus 101 (step S202). The acquisition unit 122 may acquire a form image from the image reading apparatus via an interface circuit (not shown) such as USB (Universal Serial Bus). Alternatively, the information processing apparatus 100 may include an imaging device, and the acquisition unit 122 may acquire a document image by imaging a form using the imaging device.

Next, the feature quantity calculation unit 123 calculates a feature quantity from the form image acquired by the acquisition unit 122, as in the process of step S111 in FIG. 7 (step S203). The feature amount calculation unit 123 calculates the same feature amount as the feature amount calculated by the rule dictionary generation unit 222 of the generation device 200. When the rule dictionary generation unit 222 stores the feature amount ID in association with the rule list ID in the rule dictionary, the feature amount calculation unit 123 calculates the feature amount calculated by the same method as the rule dictionary generation unit 222. Assign feature quantity ID. In that case, the information processing apparatus 100 executes the following process using the feature amount ID instead of the feature amount.

Next, the rule list extraction unit 124 determines whether the feature amount calculated by the feature amount calculation unit 123 is stored in the rule dictionary stored in the first storage device 110 (step S204). Note that the rule list extraction unit 124 is also configured to store the feature amount that is similar to the feature amount calculated by the feature amount calculation unit 123, as well as the feature amount that is similar to the feature amount calculated by the feature amount calculation unit 123. It may be determined that the feature quantity calculated by the above is stored. In such a case, the rule list extraction unit 124 determines that the two feature quantities are similar if the similarity between the feature quantity calculated by the feature quantity calculation unit 123 and the feature quantity stored in the rule dictionary is equal to or greater than a threshold. judge. The similarity is, for example, the reciprocal of the Euclidean distance of each feature (feature vector), or a cosine similarity.

When the feature quantity calculated by the feature quantity calculation unit 123 is stored in the rule dictionary, the rule list extraction unit 124 extracts the rule list corresponding to the rule list ID associated with the feature quantity in the rule dictionary. (Step S205).

On the other hand, when the feature quantity calculated by the feature quantity calculation unit 123 is not stored in the rule dictionary, that is, when there is no rule list associated with the feature quantity, the rule list extraction unit 124 , And extract an alternative rule list (step S206). The rule list extraction unit 124 extracts, for example, a rule list associated with the most feature amount in the rule dictionary as an alternative rule list. As a result, the rule list extraction unit 124 can accurately detect the item value of the item to be noticed from the form image in which the feature amount is not registered in the rule dictionary. The rule list extraction unit 124 may extract the alternative rule list according to other conditions.

Next, the detection unit 125 detects an item value from the form image based on the rule list extracted by the rule list extraction unit 124 (step S207). The detecting unit 125 determines whether or not each item name defined in the rule list is included in the form image in the order of the priorities defined in the rule list (in the order of high priority) using known OCR technology. The decision is made one by one. The detection unit 125 detects an item value from an area of a predetermined size located in the display positional relationship defined in the position information associated with the item name, for the item name determined to be included first, Detect as an item value corresponding to the item name. Thus, the detection unit 125 detects the item value corresponding to each item name based on the display position relationship defined in the rule list.

Next, the output control unit 126 outputs the item value by displaying a recognition result screen including the item value detected by the detection unit 125 on the display device 103 (step S208).

FIG. 15 is a view showing an example of the recognition result screen.

As shown in FIG. 15, on the recognition result screen 1500, an item 1501, a recognition result 1502, a correction button 1503, a confirmation button 1504 and the like are displayed for each item of interest. The item value of each item detected from the form image is displayed as a recognition result 1502 in association with each item 1501. When the correction button 1503 is pressed by the user using the input device 102, the recognition result 1502 can be corrected (edited) by the operator. When the user presses the enter button 1504 using the input device 102, the item values of the items detected from the form image are corrected to the correction values designated in the recognition result 1502. Note that, instead of the correction button 1503 and the determination button 1504 displayed for each item, one correction button and a determination button commonly used for all the items may be displayed. In that case, when the correction button is pressed, all recognition results 1502 can be corrected (edited) by the operator, and when the confirmation button is pressed, all correction values designated in each recognition result 1502 are item values. Reflected in

Note that the output control unit 126 outputs the item value by transmitting the item value to a server or the like that aggregates each piece of information related to the form via the first communication device 101 instead of displaying the item value on the display device 103. May be

Next, the reception unit 127 determines whether or not the user has received an instruction to correct the item value displayed on the recognition result screen 1500 by pressing the enter button 1504 using the input device 102 by the user. (Step S209). If the user does not receive the correction instruction, the information processing apparatus 100 ends the series of steps.

On the other hand, when the correction instruction from the user is accepted, the updating unit 128 executes the updating process (step S210), and ends the series of steps. The updating unit 128 updates the rule list associated with the feature amount calculated by the feature amount calculating unit 123 in the rule dictionary in the updating process. Details of the update process will be described later.

FIG. 16 is a flowchart illustrating an example of the operation of the update process. The update process shown in FIG. 16 is executed in step S210 of FIG.

First, the updating unit 128 selects one rule list from among the rule lists stored in the first storage device 110 (step S301).

Next, the updating unit 128 extracts item values from the form image based on the selected rule list (step S302). Whether the item names specified in the rule list are included in the sample document image in the order of the priorities defined in the rule list (in the order of high priority) using the known OCR technology Whether or not it will be judged one by one. The updating unit 128 detects an item value from an area of a predetermined size located in the display positional relationship defined in the position information associated with the item name, for the item name determined to be included first, Extract as the item value corresponding to the item name.

Next, the updating unit 128 determines whether the extracted item value matches the correction value designated by the user using the input device 102 (step S303).

If the item value matches the correct value, the updating unit 128 determines that the form image matches the rule list selected in step S301. In that case, the updating unit 128 associates the feature quantity calculated by the feature quantity calculating unit 123 in step S203 with the rule list ID of the rule list selected in step S301 and stores it in the rule dictionary (step S304) Finish the step. Thus, the updating unit 128 updates the rule dictionary stored in the first storage device 110. The updating unit 128 may transmit the updated rule dictionary to the generating device 200 via the first communication device 101, and may update the rule dictionary stored in the second storage device 210. As a result, the updating unit 128 can appropriately update the rule dictionary used by the other information processing apparatus.

Further, when the feature quantity calculated by the feature quantity calculating unit 123 is already stored in the rule dictionary, the updating unit 128 updates (overwrites) the rule list ID associated with the feature quantity. Accordingly, the updating unit 128 can appropriately improve the quality of the rule dictionary by appropriately updating the rule list improperly registered in the rule dictionary. On the other hand, when the feature quantity calculated by the feature quantity calculating unit 123 is not yet stored in the rule dictionary, the updating unit 128 associates the feature quantity with the rule list ID and newly adds the feature quantity to the rule dictionary. As a result, the updating unit 128 can add new information not yet registered to the rule dictionary to efficiently improve the quality of the rule dictionary.

On the other hand, when the item value and the correction value do not match, the updating unit 128 determines whether the process has been completed for all the rule lists stored in the first storage device 110 (step S305). If there is a rule list whose processing has not been completed yet, the updating unit 128 returns the processing to step S301, and executes the processing of steps S301 to S305 on the unselected rule list. On the other hand, when the process is completed for all the rule lists, the updating unit 128 ends the series of steps.

As described in detail above, the information processing apparatus 100 detects the item value of the desired item from the form image in accordance with the priority defined in the rule list associated with the feature amount of the form image. In a form with various layouts, the entry position of the item name may be different for each form, and there may be a plurality of synonyms of the item name in one form. On the other hand, in form images having similar feature amounts, there is a high possibility that the same item name is used for the same item. It becomes possible for the information processing apparatus 100 to inspect each item name in descending order of the possibility of corresponding to the desired item from the form image in which the item name and the item value are represented for each of a plurality of items. It has become possible to detect the item value of the item more accurately.

Further, the information processing apparatus 100 detects an item value of a desired item from the form image, using a rule list in which the display position relationship of item values corresponding to each item name is defined. As a result, the information processing apparatus 100 can detect the item value of the desired item with high accuracy from the form image of the form of the multi-layout form in which the description position of the item value for the item name is different for each form. .

In addition, the generating apparatus 200 examines each item name in the order of priority defined in the rule list, and first determines an item value corresponding to the item name determined to be included in the sample form image and its correct value If there is a match, the rule list is associated with the sample form image. As a result, the generating apparatus 200 can generate a rule list capable of correctly detecting the item value of a desired item by examining each item name in order of priority.

In addition, the generating apparatus 200 generates a rule dictionary by performing learning in advance using a plurality of sample form images. As a result, the generation device 200 suppresses the number of rule lists included in the rule dictionary from becoming a huge amount, and highly accurate item values of desired items even for form images whose layout is unknown. It is possible to generate a rule dictionary that can be detected in

FIG. 17 is a block diagram showing a schematic configuration of the first processing circuit 320 in the information processing apparatus according to another embodiment.

The first processing circuit 320 is used instead of the first CPU 320, and executes detection processing instead of the first CPU 320. The first processing circuit 320 includes a reception circuit 321, an acquisition circuit 322, a feature quantity calculation circuit 323, a rule list extraction circuit 324, a detection circuit 325, an output control circuit 326, a reception circuit 327, an update circuit 328, and the like.

The receiving circuit 321 is an example of a receiving unit, and has the same function as the receiving unit 121. The receiving circuit 321 receives the rule dictionary and the rule list from the generating device via the first communication device 101, and stores the received rule dictionary and the rule list in the first storage device 110.

The acquisition circuit 322 is an example of an acquisition unit, and has the same function as the acquisition unit 122. The acquisition circuit 322 receives a form image from an image reading device (not shown) via the first communication device 101, and outputs the received form image to the feature amount calculation circuit 323 and the detection circuit 325.

The feature amount calculation circuit 323 is an example of a feature amount calculation unit, and has the same function as the feature amount calculation unit 123. The feature amount calculation circuit 323 receives the form image from the acquisition circuit 322, calculates the feature amount from the received form image, and outputs the feature amount to the rule list extraction circuit 324.

The rule list extraction circuit 324 is an example of a rule list extraction unit, and has the same function as the rule list extraction unit 124. The rule list extraction circuit 324 reads the rule dictionary from the first storage device 110, receives the feature amount from the feature amount calculation circuit 323, extracts the rule list from the rule dictionary and the feature amount, and outputs the rule list to the detection circuit 325.

The detection circuit 325 is an example of a detection unit, and has the same function as the detection unit 125. The detection circuit 325 receives a form image from the acquisition circuit 322, receives a rule list from the rule list extraction circuit 324, detects an item value of an item of interest from the form image and the rule list, and outputs the value to the output control circuit 326. Do.

The output control circuit 326 is an example of an output control unit, and has the same function as the output control unit 126. The output control circuit 326 receives an item value from the detection circuit 325, and outputs the received item value to the display device 103.

The reception circuit 327 is an example of a reception unit, and has the same function as the reception unit 127. The reception circuit 327 receives the correction value from the input device 102 and outputs the received correction value to the update circuit 328.

The update circuit 328 is an example of the update unit, and has the same function as the update unit 128. The update circuit 328 reads the rule list from the first storage device 110, receives the correction value from the reception circuit 327, and updates the rule dictionary from the rule list and the correction value.

FIG. 18 is a block diagram showing a schematic configuration of the second processing circuit 420 in the generation device according to another embodiment.

The second processing circuit 420 is used instead of the second CPU 220, and executes generation processing instead of the second CPU 220. The second processing circuit 420 includes a rule list generation circuit 421, a rule dictionary generation circuit 422, a transmission circuit 423, and the like.

The rule list generation circuit 421 is an example of a rule list generation unit, and has the same function as the rule list generation unit 221. The rule list generation circuit 421 reads out a sample form image from the second storage device 210, generates a rule list based on the read out sample form image, and stores the rule list in the second storage device 210.

The rule dictionary generation circuit 422 is an example of a rule dictionary generation unit, and has the same function as the rule dictionary generation unit 222. The rule dictionary generation circuit 422 reads out the sample form image and the rule list from the second storage device 210, generates a rule dictionary based on the read out sample form image and the rule list, and stores the rule dictionary in the second storage device 210.

The transmission circuit 423 is an example of a transmission unit, and has the same function as the transmission unit 223. The transmission circuit 423 reads the rule dictionary from the second storage device 210, and transmits the read rule dictionary to the information processing device via the second communication device 201.

As described above, even when the information processing apparatus uses the first processing circuit 320 and the generation apparatus uses the second processing circuit 420, the information processing apparatus detects the item value of the desired item with higher accuracy. It became possible.

Although the preferred embodiments of the present invention have been described above, the present invention is not limited to these embodiments. For example, the generation process may be performed by the information processing device 100 instead of being performed by the generation device 200. In that case, the first storage device 110 stores each piece of information stored in the second storage device 210, and the first CPU 120, in addition to the portions shown in FIG. 2, has each portion having the same function as each portion of the second CPU 220. Have.

Further, the number of item values of the item of interest to be detected from each form image is not limited to one. The generation device 200 may generate a rule dictionary and a rule list for each of two or more items, and the information processing device 100 may detect item values of two or more items from each form image.

100 information processing device 102 input device 103 display device 110 first storage device 121 reception unit 122 acquisition unit 123 feature amount calculation unit 124 rule list extraction unit 125 detection unit 127 reception unit 128 update unit 200 generation device 210 second storage device 221 rule List generation unit 222 Rule dictionary generation unit 223 Transmission unit

Claims (11)

1. An information processing apparatus that processes a form image in which an item name and an item value are represented for each of a plurality of items.
   A storage unit storing a rule dictionary in which each of the feature amounts for each of a plurality of sample form images is associated with each of a plurality of item names and a plurality of rule lists in which priorities of the item names are defined;
   An acquisition unit for acquiring a form image,
   A feature amount calculation unit that calculates a feature amount from the acquired form image;
   A rule list extraction unit that extracts a rule list associated with the calculated feature amount in the rule dictionary;
   It is sequentially determined whether or not each item name defined in the rule list is included in the form image in the order of the priority defined in the extracted rule list, and it is determined that the item image is first included A detection unit that detects an item value corresponding to the item name;
   An output unit that outputs the detected item value;
   An information processing apparatus comprising:
2. In the rule list, the display position relationship between the plurality of item names, the priority order of each item name, the item names in the sample form image, and the item values corresponding to the item names is defined.
   The information processing apparatus according to claim 1, wherein the detection unit detects an item value corresponding to the item name determined to be included first based on the display position relation.
3. The information processing apparatus according to claim 1, wherein the feature amount for each form image is a feature amount related to a shape of a ruled line, a feature amount related to a logo, or a feature amount related to a character recognition result.
4. The rule list extraction unit extracts a rule list associated with the largest number of feature amounts when there is no rule list associated with the calculated feature amount in the rule dictionary. The information processing apparatus according to any one of the above.
5. A receiving unit for receiving a correction instruction of the output item value from the user;
   The update method according to any one of claims 1 to 4, further comprising: an update unit for updating a rule list associated with the calculated feature amount in the rule dictionary when the correction instruction is accepted. Information processing device.
6. An information processing system comprising: a generation device; and an information processing device for processing a form image in which an item name and an item value are represented for each of a plurality of items.
   The generator is
   A storage unit that stores a plurality of sample form images, a plurality of item names, and a plurality of rule lists in which priorities of the item names are defined;
   A rule dictionary generation unit that generates a rule dictionary in which feature amounts of the sample form image that conform to each of the plurality of rule lists are associated with each rule list;
   And a transmitter configured to transmit the rule dictionary to the information processing apparatus.
   The information processing apparatus is
   A receiver for receiving the rule dictionary from the generator;
   An acquisition unit for acquiring a form image,
   A feature amount calculation unit that calculates a feature amount from the acquired form image;
   A rule list extraction unit that extracts a rule list associated with the calculated feature amount in the received rule dictionary;
   It is sequentially determined whether or not each item name defined in the rule list is included in the form image in the order of the priority defined in the extracted rule list, and it is determined that the item image is first included A detection unit that detects an item value corresponding to the item name;
   And an output unit that outputs the detected item value.
   An information processing system characterized by
7. When there is a sample form image that does not match any of the plurality of rule lists, the rule list generation unit further generates a new rule list based on the sample form image that does not match any of the plurality of rule lists. The information processing system according to claim 6.
8. The storage unit stores each of the plurality of sample form images in association with a correct value,
   The rule list generation unit is configured such that, in each sample form image not conforming to any of the plurality of rule lists, the priority is increased such that the item name with a higher ratio corresponding to the correct value becomes higher. The information processing system according to claim 7, generating a rule list.
9. The storage unit stores each of the plurality of sample form images in association with a correct value,
   The rule dictionary generation unit sequentially determines whether each item name defined in the rule list is included in the sample document image in the order of the priority defined in the rule list, and is first included It is determined that the rule list and the sample form image match if the item value corresponding to the item name determined to be identical with the correct value associated with the sample form image. Information processing system.
10. A control method of an information processing apparatus, which has a storage unit and an output unit and processes a form image in which an item name and an item value are represented for each of a plurality of items.
    Storing, in the storage unit, a rule dictionary in which each of the feature amounts for each of a plurality of sample form images is associated with each of a plurality of item names and a plurality of rule lists in which the priorities of the item names are defined;
    Get form image,
    Calculate the feature amount from the acquired form image,
    Extracting a rule list associated with the calculated feature amount in the rule dictionary;
    It is sequentially determined whether or not each item name defined in the rule list is included in the form image in the order of the priority defined in the extracted rule list, and it is determined that the item image is first included Detect the item value corresponding to the item name,
    Outputting the detected item value to the output unit;
    Control method characterized in that it includes.
11. A control program of an information processing apparatus having a storage unit and an output unit and processing a form image in which an item name and an item value are represented for each of a plurality of items,
    Storing, in the storage unit, a rule dictionary in which each of the feature amounts for each of a plurality of sample form images is associated with each of a plurality of item names and a plurality of rule lists in which the priorities of the item names are defined;
    Get form image,
    Calculate the feature amount from the acquired form image,
    Extracting a rule list associated with the calculated feature amount in the rule dictionary;
    It is sequentially determined whether or not each item name defined in the rule list is included in the form image in the order of the priority defined in the extracted rule list, and it is determined that the item image is first included Detect the item value corresponding to the item name,
    Outputting the detected item value to the output unit;
    Control program for causing the information processing apparatus to execute the control program.

Images