JP2021012487A - Image processing program, image processing method, and image processing device - Google Patents

Abstract

To provide a technology capable of extracting the seal impression of a registered seal from an image of a document by eliminating the need of predefining the position of a registered seal column.SOLUTION: An image processing program executes first detection processing for detecting an area of a keyword showing an item name of a registered seal in an image of a document, and second detection processing for detecting an area of a seal impression in the image of the document, and causes a computer to execute processing for specifying the area of the seal impression of the registered seal in the image of the document on the basis of execution results of the first detection processing and the second detection processing.SELECTED DRAWING: Figure 3

Description

The present invention relates to an image processing program, an image processing method, and an image processing apparatus.

As one of the form processing systems used in banks and the like, a system that handles forms with a notification stamp is known. This type of form processing system includes an image processing device. The image processing device acquires an optically read image of the form, extracts the imprint of the notification stamp from the acquired image of the form, and electronically converts the imprint of the extracted notification stamp and the imprint of the pre-registered notification stamp. Perform processing such as collation (so-called seal collation).

The following method is known as a method for extracting the imprint of the notification stamp from the image of the form. For example, a method (method 1) of extracting the imprint of the notification stamp based on the position of the notification stamp column defined in advance is known (see, for example, Patent Document 1). In addition, for example, the position of the notification stamp column in the image of the form is detected by using the character string indicating the item name of the notification stamp (for example, "delivery stamp"), and the notification stamp is based on the position of the detected notification stamp column. A method (method 2) for extracting the imprint of the above is also known. Further, for example, a method is also known in which a customer imprint is extracted by detecting an imprint from an image of a form and excluding a pre-registered exclusion imprint (for example, a seal imprint) from the detected imprints. (See, for example, Patent Document 2).

Further, as a method of extracting the imprint of the notification stamp from the image of the form without using the character string indicating the item name of the notification stamp as in method 2, for example, object detection AI (artificial) such as deep learning (deep learning) is performed. A method (method 3) of extracting the imprint of the registered seal by using the intelligence) technique or the date stamp area specifying technique described in the patent application of Japanese Patent Application No. 2018-091876 by the applicant of the present application is also conceivable.

Japanese Unexamined Patent Publication No. 2011-197787 Japanese Unexamined Patent Publication No. 2013-167928

However, in Method 1, it is necessary to define the position of the notification stamp column in advance, so when handling various forms with different positions of the notification stamp column, it takes a lot of labor to define the position. Become. For example, since the account transfer request form, which is one of the forms, is created by the storage institution (payee), the account transfer request form with a different layout (that is, the account transfer request form with a different position of the notification stamp column). There are many types, and it is not easy to individually define the position of the notification stamp column.

In the method 2, when the design of the notification stamp column is high, the position of the notification stamp column may not be detected by using the character string indicating the item name of the notification stamp. In general, the notification stamp column tends to have a high design, unlike the account number column and the amount column. FIG. 13 is a diagram illustrating a notification stamp column having a high design. The notification stamp column illustrated in FIG. 13 is a notification stamp column in which a character string indicating an item name of the notification stamp is shown in white. In the case of such a notification stamp column, it may not be possible to detect the position of the notification stamp column using the character string indicating the item name of the notification stamp.

In the method 3, when there is a correction stamp, a stamp, a sample stamp, or the like in addition to the stamp of the notification stamp, the stamp of the notification stamp cannot be uniquely specified. FIG. 14 is a diagram illustrating imprints other than the notification stamp. The imprint illustrated in (a) of FIG. 14 is an imprint of a stamp, and the imprint illustrated in (b) of FIG. 14 is an imprint of a sample stamp. When there is an imprint other than the notification stamp as described above, the imprint of the notification stamp cannot be uniquely specified.

In view of the above circumstances, it is an object of the present invention to provide a technique capable of extracting the imprint of a notification stamp from an image of a form without the need to predefine the position of the notification stamp column.

One aspect of the program is an image processing program, which is a first detection process for detecting an area of a keyword indicating an item name of a notification stamp in an image of a form, and a second detection process for detecting an area of an imprint in the image of the form. Is executed, and the computer is made to execute the process of specifying the imprint area of the notification stamp in the image of the form based on the execution results of the first detection process and the second detection process. It is a feature.

One aspect of the method is an image processing method, in which a first detection process for detecting an area of a keyword indicating an item name of a notification stamp in an image of a form and a second detection process for detecting an area of an imprint in the image of the form. Is executed, and the region of the imprint of the notification stamp in the image of the form is specified based on the execution results of the first detection process and the second detection process.

One aspect of the device is an image processing device, which is a notification stamp item name detection unit that executes a first detection process for detecting a keyword area indicating a notification stamp item name in a form image, and an image of the form. Based on the imprint detection unit that executes the second detection process that detects the imprint area in the above, and the execution results of the first detection process and the second detection process, the imprint of the notification stamp in the image of the form. It is characterized by including a notification stamp imprint specifying part that specifies an area.

According to the present invention, it is possible to extract the imprint of the notification stamp from the image of the form without the need to predefine the position of the notification stamp column.

It is a figure which exemplifies the structure of the form processing system. It is a figure which illustrates the functional structure of the image processing apparatus which concerns on one Embodiment. It is a flowchart which illustrates the image processing which concerns on one Embodiment. It is a figure which shows typically the specific example of the image processing which concerns on one Embodiment. It is a flowchart which illustrates the details of the process of S5. It is a figure which shows typically the specific example of the process of S5. It is a flowchart which illustrates the details of the process of S51. It is a figure (the 1) which shows typically the specific example of the process of S51. It is a figure (the 2) which shows typically the specific example of the process of S51. It is a flowchart which illustrates the detail of the process of S53. It is a figure which shows typically the specific example of the process of S53. It is a figure which illustrates the hardware structure of a computer. It is a figure which illustrates the notification stamp column with high design. It is a figure which illustrates the imprint other than the notification seal.

Hereinafter, embodiments of the present invention will be described with reference to the drawings.
FIG. 1 is a diagram illustrating the configuration of a form processing system.

The form processing system 1 illustrated in FIG. 1 is a system that handles a form with a notification stamp stamped on it. The form processing system 1 includes an image processing device 2, a scanner device 3, an input device 4, a display device 5, and a server device 6.

The image processing device 2 extracts the imprint of the notification stamp and other necessary information from the image of the form, and electronically collates, for example, the imprint of the extracted notification stamp with the imprint of the previously registered notification stamp. The image processing device 2 acquires, for example, an image of a form from the scanner device 3. The scanner device 3 is a device that converts a paper medium such as a form into digital data (image).

The input device 4 is a device for inputting various information to the image processing device 2. The input device 4 is, for example, a keyboard device. The display device 5 is a device that displays various information such as the content of the processing performed by the image processing device 2 and the result of the processing, the information input by the input device 4, the image of the form, and the like. The display device 5 is, for example, a liquid crystal display.

The server device 6 holds a character recognition dictionary that the image processing device 2 refers to when recognizing characters in an image, and a keyword table that the image processing device 2 refers to when extracting keywords from the character recognition result. It is a device. The server device 6 is communicably connected to the image processing device 2 via the network 7.

FIG. 2 is a diagram illustrating a functional configuration of an image processing device according to an embodiment.
As illustrated in FIG. 2, the image processing device 2 includes a scanner control unit 210, a character recognition processing unit 220, a character string extraction unit 230, an imprint detection unit 240, a notification imprint identification unit 250, and display control. A unit 260 and a main control unit 270 are included. Further, the image processing device 2 includes a storage unit (not shown).

The scanner control unit 210 controls the process of acquiring the image of the form from the scanner device 3.
The character recognition processing unit 220 refers to the character recognition dictionary 610 held by the server device 6 and recognizes the characters included in the image of the form.

The character string extraction unit 230 refers to the keyword table 620 held by the server device 6 and extracts a keyword (character string) indicating an item name from the character recognition result by the character recognition processing unit 220. In the keyword table 620, a character string that can be a keyword indicating an item name is registered in advance. For example, "account number", "account", "account transfer number", etc. are registered as a character string that can be a keyword indicating the item name of the account number, and "notification stamp" is a character string that can be a keyword indicating the item name of the notification stamp. , "Delivery stamp", "Registration stamp", etc. are registered.

Further, the character string extraction unit 230 extracts a character string (for example, "1234567") of the content for a keyword (for example, "account number") indicating an item name from the character recognition result by the character recognition processing unit 220.

In addition, the character string extraction unit 230 includes a notification stamp item name detection unit 231.
The notification stamp item name detection unit 231 executes the first detection process for detecting a keyword area (for example, a rectangular area including the keyword) indicating the item name of the notification stamp in the image of the form. The first detection process is performed using, for example, existing technology.

The imprint detection unit 240 executes a second detection process for detecting an imprint region (for example, a rectangular region including the imprint) in the image of the form. The second detection process is performed by using, for example, an object detection AI technique such as deep learning or a date stamp region specifying technique described in the patent application of Japanese Patent Application No. 2018-091876 by the applicant of the present application.

The notification stamp imprint identification unit 250 determines the area of the notification stamp imprint in the image of the form based on the execution results of the first detection process by the notification mark item name detection unit 231 and the second detection process by the imprint detection unit 240. Identify. For example, by executing the first detection process by the notification stamp item name detection unit 231 and the second detection process by the imprint detection unit 240, one or more keyword regions indicating the item names of the notification stamp are detected and the imprint region. When one or more are detected, the imprint area of the notification stamp is specified based on the positional relationship between the keyword area indicating the item name of the notification stamp and the imprint area. In the specification in this case, for example, the area of the imprint that satisfies the predetermined condition is specified as the area of the imprint of the notification stamp. The predetermined condition is, for example, the condition that the item exists on the right side or the lower side of the keyword area indicating the item name of the notification stamp and the distance from the keyword area is equal to or less than the predetermined value, or the keyword indicating the item name of the notification mark. It is a condition that it exists on the right side or the lower side of the area, the distance from the keyword area is equal to or less than a predetermined value, and the area size is larger than the predetermined size. Further, for example, when only one imprint area is detected by executing the first detection process by the notification stamp item name detection unit 231 and the second detection process by the imprint detection unit 240, the imprint area is used. , Specify as the area of the imprint of the notification stamp.

The display control unit 260 controls the display of the display device 5. For example, the display control unit 260 generates screen data including the result of processing on the image of the form, and displays (outputs) the screen data on the display device 5.

The main control unit 270 controls a series of processes performed by the scanner control unit 210, the character recognition processing unit 220, the character string extraction unit 230, the imprint detection unit 240, the notification imprint identification unit 250, and the display control unit 260.

FIG. 3 is a flowchart illustrating image processing according to the embodiment. FIG. 4 is a diagram schematically showing a specific example of image processing according to the embodiment.

When the image processing device 2 starts image processing for extracting necessary information from the form, it first acquires an image of the form (S1), as illustrated in FIG. The scanner control unit 210 performs the processing of S1.

Next, the image processing device 2 refers to the character recognition dictionary 610 held by the server device 6 and recognizes the characters included in the image of the form (S2). The processing of S2 is performed by the character recognition processing unit 220.

Next, the image processing device 2 refers to the keyword table 620 held by the server device 6, extracts a keyword indicating the item name from the character recognition result, and indicates the item name of the notification mark in the image of the form. The first detection process for detecting the keyword region is executed (S3). The processing of S3 is performed by the character string extraction unit 230, and in particular, the first detection processing of the processing of S3 is performed by the notification stamp item name detection unit 231 included in the character string extraction unit 230.

Next, the image processing device 2 extracts a character string of the content for the keyword indicating the item name from the character recognition result (S4). The processing of S4 is performed by the character string extraction unit 230.

Next, the image processing device 2 executes a second detection process for detecting the imprint region in the image of the form, and based on the execution results of the first detection process and the second detection process, the image of the form. The area of the imprint of the notification stamp in (S5) is specified. The processing of S5 is performed by the imprint detection unit 240 and the notification imprint identification unit 250.

Next, the image processing device 2 causes the display device 5 to display the processing result for each item name (S6). For example, as the processing result of the item name of the notification stamp, the coordinates indicating the area of the imprint of the specified notification stamp (for example, the coordinates of two diagonal points of the rectangular area including the imprint) are displayed, and the items other than the notification stamp are displayed. As the processing result of the name, the character string of the content for the keyword indicating the item name and the coordinates indicating the area of the character string (for example, the coordinates of two diagonal points of the rectangular area including the character string) are displayed. The processing of S6 is performed by the display control unit 260.

According to such image processing, for example, in the processing of S1, the image 8 of the form as schematically shown in FIG. 4A is acquired. Image 8 of the form shows the area 801 of the keyword (“account number”) indicating the item name of the account number, the area 802 of the content character string (“12344567”) for the keyword indicating the item name of the account number, and the notification stamp. Areas 803, 804, 805 of the keyword (“notification stamp”) indicating the item name of the above, the area 806 of the imprint stamped with the notification stamp, and the area of the imprint stamped with a mark other than the notification stamp (for example, a correction stamp). 807 and include. In the keyword areas 803, 804, and 805 indicating the item name of the notification stamp, the area 803 indicates the area of the keyword indicating the item name of the original notification stamp, and the areas 804 and 805 are the notification stamps included in the explanatory text and the like. Indicates the area of the keyword indicating the item name of.

After that, in the process of S3, as schematically shown in FIG. 4B, the character string (“account number”) included in the area 801 in the image 8 of the form is a keyword indicating the item name of the account number. The character string (“notification stamp”) included in the areas 803, 804, and 805 is extracted as a keyword indicating the item name of the notification mark, and the areas 803, 804, and 805 indicate the item name of the notification mark. Detected as a keyword area.

After that, in the processing of S4 and S5, as schematically shown in FIG. 4 (c), the character string (“12345667”) included in the area 802 indicates the item name of the account number in the image 8 of the form. It is extracted as a character string of the content for the keyword, and the area 806 is specified as the area of the imprint of the notification stamp.

After that, in the processing of S6, as shown in FIG. 4D, the processing result (contents, coordinates) for each item name is displayed on the display device 5. For example, as a result of processing the item name of the account number, the character string ("1234567") of the content for the keyword indicating the item name of the account number and the coordinates of the area of the character string ("(x1, y1, x2, y2)). ") Is displayed, and as a result of processing the item name of the notification stamp, the coordinates of the area of the imprint of the notification stamp ("(x3, y3, x4, y4) ") are displayed. Since the character string of the content for the keyword indicating the item name of the notification stamp does not originally exist, it is not included in the processing result of the item name of the notification stamp.

FIG. 5 is a flowchart illustrating the details of the process of S5. FIG. 6 is a diagram schematically showing a specific example of the processing of S5.

As illustrated in FIG. 5, when the process of S5 is started, the imprint detection unit 240 first executes a second detection process of detecting the imprint area in the image of the form (S51). The imprint area detected by the process of S51 is also a candidate for the imprint area of the notification stamp in the image of the form.

Next, the notification imprint identification unit 250 determines whether or not a plurality of imprint regions have been detected by executing the second detection process by the imprint detection unit 240 (S52).

When the determination result of S52 is YES, the notification stamp imprint identification unit 250 determines the form based on the execution results of the first detection process by the notification mark item name detection unit 231 and the second detection process by the imprint detection unit 240. The region of the imprint of the notification stamp in the image is specified (S53), and the coordinates indicating the region of the imprint of the identified notification stamp are output to the main control unit 270 (S54). However, if the area of the imprint of the notification stamp cannot be specified by the processing of S53, the coordinates are not output.

On the other hand, when the determination result of S52 is NO, the notification imprint identification unit 250 has 1 or 0 imprint regions detected by the execution of the second detection process by the imprint detection unit 240. The next process is performed according to the result (S54). When the number of detected imprint areas is one, the imprint area is specified as the imprint area of the notification mark in the image of the form, and the coordinates indicating the imprint area of the specified notification mark are mainly controlled. Output to unit 270. On the other hand, when the number of detected imprint areas is 0, it is assumed that there is no imprint area of the notification stamp in the image of the form, and the coordinates are not output.

According to the process illustrated in FIG. 5, for example, in the process of S51, the areas 806 and 807 in the image 8 of the form are detected as the imprint area, as schematically shown in FIG. 6A. Will be done.

After that, in the process of S53, as schematically shown in FIG. 6B, the area 806 in the image 8 of the form is specified as the area of the imprint of the notification stamp.

After that, in the process of S54, as shown in FIG. 6C, the coordinates (“(x3, y3, x4, y4)” indicating the region of the imprint of the specified notification stamp) are output to the main control unit 270. Will be done.

FIG. 7 is a flowchart illustrating the details of the process of S51. 8 and 9 are diagrams schematically showing a specific example of the processing of S51.

As described above, the imprint detection unit 240 that performs the processing of S51 describes the second detection processing in the object detection AI technology such as deep learning and the patent application of Japanese Patent Application No. 2018-091876 by the applicant of the present application. This is done by using the date stamp area identification technology of. Here, a case where the object detection AI technology such as deep learning is used will be described as an example.

As illustrated in FIG. 7, when the processing of S51 is started, the imprint detection unit 240 first resizes the image of the form so as to fit the trained neural network model (S511). The trained neural network model outputs the coordinates indicating the region of the object in the image and the feature quantity indicating the imprint of the object in response to the input of the image of a predetermined size. The area of an object is an area determined as a group of objects.

Next, the imprint detection unit 240 reads the trained neural network model (S512). The trained neural network model can be managed by the storage unit (not shown) included in the image processing device 2 or the server device 6.

Next, the imprint detection unit 240 inputs the size-converted image of the form into the read neural network model, and acquires the coordinates indicating the area of the object which is the output and the feature amount indicating the imprint-likeness (S513). ).

Next, the imprint detection unit 240 uses the acquired coordinates indicating the area of the object and the feature amount indicating the imprint-likeness to obtain the coordinates indicating the area of the object in which the feature amount representing the imprint-likeness exceeds a predetermined value. It is determined as the coordinates to be shown (S514).

Next, the imprint detection unit 240 resizes the image of the converted form to the original size, and resizes the coordinates indicating the area of the object determined as the coordinates indicating the imprint area to the original size. It is converted into the coordinates in the image of (S515).

Next, the imprint detection unit 240 outputs the converted coordinates to the main control unit 270 as coordinates indicating the imprint region (candidate for the imprint region of the notification stamp) in the image of the form (S516).

In this way, the imprint detection unit 240 detects the imprint area in the image of the form and outputs the coordinates indicating the area.

According to the process illustrated in FIG. 7, for example, in the process of S511, the image 8 of the form has a size suitable for the trained neural network model, as schematically shown in FIG. 8A. Here, the size is converted to 512 pixels × 512 pixels in size).

After that, in the process of S513, as schematically shown in FIG. 8B, the image 8a of the form after size conversion is input to the trained neural network model 9, and the area of the object which is the output thereof is input. The coordinates to be shown and the feature amount representing the imprint-likeness are acquired. In FIG. 8B, the coordinates of the regions 808, 809, 810, 811, 812, 813, and 814 are acquired as the coordinates of the region of the object. Further, although omitted in FIG. 8B, by performing the processing of S513, a feature amount representing the imprint-likeness of the objects in each of the areas 808, 809, 810, 811, 812, 813, and 814 Is obtained.

After that, in the processing of S514, as schematically shown in FIG. 9A, the coordinates indicating the regions 811 and 812 of the object whose feature amount representing imprint-likeness exceeds a predetermined value (0.9 in this case) are set. It is determined as the coordinates indicating the area of the imprint. The value of the feature amount is a value that approaches 1 as the object resembles the imprint.

After that, in the process of S515, as schematically shown in FIG. 9B, the size-converted form image 8a is resized to the original size, and the object is determined as the coordinates indicating the imprint area. The coordinates indicating the areas 811 and 812 are converted into the coordinates in the image 8 of the form whose size has been converted to the original size. As a result, the coordinates indicating the areas 806 and 807 in the image 8 of the form are acquired as the coordinates after the conversion.

After that, in the processing of S516, as shown in (c) of FIG. 9, the coordinates indicating the region 806 (“(x3, y3, x4, y4)”) and the coordinates indicating the region 807, which are the coordinates after the conversion, are used. "(X5, y5, x6, y6)") is output to the main control unit 270 as coordinates indicating the imprint area (candidate for the imprint area of the notification stamp) in the image 8 of the form.

FIG. 10 is a flowchart illustrating the details of the process of S53. FIG. 11 is a diagram schematically showing a specific example of the processing of S53.

As illustrated in FIG. 10, when the processing of S53 is started, the notification stamp imprint specifying unit 250 first indicates a keyword area indicating the item name of the notification stamp detected by the notification stamp item name detection unit 231. The coordinates and the coordinates indicating the area of the imprint detected by the imprint detection unit 240 are mapped to the coordinate system corresponding to the image of the form (S531). That is, the area of the keyword indicating the item name of the notification stamp and the area of the imprint are mapped.

Next, the notification stamp imprint specifying unit 250 identifies the imprint area of the notification stamp based on the positional relationship between the keyword area indicating the item name of the mapped notification stamp and the imprint area (S532). In this identification, for example, an imprint area satisfying a predetermined condition is specified as an imprint area of a notification stamp. The predetermined condition is, for example, the condition that the item exists on the right side or the lower side of the keyword area indicating the item name of the notification stamp and the distance from the keyword area is equal to or less than the predetermined value, or the keyword indicating the item name of the notification mark. It is a condition that it exists on the right side or the lower side of the area, the distance from the keyword area is equal to or less than a predetermined value, and the area size is larger than the predetermined size.

Next, the notification stamp imprint specifying unit 250 outputs the coordinates indicating the region of the specified notification stamp imprint in the image of the form to the main control unit 270 (S533).

According to the process illustrated in FIG. 10, for example, in the process of S531, as shown schematically in (a) of FIG. 11, the item of the notification stamp is set in the coordinate system 10 corresponding to the image 8 of the form. The keyword areas 101, 102, 103 indicating the name and the imprint areas 104, 105 are mapped.

After that, in the process of S532, among the imprint areas 104 and 105, the imprint area 104 satisfying a predetermined condition as shown in FIG. 11B is specified as the imprint area of the notification stamp. The predetermined condition shown in FIG. 11B exists on the right side or the lower side of the keyword area indicating the item name of the notification stamp, and the distance from the keyword area indicating the item name of the notification stamp is equal to or less than the predetermined value. (Here, 1 cm or less), and the area size is larger than the predetermined size (here, the size of 2.4 cm × 2.4 cm). The distance from the keyword area indicating the item name of the notification stamp is equal to or less than the predetermined value. For example, the distance between the reference point of the keyword area indicating the item name of the notification stamp and the reference point of the imprint area is a predetermined value. It is as follows. The reference point is, for example, the center point of the region.

After that, in the process of S533, as shown in (c) of FIG. 11, the coordinates ("(x3, y3, x4) indicating the region of the imprint of the specified notification stamp (here, region 104 (corresponding to region 806))). , Y4) ”) is output to the main control unit 270.

As described above, according to the image processing performed by the image processing device 2, a process of detecting the keyword area indicating the item name of the notification stamp and the imprint area in the image of the form is executed, and based on the processing result. Since the area of the imprint of the notification stamp in the image of the form is specified, it is not necessary to predefine the position of the notification stamp column, and the imprint of the notification stamp can be extracted. In addition to the imprint of the notification stamp, the area of the imprint of the notification stamp can be specified for the image of the form including the imprint of the correction stamp, the stamp, the sample stamp, etc., and the imprint of the notification stamp can be extracted. become. Furthermore, for the image of the form including the notification stamp column with high design, even if the keyword area indicating the item name of the notification stamp is not detected, if only one imprint area is detected, the imprint is found. Since the area of is specified as the area of the imprint of the notification stamp, the imprint of the notification stamp can be extracted.

The functional configuration of the image processing device 2 is not limited to the configuration illustrated in FIG. 2, and can be appropriately changed. For example, one or both of the character recognition dictionary 610 referred to by the character recognition processing unit 220 and the keyword table 620 referred to by the character string extraction unit 230 can be managed by a storage unit (not shown) included in the image processing device 2. Is. Further, the image processing device 2 may acquire an image of a form provided via a portable recording medium such as a memory card or an optical disk, or a network, and perform the above image processing.

In addition, the image processing device 2 can be realized by a computer and a program executed by the computer. Hereinafter, the image processing device 2 realized by the computer and the program will be described with reference to FIG.

FIG. 12 is a diagram illustrating a hardware configuration of a computer.
As illustrated in FIG. 12, the computer 11 includes a CPU (Central Processing Unit) 1101, a memory 1102, an auxiliary storage device 1103, an input device 1104, and a display device 1105. Further, the computer 11 includes an input / output interface 1106, a communication control device 1107, and a medium drive device 1108. These elements 1101 to 1108 in the computer 11 are connected to each other by a bus 1109, and data can be exchanged between the elements.

The CPU 1101 controls the overall operation of the computer 11 by executing various programs including an operating system. For example, the CPU 1101 executes an image processing program including each process described with reference to FIGS. 3 to 11.

The memory 1102 includes a ROM (Read Only Memory) and a RAM (Random Access Memory) (not shown). In the ROM of the memory 1102, for example, a predetermined basic control program or the like read by the CPU 1101 when the computer 11 is started is recorded in advance. Further, the RAM of the memory 1102 is used as a work storage area by the CPU 1101 as needed when executing various programs. The RAM of the memory 1102 can be used, for example, for temporary storage of a form image, coordinates indicating a keyword area indicating an item name of a notification stamp, coordinates indicating an imprint area, and the like.

The auxiliary storage device 1103 is, for example, a magnetic disk such as an HDD (Hard Disk Drive) and a non-volatile memory such as a flash memory. The auxiliary storage device 1103 can store various programs, various data, and the like executed by the CPU 1101. The auxiliary storage device 1103 can be used for storing, for example, the image processing program described above. Further, the auxiliary storage device 1103 can be used for storing, for example, an image of a form, a processing result for each item name, and the like.

The input device 1104 is, for example, a keyboard device, a mouse device, a touch panel device, and the like. When the operator of the computer 11 performs a predetermined operation on the input device 1104, the input device 1104 transmits the input information associated with the operation content to the CPU 1101. The input device 1104 corresponds to the input device 4 in the form processing system 1 of FIG.

The display device 1105 is, for example, a liquid crystal display device. The display device 1105 can be used, for example, to display the operating state of the computer 11, the image of the form, the processing result for each item name, and the like. The display device 1105 corresponds to the display device 5 in the form processing system 1 of FIG.

The input / output interface 1106 connects the computer 11 to electronic components, other electronic devices (for example, a scanner device 3), and the like. The input / output interface 1106 can also be used for connecting a flash memory provided with a USB (Universal Serial Bus) standard connector, for example.

The communication control device 1107 is a device that connects the computer 11 to a network and controls various communications between the computer 11 and other electronic devices via the network. The communication control device 1107 can be used, for example, to refer to the character recognition dictionary or keyword table held by the server device 6 in the form processing system 1 of FIG.

The medium drive device 1108 reads out the programs and data recorded in the portable recording medium 12, and writes the data stored in the auxiliary storage device 1103 to the portable recording medium 12. As the portable recording medium 12, for example, there is a Secure Digital (SD) standard memory card (flash memory). The portable recording medium 12 can be used for storing the above-mentioned image processing program, an image of a form, a processing result for each item name, and the like. When the computer 11 is equipped with an optical disk drive that can be used as the medium drive device 1108, various optical disks that can be recognized by the optical disk drive can be used as the portable recording medium 12. Optical discs that can be used as the portable recording medium 12 include, for example, Compact Disc (CD), Digital Versatile Disc (DVD), Blu-ray Disc (Blu-ray is a registered trademark), and the like.

When the operator operates the input device 1104 to input the start command of the image processing program, the computer 11 reads the above-mentioned image processing program from the auxiliary storage device 1103 or the like and executes the computer 11. While executing the image processing program, the CPU 1101 includes a character recognition processing unit 220 in the image processing device 2 illustrated in FIG. 2, a character string extraction unit 230 including a notification stamp item name detection unit 231, a stamp imprint detection unit 240, and a notification. It functions (operates) as a stamp imprint specifying unit 250 and a display control unit 260. Further, when the image of the form is acquired from the scanner device 3, the CPU 1101 and the input / output interface 1106 function as the scanner control unit 210 in the image processing device 2 illustrated in FIG. Further, when characters are recognized by referring to the character recognition dictionary 610 held by the server device 6 or the like other than the computer 11, the communication control device 1107 takes a part of the functions of the character recognition processing unit 220. Further, when the keyword is extracted by referring to the keyword table 620 held by the server device 6 or the like other than the computer 11, the communication control device 1107 takes a part of the function of the character string extraction unit 230.

The computer 11 operated as the image processing device 2 does not need to include all of the elements 1101 to 1108 illustrated in FIG. 12, and some elements may be omitted depending on the application and conditions. For example, in the computer 11, the medium driving device 1108 and the communication control device 1107 may be omitted.

Although the embodiments of the present invention have been described above, the present invention is not limited to the above-described embodiments, and various improvements and changes can be made without departing from the gist of the present invention.

1 Form processing system 2 Image processing device 3 Scanner device 4 Input device 5 Display device 6 Server device 7 Network 8, 8a Image 9 Neural network model 10 Coordinate system 11 Computer 12 Portable recording medium 101, 102, 103, 104, 105 areas 210 Scanner control unit 220 Character recognition processing unit 230 Character string extraction unit 231 Notification mark item name detection unit 240 Imprint detection unit 250 Notification mark imprint identification unit 260 Display control unit 270 Main control unit 610 Character recognition dictionary 620 Keyword table 801, 802, 803, 804, 805 area 806, 807, 808, 809, 810 area 811, 812, 813, 814 area 1101 CPU
1102 Memory 1103 Auxiliary storage 1104 Input device 1105 Display device 1106 Input / output interface 1107 Communication control device 1108 Media drive device 1109 Bus

Claims (8)

The first detection process for detecting the area of the keyword indicating the item name of the notification stamp in the image of the form and the second detection process for detecting the area of the imprint on the image of the form are executed.
An image processing program characterized in that a computer is made to execute a process of specifying an area of an imprint of a notification stamp in an image of the form based on the execution results of the first detection process and the second detection process. The specific processing is performed when one or more regions of the keyword are detected and one or more regions of the imprint are detected by executing the first detection process and the second detection process. The image processing program according to claim 1, wherein the area of the imprint of the notification stamp is specified based on the positional relationship between the area and the area of the imprint. The image processing program according to claim 2, wherein the specific processing specifies the region of the imprint that satisfies a predetermined condition as the region of the imprint of the notification stamp. The image processing program according to claim 3, wherein the predetermined condition is a condition that exists on the right side or the lower side of the keyword region and the distance from the keyword region is equal to or less than a predetermined value. The predetermined condition is a condition that the keyword exists on the right side or the lower side of the keyword region, the distance from the keyword region is equal to or less than a predetermined value, and the region size is larger than the predetermined size. 3 The image processing program described. In the specific processing, when only one imprint area is detected by executing the first detection process and the second detection process, the imprint area is used as the imprint area of the notification mark. The image processing program according to claim 1, wherein the image processing program is specified. The first detection process for detecting the area of the keyword indicating the item name of the notification stamp in the image of the form and the second detection process for detecting the area of the imprint on the image of the form are executed.
An image processing method characterized in that an area of an imprint of a notification stamp in an image of the form is specified based on the execution results of the first detection process and the second detection process. The notification stamp item name detection unit that executes the first detection process that detects the area of the keyword indicating the item name of the notification stamp in the image of the form,
An imprint detection unit that executes a second detection process for detecting an imprint area in the image of the form, and
An image processing apparatus including a notification stamp imprint specifying unit that specifies an area of a notification stamp imprint in the image of the form based on the execution results of the first detection process and the second detection process. ..

Images