JP2022020412A - Image processing device and information processing system - Google Patents

Abstract

To read an image of a document, to classify the image of the document on the basis of a feature of the entire image of the document, and to store image data showing the image of the document in a folder corresponding to specified classification.SOLUTION: An image processing device 10 is provided with a control part 31 for extracting a feature of an image of an acquired document, and collecting data showing the feature of the image of the document as learning data to discriminate classification of the document in the case of a training mode, and extracting the feature of the image of the document read by an image reading part 11, discriminating the classification of the document on the basis of the feature of the image of the document and the learning data, adding information showing the discriminated classification to the image data of the document, confirming the existence/nonexistence of a destination folder of a name matching the discriminated classification in a folder of a predetermined storage destination, and storing the image data of the document in the destination folder of a matching name in the case of a classification mode.SELECTED DRAWING: Figure 2

Description

The present invention relates to an image processing device and an information processing system for acquiring image data of a manuscript, and more particularly to a technique for classifying and storing image data of a manuscript.

The image processing device includes a multifunction device (MFP), a copier, a facsimile machine, and the like. In such an image processing apparatus, it is desirable that the image data of the scanned document can be automatically classified.

Japanese Unexamined Patent Publication No. 2001-101213

In the document management device described in Patent Document 1, it is necessary for the user to set in advance whether the classification of the image data of the read document is a private document or a public document. Therefore, it is not possible to automatically set the classification (private document or public document) of the image data of the read document.

Further, the document management device described in Patent Document 1 classifies the image data of the read document into "private document" or "public document" as a major classification according to the above setting by the user, and further, it is read. Layout analysis is performed on the image data of the document, and based on the layout analysis result, it is specified whether the classification is "photo document", "office document" or "information gathering document" as the middle classification. The image data of the read document is stored in the folder corresponding to the specified classification. As described above, since the image data of the read document is classified based on the layout of the document image, it is not classified based on the content or the feature of the entire document image. Therefore, it cannot be said that the document images are properly classified.

The present invention has been made in view of the above circumstances, and makes it possible to specify the classification of the image of the manuscript based on the characteristics of the entire image of the manuscript. The purpose is to be able to save image data indicating an image.

The image processing apparatus according to one aspect of the present invention has an image acquisition unit for acquiring image data of a document, an operation unit for inputting settings and instructions by user operation, and a training mode according to instructions input to the operation unit. Alternatively, the control unit includes a control unit for setting the classification mode, and the control unit is an image of a document whose image data is acquired by the image acquisition unit when the training mode is set by operating the operation unit. When the characteristics of the original are extracted, data showing the characteristics of the image of the original is collected as learning data for determining the classification of the original, and the classification mode is set by the operation to the operation unit. , The characteristics of the image of the manuscript whose image data has been acquired by the image acquisition unit are extracted, the classification of the manuscript is discriminated based on the characteristics of the image of the manuscript and the learning data, and the information indicating the discriminated classification is obtained. It is added to the image data of the manuscript, and it is confirmed whether there is a destination folder with a name that matches the determined classification in the predetermined storage destination folder, and if there is no destination folder with the matching name, the match is made. A destination folder with a name is created, and the image data of the manuscript is stored in the destination folder.

Further, the information processing system according to one aspect of the present invention is an information processing system including the above image processing device and the information processing device, and the image processing device and the information processing device can be connected to each other via a network. Therefore, the information processing device includes a storage unit having a shared folder as a predetermined storage destination folder, and the control unit of the image processing device inputs information for establishing a connection to the operation unit. Establishes a connection state to the information processing device, makes it possible to access the shared folder of the storage unit of the information processing device, and a destination having a name matching the determined classification in the shared folder. The existence of the folder is confirmed, and if the destination folder with the matching name does not exist, the destination folder with the matching name is created, and the image data of the manuscript is stored in the destination folder.

According to the present invention, the image of the manuscript can be read and the image of the manuscript can be classified based on the characteristics of the entire image of the manuscript, and the image data showing the image of the manuscript can be stored in the folder corresponding to the specified classification. Can be saved.

It is sectional drawing which shows the image processing apparatus of 1st Embodiment of this invention. It is a block diagram which shows the main internal structure of the image processing apparatus of 1st Embodiment. It is a flowchart which shows the processing procedure of the training mode for generating the learning data of deep learning. It is a figure which shows the classification function input screen displayed on the display part. It is a figure which shows an example of the training mode setting screen displayed on the display part. It is a figure which shows the concept of deep learning. It is a chart exemplifying the correspondence between the hue, contrast, the number of colors, and the density of the image, which are the characteristics of the image of the manuscript, and the invitation letter and the receipt, which are the tag names. It is a flowchart which shows the processing procedure in a classification mode. It is a figure which shows the classification mode setting screen displayed on the display part. It is a figure which shows the pop-up screen displayed on the display part. It is a figure which shows the other pop-up screens displayed on the display part. It is a figure which shows the property of the image of the original which was displayed on the display part. It is a figure which shows the other property of the image of a manuscript displayed on the display part. It is a figure which shows the list of the file displayed on the display part. It is a flowchart which shows an example of the manuscript image management processing of 1st Embodiment. It is a figure which shows an example of the setting screen of the management by classification displayed on the display part. It is a figure for demonstrating pixel matching. It is a figure which shows the formula for calculating a correlation value using pixel matching, and the calculation example of a correlation value. It is a figure which shows an example of saving the manuscript data when the folder of the name which matches the classification of the scanned manuscript does not exist in the shared folder. It is a figure which shows an example of saving the manuscript data when the folder of the name which matches the classification of the scanned manuscript is in a shared folder. It is a figure which shows an example of saving the manuscript data when the folder of the name matching the classification of the manuscript read this time is not in a shared folder, and there is a folder of another name. It is a figure which shows an example of saving the manuscript data when a folder of a name matching the classification of a scanned manuscript is in a shared folder, and a plurality of image data are saved in the folder of the said name. It is a block diagram which shows the main internal structure of the information processing system of 2nd Embodiment. It is a flowchart which shows an example of the manuscript image management processing of 2nd Embodiment. It is a figure which shows an example of the transmission setting screen displayed on the display part.

Hereinafter, the image processing apparatus according to the embodiment of the present invention will be described with reference to the drawings.

FIG. 1 is a cross-sectional view showing an image processing apparatus according to the first embodiment of the present invention. As shown in FIG. 1, the image processing apparatus 10 of the present embodiment is an MFP (image forming apparatus) having a plurality of functions such as a copy function, a printer function, and a facsimile function. The image processing device 10 includes an image reading unit 11 and an image forming unit 12.

The image reading unit 11 has an image pickup element that optically reads the image of the document, and the analog output of the image pickup element is converted into a digital signal to generate image data indicating the image of the document. The image reading unit 11 is an example of an image acquisition unit within the scope of claims.

The image forming unit 12 forms the image shown by the image data on the recording paper, and is used for the image forming unit 3M for magenta, the image forming unit 3C for cyan, the image forming unit 3Y for yellow, and the black. It is equipped with an image forming unit 3Bk. In each of the image forming units 3M, 3C, 3Y, and 3Bk, the surface of the photoconductor drum 4 is uniformly charged, the surface of the photoconductor drum 4 is exposed, and an electrostatic latent image is formed on the surface of the photoconductor drum 4. Is formed, the electrostatic latent image on the surface of the photoconductor drum 4 is developed into a toner image, and the toner image on the surface of the photoconductor drum 4 is primarily transferred to the intermediate transfer belt 5. As a result, a color toner image is formed on the intermediate transfer belt 5. The toner image of this color is secondarily transferred to the recording paper P conveyed from the paper feed unit 14 through the transfer path 8 in the nip area N between the intermediate transfer belt 5 and the secondary transfer roller 6.

After that, the recording paper P is heated and pressurized by the fixing device 15, the toner image on the recording paper P is fixed by thermocompression bonding, and the recording paper P is discharged to the discharge tray 17 through the discharge roller 16.

Next, FIG. 2 is a block diagram showing a main internal configuration of the image processing apparatus 10 of the present embodiment. As shown in FIG. 2, the image processing device 10 of the present embodiment includes an image reading unit 11, an image forming unit 12, a display unit 21, an operation unit 22, a touch panel 23, a network interface unit 91, and an image memory. 26, an image processing unit 27, a storage unit 28, and a control unit 29 are provided. These components are capable of transmitting and receiving data or signals through the bus to each other.

The display unit 21 is a display device such as a liquid crystal display (LCD) or an organic EL (OLED: Organic Light-Emitting Diode) display. The operation unit 22 includes physical keys such as a numeric keypad, an enter key, and a start key.

A touch panel 23 is arranged on the screen of the display unit 21. The touch panel 23 is a touch panel of a so-called resistance film method or a capacitance method, and detects contact (touch) of a user's finger or the like with the touch panel 23 together with the contact position, and a detection signal indicating the coordinates of the contact position. Is output to the control unit 31 or the like, which will be described later, of the control unit 29.

The network interface unit 91 transmits and receives various data to and from an external device such as a terminal device or a server in the local area or on the Internet. The network interface unit 91 is an example of an image acquisition unit within the scope of claims. The image processing apparatus according to the embodiment of the present invention includes at least one of an image reading unit 11 and a network interface unit 91.

The image memory 26 stores image data indicating an image of a document read by the image reading unit 11. Further, the control unit 31 functions as an image processing unit 27. The image processing unit 27 performs various image processing such as shading correction on the image data in the image memory 26.

The image processing unit 27 may be provided separately from the control unit 31 and the control unit 29. In this case, the control unit 31 and the image processing unit 27 are examples of the control unit within the scope of the claims.

The storage unit 28 is a large-capacity storage device such as an SSD (Solid State Drive) or an HDD (Hard Disk Drive), and stores various application programs and various data.

The control unit 29 is composed of a processor, a RAM (Random Access Memory), a ROM (Read Only Memory), and the like. The processor is, for example, a CPU (Central Processing Unit), an ASIC (Application Specific Integrated Circuit), an MPU (Micro Processing Unit), or the like. The control unit 29 functions as the control unit 31 when the control program stored in the ROM or the storage unit 28 is executed by the processor.

The control unit 31 comprehensively controls the image processing device 10. The control unit 29 is connected to an image reading unit 11, an image forming unit 12, a display unit 21, an operation unit 22, a touch panel 23, a network interface unit 91, an image memory 26, an image processing unit 27, a storage unit 28, and the like. .. The control unit 31 controls the operation of these components and transmits / receives signals or data to / from each component.

The control unit 31 serves as a processing unit that executes various processes necessary for image formation by the image processing device 10. Further, the control unit 31 receives an operation instruction input by the user based on the detection signal output from the touch panel 23 or the operation of the physical key of the operation unit 22. Further, the control unit 31 has a function of controlling the display operation of the display unit 21 and a function of controlling the communication operation of the network interface unit 91.

In the image processing device 10 having such a configuration, for example, when the user sets the document in the image reading unit 11 and operates the operation unit 22 or the touch panel 23 to instruct the execution of copying, the control unit 31 displays the image. The image of the original is read by the reading unit 11, and the image data indicating the image of the original is stored in the image memory 26. The image processing unit 27 performs various processes on the image data. The image forming unit 12 acquires the image data from the image memory 26 and forms the image of the original document indicated by the image data on the recording paper.

Further, by the user's operation, the tag name indicating the type (that is, classification) of the original, the setting of the training mode, and the execution instruction of the image reading unit 11 to read the image of the original are input to the operation unit 22 or the touch panel 23. When the control unit 31 receives the training mode setting, the tag name, and the reading execution instruction, the control unit 31 sets the image processing device 10 to the training mode, and the image reading unit 11 reads the image of the original. , The image data is stored in the image memory 26, and the image data is processed by the image processing unit 27. Further, the control unit 31 executes a process of extracting the features of the image of the read original based on the image data stored in the image memory 26. Then, the control unit 31 collects the data indicating the characteristics of the image of the manuscript extracted in this way as learning data for discriminating the classification of the manuscript indicated by the tag name.

After that, the user sets a further document in the image reading unit 11, and the user operates the operation unit 22 or the touch panel 23 to set the classification mode and execute the image reading unit 11 to read the image of the document. When the instruction is received, the control unit 31 sets the classification mode, the image reading unit 11 reads the image of the original, the image data is stored in the image memory 26, and the image processing unit 27 processes the image data. To do. After that, the control unit 31 further executes a process of extracting the features of the image of the manuscript based on the image data in the image memory 26, and collects the features of the image of the manuscript and as described above. Based on the learning data, the tag name indicating the classification of the manuscript is determined, and the information indicating the determined tag name is added to the image data of the manuscript. That is, the control unit 31 classifies the other manuscript by adding the information indicating the tag name in this way.

Therefore, in the state where the control unit 31 is set to the training mode, the data indicating the characteristics of the image of the document read by the image reading unit 11 is the document indicated by the tag name input to the operation unit 22 or the touch panel 23. It is collected as learning data for discriminating as a classification of. Further, in the state where the control unit 31 is set to the classification mode, the characteristics of the image of the original document extracted from the image of the original document read by the image reading unit 11 and the learning data collected at this time are obtained. By comparing, the tag name associated with the learning data indicating the feature that matches or approximates the feature of the image of the manuscript is determined. Then, the control unit 31 classifies the image of the scanned document by adding the information indicating the tag name to the image of the scanned document. The control unit 31 collects data indicating the characteristics of the image of the original document read by the image reading unit 11 as learning data for discriminating the tag name even in the state of being set to the classification mode.

Next, the processing procedure in the training mode will be described in detail with reference to the flowchart shown in FIG.

First, the administrator (specific user) operates the operation unit 22 or the touch panel 23 to input an instruction to execute the document classification function. When the operation unit 22 or the touch panel 23 receives the input of the instruction, the control unit 31 causes the display unit 21 to display the classification function input screen G1 as shown in FIG. 4 in response to the instruction (S101). In the classification function input screen G1 shown in FIG. 4, a key K11 for inputting a training mode execution instruction, a key K12 for inputting a classification mode execution instruction, and a text box for inputting an identification information ID are entered. TB1 is displayed.

By the operation of the operation unit 22 or the touch panel 23 by the administrator, the identification information ID of the administrator is input to the text box TB1, and when the key K11 of the training mode is pressed, the control unit 31 is input to the text box TB1. It authenticates whether or not the identification information ID matches the identification information ID of the administrator (S102). When the control unit 31 determines that the identification information ID input to the text box TB1 matches the identification information ID of the administrator (YES in S102), the image processing device 10 is set to the training mode (S103). When the control unit 31 determines that the identification information ID input to the text box TB1 does not match the identification information ID of the administrator (NO in S102), the control unit 31 does not set the training mode. The process ends.

When the administrator sets the document in the image reading unit 11 and operates the operation unit 22 or the touch panel 23 in the state where the training mode is set in this way, the control unit 31 inputs a scan execution instruction. According to the input scan execution instruction, the image reading unit 11 reads the image of the original, and the image data indicating the image of the original output from the image reading unit 11 is stored in the image memory 26. The image processing unit 27 performs various processes on the image data (S104). Various processes by the image processing unit 27 in this training mode are preliminary processes performed in order to accurately extract the features of the image of the original in S109 described later, and are image density, contrast, edges, and corners. It is a process to emphasize these.

After that, the control unit 31 causes the display unit 21 to display the training mode setting screen G2 as shown in FIG. 5 (S105). On the training mode setting screen G2 shown in FIG. 5, the image D1 of the document read by the image reading unit 11, the respective keys K21, K22, K23, K24 associated with each tag name already set, are displayed. K25 ..., a text box TB2 for entering a tag name indicating a new document classification, and an OK key K20 are displayed.

When the administrator confirms the image D1 of the manuscript displayed on the display unit 21 and determines that any of the tag names indicated by the keys K21, K22, ... Indicates the classification of the manuscript, each key. Of K21, K22, ..., Touch the key indicating the tag name indicating the classification of the manuscript, and none of the tag names indicated by the keys K21, K22, ... Is the tag name indicating the classification of the manuscript. If it is determined, the operation unit 22 or the touch panel 23 is operated, a tag name indicating a new classification of the original document is input to the text box TB2, and the OK key K20 is touch-operated.

When the control unit 31 detects a touch operation of any of the keys K21, K22, ... Through the touch panel 23 (S106 "key"), the control unit 31 acquires the tag name associated with the instructed key (S107). .. Further, when the control unit 31 detects the touch operation of the OK key K20 through the touch panel 23 (S106 “tag name”), the control unit 31 acquires the tag name input to the text box TB2 (S108).

Then, the control unit 31 analyzes the image data indicating the image D1 of the original in the image memory 26 and extracts the characteristics of the image D1 of the original (S109).

Further, the control unit 31 uses the data indicating the characteristics of the image D1 of the manuscript extracted in S109 as the learning data for, for example, deep learning for discriminating the classification of the manuscript indicated by the tag name acquired in S107 or S108. Collect as (S110). The control unit 31 stores the deep learning learning data in the storage unit 28 (S110).

In the above, the target of image data analysis and image feature extraction by the control unit 31 is the original image read by the image reading unit 11, but instead of this, the network interface unit 91 uses the network interface unit 91. An image acquired from the terminal device, a server, or the like may be used as a target for image data analysis and image feature extraction.

For example, the tag name indicating the classification of the manuscript may be a contract, a receipt, a certificate, an invitation, or a poster. Further, the control unit 31 extracts elements such as hue, contrast, number of colors, or density of the image as features of the image of the original by the image processing unit 27. The learning data is used for learning by the deep learning performed by the control unit 31.

For example, the administrator sets the invitation card in the image reading unit 11, operates the operation unit 22 or the touch panel 23, inputs the training mode setting, inputs the "invitation card" as the tag name, and scans. It is assumed that the execution is instructed. The control unit 31 acquires an "invitation card" as a tag name, causes the image reading unit 11 to read the image of the invitation card, stores the image data indicating the image of the invitation card in the image memory 26, and stores the image data indicating the image of the invitation card in the image processing unit. In 27, various processes are applied to the image data. Then, the control unit 31 extracts the features of the image of the invitation and collects the data showing the features of the image of the invitation as learning data for discriminating the classification of the manuscript.

FIG. 7 is Chart A illustrating the correspondence between the hue, contrast, number of colors, and densities of the image, which are the characteristics of the image of the manuscript, and the invitation letter and the receipt, which are the tag names. In the training mode, as shown in Chart A of FIG. 7, an example is shown in which a combination of hue, contrast, number of colors, and density of an image is used as a feature of an image corresponding to each tag name input by the administrator. There is. In this example, an invitation and a receipt are used as the tag name.

With the training mode set, the administrator sets the document in the image reading unit 11, operates the operation unit 22 or the touch panel 23, inputs a tag name indicating the classification of the document, and executes scanning. Instruct. The control unit 31 extracts the characteristics of the image of the original document read by the image reading unit 11, and collects the data indicating the characteristics of the image of the original document as learning data for discriminating the classification of the original document.

In the present embodiment, deep learning is used for the process for determining the classification of the manuscript. FIG. 6 is a diagram showing the concept of deep learning performed by the control unit 31. In the deep learning shown in FIG. 6, the feature of the image is input from the input layer IL, and the tag name is output from the output layer OL via the intermediate layer (plurality of nodes ML). By adjusting how each node ML is connected, or "what kind of input and what kind of output is performed" by each node ML, that is, learning data is learned and generated. By doing or updating, the tag name to be acquired is determined and output based on the characteristics of the input image.

In this deep learning, in the training mode, each element indicating the characteristics of the image input from the input layer IL as the characteristics of the image corresponding to the tag name input by the administrator is stored as a node ML. The set of the node ML is an example of learning data.

Further, in the classification mode, in the output layer OL, the tag name corresponding to the feature of the image input from the input layer IL is tentatively determined and output. If this provisional tag name is approved by the administrator as indicating a formal tag name, each node ML is stored in association with the approved tag name. If this tentative tag name is changed by the administrator, each node ML is updated, and the tentatively determined tag name matches the tag name changed by the administrator. Updates are made (ie, the training data is updated). The deeper the hierarchy of each node ML and the larger the number of each node ML, the more complicated classification and judgment can be performed, and the tag name corresponding to the feature of the input image can be accurately determined. Therefore, the learning data of deep learning becomes more effective by increasing the number of images of the manuscript processed in the training mode and the classification mode, thereby deepening the hierarchy of each node ML and increasing the number of each node ML. Become. Preferably, the training mode is such that the acquisition rate of the correct tag name (the rate at which the tentatively determined tag name is approved by the administrator as indicating the official tag name) is, for example, 95% or more. Increase the number of images of the original processed in.

In addition, the control unit 31 uses the characteristics of the image of the collected manuscript in association with the tag name without using deep learning in the process for determining the classification of the manuscript, and the tag name and the image of the collected manuscript. The classification of the manuscript may be determined based on the correspondence with the characteristics of.

Next, the processing procedure in the classification mode will be described in detail with reference to the flowchart shown in FIG.

The user operates the operation unit 22 or the touch panel 23 to instruct the execution of the document classification function. The control unit 31 causes the display unit 21 to display the classification function input screen G1 shown in FIG. 4 in response to an instruction to execute the document classification function (S201).

The user operates the operation unit 22 or the touch panel 23, enters his / her identification information ID in the text box TB1, and operates the key K12 in the classification mode. The control unit 31 sets the classification mode when the classification mode setting instruction is input to the operation unit 22 or the touch panel 23 based on this operation (YES in S202) (S203). If this setting instruction is not input (NO in S202), the process ends.

With the classification mode set in this way, when the user sets the document in the image reading unit 11 and operates the operation unit 22 or the touch panel 23 to instruct the execution of scanning, the control unit 31 reads the image. The image of the original is read by the unit 11, the image data indicating the image of the original is stored in the image memory 26, and the image processing unit 27 performs various processing on the image data (S204). Similar to the training mode, the various processes by the image processing unit 27 in this classification mode are also the processes for emphasizing the density, contrast, edges, and corners of the image, which are performed to accurately extract the features of the image of the original. be.

Further, the control unit 31 causes the display unit 21 to display the classification mode setting screen G3 as shown in FIG. 9 (S205). The classification mode setting screen G3 shown in FIG. 9 displays image data showing an image D2 of a document read by the image reading unit 11 and a start key K31.

For example, when the user touches the start key K31 displayed on the display unit 21, the control unit 31 receives an instruction to execute the classification process based on the touch operation of the start key K31 through the touch panel 23, and follows this instruction. The image processing unit 27 analyzes the image data indicating the image D2 of the original in the image memory 26, extracts the characteristics of the image D2 of the original, and stores the characteristics of the image D2 of the original and the storage unit 28. Based on the learning data obtained by the deep learning, the classification (that is, the tag name) of the image D2 of the manuscript is determined (S206). Then, the control unit 31 displays the pop-up screen P1 as shown in FIG. 10 in the classification mode setting screen G3. On the pop-up screen P1 shown in FIG. 10, the OK key K32 and the NG key K33 are displayed together with the tag name of the image D2 of the original document determined in S206.

When the user determines that the tag name of the image D2 of the document on the pop-up screen P1 is correct, the user touches the OK key K32. When the control unit 31 detects the touch operation of the OK key K32 through the touch panel 23 (S207 “OK”), the control unit 31 adds information indicating the tag name determined in S206 to the image data indicating the image D2 of the document (S208). That is, the control unit 31 classifies the image D2 of the original. Then, the control unit 31 updates the deep learning learning data for determining the tag name indicating the classification of the manuscript by using the feature of the image D2 of the manuscript extracted in S206, and stores this learning data in the storage unit 28. To memorize (S209).

Further, when the user determines that the tag name of the image D2 of the document on the pop-up screen P1 is incorrect, the user touches the NG key K33. When the control unit 31 detects the touch operation of the NG key K33 through the touch panel 23 (S207 “NG”), the pop-up screen P2 as shown in FIG. 11 is displayed in the classification mode setting screen G3. In the pop-up screen P2 shown in FIG. 11, a text box TB3 for inputting a tag name indicating a new document classification, and keys K21, K22, and K23 associated with each already set tag name are displayed. , K24, K25 ..., And OK key K41 are displayed.

The user confirms the image D2 of the manuscript displayed on the display unit 21, and if any of the keys K21, K22, ... Is associated with the tag name indicating the classification of the manuscript, the corresponding key is used. If none of the keys K21, K22, ... Is associated with the tag name indicating the classification of the original, the operation unit 22 or the touch panel 23 is operated to indicate the new classification of the original. Is entered in the text box TB3, and the OK key K41 is touch-operated. When the control unit 31 detects a touch operation of any of the keys K21, K22, ... Through the touch panel 23, the control unit 31 assigns a tag name indicating the classification of the document associated with the instructed key to the image data of the document. It is acquired as a formal tag name to be added to (S208). On the other hand, when the control unit 31 detects the touch operation of the OK key K41, the control unit 31 adds a tag name indicating the classification of the new manuscript entered in the text box TB3 as a formal tag name to be added to the image of the manuscript. Acquire (S210). The control unit 31 adds the information indicating the tag name acquired in S210 to the image data indicating the image D2 of the original.

Then, the control unit 31 collects the characteristics of the image D2 of the manuscript used here as learning data for discriminating the tag name acquired in S210 or S208, and uses the characteristics of the image D2 of this manuscript to describe the above. The learning data is updated by deep learning in this manner, and the updated learning data is stored in the storage unit 28 (S209).

In the above, the target of image data analysis and image feature extraction by the control unit 31 is the original image read by the image reading unit 11, but instead of this, the network interface unit 91 uses the network interface unit 91. An image acquired from the terminal device, a server, or the like may be used as a target for image data analysis and image feature extraction.

Therefore, the user can automatically classify the image of the original by simply setting the classification mode and having the image reading unit 11 read the image of the original, and the image of the original is correctly classified. If not, you can enter the correct tag name to indicate the classification of the new manuscript and update the deep learning training data so that the correct tag name you entered is derived based on the characteristics of the image of the new manuscript. ..

12 to 14 show an example of adding information indicating a tag name to the image data of the manuscript.

FIG. 12 shows the property PR1 of the image of the original displayed on the display unit 21. The control unit 31 adds information indicating a tag name for the image of the manuscript to the image data of the manuscript and stores it together with the image data of the manuscript, and displays the tag name for the image of the manuscript. In the property PR1 as, it is displayed in the item Q1 (document classification) as one of the display items.

FIG. 13 shows the property PR2 of the image of the original displayed on the display unit 21. The control unit 31 adds information indicating a tag name for the image of the manuscript to the image data of the manuscript and stores it together with the image data of the manuscript, and displays the tag name for the image of the manuscript. In the property PR2 as, it is displayed in the item Q2 (tag) as one of the display items.

FIG. 14 shows a list FR of files displayed on the display unit 21 by the control unit 31. The control unit 31 adds information indicating a tag name for the image of the manuscript to the image data of the manuscript and stores it together with the image data of the manuscript, and stores the tag name as an image showing a list of files. It is displayed in the item Q3 of the tag name in the image FR.

By displaying the tag name for such a property or a list of files, it is possible to easily confirm the classification of the image of the executed manuscript.

Returning to FIG. 8, the control unit 31 saves and manages the image data indicating the image D2 of the original document read by the image reading unit 11 in S204 in a predetermined storage destination folder after the processing of S209. Manuscript image management process (S300) is performed. The manuscript image management process (S300) will be described with reference to FIG. FIG. 15 is a flowchart showing an example of the manuscript image management process of the first embodiment.

In the present embodiment, the storage unit 28 of the image processing device 10 has, for example, a shared folder as a predetermined storage destination folder. The control unit 31 determines whether or not there is a destination folder having a name matching the tag name (that is, classification) determined in S206 or the tag name corrected in S210 (that is, classification) in the shared folder of the storage unit 28. If there is no destination folder with the matching name, the destination folder with the matching name is created, and the image data indicating the image D2 of the manuscript is stored in the destination folder.

The control unit 31 causes the display unit 21 to display the classification management setting screen G4 shown in FIG. 16 (S301). FIG. 16 is a diagram showing an example of a setting screen for management by classification displayed on the display unit. On the classification management setting screen G4, the image D2 of the original read by the image reading unit 11, the display DP1 indicating that the tag name (classification) already set is "invitation", and the classification management are displayed. Key K51 associated with "on" indicating to perform, key K52 associated with "off" indicating not to perform classification management, and input box NB1 for inputting a predetermined sensitivity value Sen. , And the OK key K54 are displayed.

The control unit 31 determines whether or not a setting instruction for classification management (for example, on or off of classification management and sensitivity value Sen) has been input (S302). The control unit 31 accepts the presence / absence of classification management and the input of the sensitivity value Sen by the user's operation on the operation unit 22 or the touch panel 23, and when the OK key K54 is operated (YES in S302), the sensitivity value Sen. Is set (S303).

Here, the user touch-operates the key K51 to set it to "on" indicating that the management by classification is performed, and touch-operates the input box NB1 to input a numerical value of, for example, "70" in the operation unit 22. When the OK key K54 is operated (YES in S302), the control unit 31 sets the sensitivity value Sen to "70% (= +0.7)" (S303).

In S303, the control unit 31 has a predetermined range (“51” or more to “100” or less, “70” or more to “100” or less, etc.) as the sensitivity value Sen for the numerical input to the input box NB1. Only the numerical values in the above may be accepted. In this case, an appropriate sensitivity value Sen is set, and it becomes possible to suitably determine the similarity between the images of the original document described later. Further, the control unit 31 can set the sensitivity value Sen when the administrator is logged in, and on the classification management setting screen G4 shown in FIG. 16, the sensitivity value Sen is "70%" for the user. It may be displayed that it is.

The control unit 31 sets the classification of the tag name (S304). Set the tag name (classification) "invitation" as the classification.

The control unit 31 confirms the destination folder (S305). For example, the control unit 31 confirms whether or not the storage unit 28 has a shared folder as a predetermined storage destination folder. If there is no shared folder, the control unit 31 creates a shared folder. Then, the control unit 31 confirms whether or not the destination folder is included in the shared folder.

The control unit 31 determines whether or not there is a folder named "invitation" which is a tag name (classification) in the shared folder (S306), and if there is a folder named "invitation" (YES in S306). , It is determined whether or not the image data of the manuscript is in the folder named "invitation" (S307). If the image data of the manuscript is in the folder named "invitation" (YES in S307), the control unit 31 executes pixel matching (S308).

FIG. 17 is a diagram for explaining pixel matching. Using known pixel matching, the control unit 31 stores an image of a document similar to the image of the document read by the image reader 11 (the image of the scanned document) in the destination folder of the shared folder. Determine if it is. Pixel matching is a known matching technique that determines the degree of similarity between an input image (here, an image of a scanned document) and an output image (here, an image of a document stored in a folder) on a pixel-by-pixel basis. .. Pixel matching moves the input image to all possible positions in the output image and calculates a numerical index that indicates how well the input image matches the output image at that position. Matching is done on a pixel-by-pixel basis. Pixel matching gets the gray value of each pixel. This is calculated by the average value of R, G, B of each pixel. For example, when R, G, B of a certain pixel is R = 229, G = 255, B = 204, the gray value of this pixel is (229 + 255 + 204) / 3≈230.

Specifically, the control unit 31 uses the equation (1) shown in FIG. 18 to input an input image (here, an image of a scanned document) and an output image (here, an image of a document stored in a folder). ), The correlation value Col indicating the correlation with) is calculated as any value from “-1” to “+1”. Further, the predetermined sensitivity value Sen is a value having a positive correlation of "+1" or less, and is a value closer to "+1" than "0". In the present embodiment, the sensitivity value Sen is set to a value of "+0.7 (= 70%)".

FIG. 18 is a diagram showing an equation for calculating a correlation value using pixel matching and an example of calculating the correlation value. As shown in FIG. 18, the control unit 31 uses the equation (1) to obtain a 4-pixel source image (scanned original image) and a 4-pixel destination image (image of the original stored in the folder). The case of obtaining the correlation value with and will be described. It is assumed that the gray values of the four-pixel source image are 100, 120, 70, and 30, respectively, and the gray values of the four-pixel destination image are 100, 125, 60, and 30, respectively. Then, the control unit 31 calculates that the correlation value Col is "0.991" by using the equation (1). That is, the control unit 31 has a similarity between the 4-pixel source image (scanned original image) and the 4-pixel destination image (image of the original stored in the folder) of 99.1% (= +0. It is calculated as 991). Then, as shown in FIG. 16, the sensitivity value Sen is set to 70% (= + 0.7), and the control unit 31 controls the correlation value Col (= 99.1%)> the sensitivity value Sen (= 70). %), And it is determined that the 4-pixel source image (scanned original image) and the 4-pixel destination image (image of the original stored in the folder) are similar. If the correlation value Col is 100%, it is determined to be the same. When the control unit 31 determines that the correlation value Col <sensitivity value Sen, the control unit 31 has a 4-pixel source image (scanned original image) and a 4-pixel destination image (image of the original stored in the folder). ) Is dissimilar.

The control unit 31 determines whether or not the correlation value Col ≧ sensitivity value Sen (= 70%) (S309). When the control unit 31 determines that the correlation value Col ≧ sensitivity value Sen (= 70%) (YES in S309), the image D2 of the manuscript read by the image reading unit 11 is a folder named “invitation”. Since it is similar to the stored original in the above, it accepts an instruction as to whether or not to overwrite and save (S310).

When the control unit 31 has an operation of overwriting and saving to the operation unit 22 or the touch panel 23 by the user (YES in S310), the control unit 31 uses the image data indicating the image D2 of the document read by the image reading unit 11 as an “invitation letter”. When the image data of the stored manuscript in the folder named is overwritten and saved (S311) and there is another operation of saving to the operation unit 22 or the touch panel 23 by the user (NO in S310), "invitation letter" The image data indicating the image D2 of the original document read by the image reading unit 11 is saved while leaving the image data of the stored original document in the folder named (S312).

In S302 shown in FIG. 15, the control unit 31 is set to "Off" indicating that the user touch-operates the key K52 and does not perform classification-specific management, and the OK key K54 is operated (in S302). NO), instead of saving the image data indicating the image D2 of the document read by the image reading unit 11 by classification, normal saving is performed in a predetermined folder other than the folder named "invitation". Do (S314).

Further, when the control unit 31 determines in S306 shown in FIG. 15 that there is no folder named "invitation" (NO in S306), the control unit 31 creates a folder named "invitation" in the shared folder (NO). S313), the image data showing the image D2 of the original document read by the image reading unit 11 is saved in the generated folder named "invitation" (S312).

The control unit 31 terminates this process after the process of S311, after the process of S312, or after the process of S314.

Here, the cases of FIGS. 19 to 21 will be specifically described. FIG. 19 is a diagram showing an example of saving manuscript data when there is no folder having a name matching the classification of the scanned manuscript in the shared folder.

Regarding the image data showing the image D3 of the original document read by the image reading unit 11, the control unit 31 indicates that the classification (tag name) of the original document in S206 shown in FIG. 8 is an “invitation letter” as shown in FIG. It is assumed that it is determined to be. Then, as shown in FIG. 19, the control unit 31 determines the classification name (here, “invitation letter”) in the shared folder in S306 of FIG. 15 showing the contents of the document image management process (S300) shown in FIG. If it is determined that there is no folder (NO in S306), a folder with the name of "invitation" is generated in the shared folder (S313), and the image D3 of the manuscript read by the image reading unit 11 is shown. The image data is saved in the generated folder named "invitation" (S312).

Next, the case of FIG. 20 will be described. FIG. 20 is a diagram showing an example of saving manuscript data when a folder having a name matching the classification of the scanned manuscript is in the shared folder.

The control unit 31 determines whether or not there is a folder named "invitation" which is a tag name (classification) in the shared folder (S306), and if there is a folder named "invitation" (YES in S306). , It is determined whether or not the image data of the manuscript is in the folder named "invitation" (S307). In FIG. 20, the control unit 31 determines that there is image data indicating the image D31 of the original in the folder named “invitation” (YES in S307), and the image D3 of the original read by the image reading unit 11. And the image D31 of the stored manuscript in the folder named "invitation", pixel matching is executed to calculate the correlation value Col (S308), and the correlation value Col ≧ sensitivity value Sen (= 70%). If (YES in S309), it is determined that the image D3 is the same as or similar to the image D31, and the image data of the image D3 is overwritten and saved over the image data of the image D31 in the folder named "invitation". (S311).

On the other hand, if the correlation value Col <sensitivity value Sen (= 70%) (NO in S309), the control unit 31 determines that the image D3 is dissimilar to the image D31, and the folder named "invitation". The image data of the image D3 is stored separately from the image data of the image D31 (S312).

Next, the case of FIG. 21 will be described. FIG. 21 is a diagram showing an example of saving manuscript data when there is no folder having a name matching the classification of the manuscript read this time in the shared folder and there is a folder having another name.

The control unit 31 has a "receipt" for the image data indicating the image D4 of the original document read by the image reading unit 11 so that the classification (tag name) of the original document is as shown in FIG. 21 in S206 shown in FIG. It is assumed that it is determined to be. Then, in S306 of FIG. 15, which shows the contents of the manuscript image management process (S300) shown in FIG. 8, the control unit 31 has a tag name (classification) “receipt” in the shared folder as shown in FIG. It is determined whether or not there is a folder named "" (S306). Here, the control unit 31 determines that there is a folder named "invitation" which is a tag name (classification) in the shared folder, and a folder named "receipt" which is a tag name (classification) is created. No (NO in S306). Then, the control unit 31 creates a folder named "receipt" in the shared folder (S313), and generates image data indicating the image D4 of the original document read by the image reading unit 11 in the generated "receipt". It is saved in the folder named "(S312). Since it is not necessary to change the image data for the folders other than the folder named "receipt", the image data of the image D31 in the folder named "invitation" is saved as it is.

According to the first embodiment described above, when the training mode is set by operating the operation unit 22 or the touch panel 23, the control unit 31 is subjected to an image acquisition unit (image reading unit 11 or network interface unit 91). The characteristics of the image of the manuscript from which the image data has been acquired are extracted, and the data indicating the characteristics of the image of the manuscript is collected as training data for determining the classification of the manuscript. Further, when the classification mode is set by operating the operation unit 22 or the touch panel 23, the control unit 31 is a document whose image data has been acquired by the image acquisition unit (image reading unit 11 or network interface unit 91). The characteristics of the image are extracted, the classification of the manuscript is discriminated based on the characteristics of the image of the manuscript and the learning data, and the information indicating the discriminated classification is added to the image data of the manuscript, and a predetermined storage destination is specified. Check if there is a destination folder with a name that matches the identified classification (tag name) in the folder (shared folder), and if there is no destination folder with the matching name, create a destination folder with the matching name. The image data of the manuscript is stored in the destination folder. As a result, the classification (tag name) of the image of the manuscript can be specified based on the characteristics of the entire image of the manuscript, and the image data of the manuscript is stored in the destination folder corresponding to the specified classification (tag name). Can be done.

Further, when the control unit 31 has a destination folder having a matching name in a predetermined storage destination folder (shared folder), the control unit 31 has the image indicated by the stored image data in the destination folder and the image data of the manuscript. The correlation value Col indicating the correlation with the image shown by is calculated, and (i) the image data of the manuscript is stored image data when the calculated correlation value Col is less than the predetermined sensitivity value Sen. The image data of the manuscript is stored in the destination folder, and (ii) the image data of the manuscript is obtained when the calculated correlation value Col is equal to or higher than the predetermined sensitivity value Sen. Is determined to be the same as or similar to the stored image data, and the image data of the original is overwritten and stored. Therefore, when saving in the destination folder, it is possible to determine whether or not image data that is the same as or similar to the content of the image data of the manuscript already exists in the destination folder, and if it is the same or similar, the image of the manuscript. The data can be overwritten and stored, and if they are dissimilar, the image data will not be duplicated, so the image data of the original is stored. As a result, it is possible to prevent a plurality of image data of the original document that is the same as or similar to the image data of the read document from being duplicately stored in the destination folder corresponding to the specified classification (tag name).

Further, the control unit 31 determines that the image data of the manuscript is the same as or similar to the stored image data when the calculated correlation value Col is equal to or higher than the predetermined sensitivity value Sen, and the operation unit 22 determines. Alternatively, when an overwrite instruction is given by operating the touch panel 23, the image data of the manuscript is overwritten and stored, and when another storage instruction is given by operating the operation unit 22 or the touch panel 23, the image data of the manuscript is stored. Store it separately from the completed image data. As a result, if stored image data that is the same as or similar to the image data of the read document is already stored in the destination folder, the read document can be read according to an overwrite instruction or another storage instruction by the user. The image data can be overwritten or stored separately. That is, it is possible to select whether to save the image data of the original by overwriting or to save it separately. Therefore, the degree of freedom of choice regarding the storage of image data can be expanded.

Further, the control unit 31 shows the correlation between the saved image and the source image by performing pixel matching between the saved image indicated by the stored image data in the destination folder and the source image indicated by the image data of the original. The correlation value Col is calculated as any value from "-1" to "+1". The predetermined sensitivity value Sen is a value having a positive correlation of "+1" or less, and is a value closer to "+1" than "0". In the present embodiment, the sensitivity value Sen is set to a value of "+0.7 (= 70%)". If the correlation value is "+1", it can be determined to be the same, and if the correlation value is equal to or greater than a predetermined sensitivity value (that is, a value closer to "+1" than "0"), it is determined to be similar. can. This makes it possible to preferably determine whether the stored image and the source image are the same or similar.

In addition, although it is assumed that one image data is stored in the folder named "invitation" in FIG. 20, a case where a plurality of image data are stored will be described with reference to FIG. 22.

FIG. 22 is a diagram showing an example of saving manuscript data when a folder having a name matching the classification of the scanned manuscript is in the shared folder and a plurality of image data are saved in the folder having the name. ..

The control unit 31 determines whether or not there is a folder named "invitation" which is a tag name (classification) in the shared folder (S306), and if there is a folder named "invitation" (YES in S306). , It is determined whether or not the image data of the manuscript is in the folder named "invitation" (S307). In FIG. 22, the control unit 31 determines that there is image data of each image D32 to D34 of a plurality of (three) documents in the folder named “invitation” (YES in S307).

Then, the control unit 31 executes pixel matching for the image D3 of the document read by the image reading unit 11 and the images D32 to D34 of the three stored documents in the folder named "invitation". Then, the correlation value Col is calculated (S308).

Then, when (i) all the calculated correlation value Cols are all less than the predetermined sensitivity value Sen (= 70%) (NO in S309), the control unit 31 said, as shown in FIG. 22. It is determined that the image D3 of the manuscript is dissimilar to each of the images D32 to D34 indicated by the three stored image data, and the folder named "invitation" contains the image data indicating each image D32 to D34. Separately, image data indicating the image D3 of the manuscript is stored (S312).

On the other hand, when (ii) one or more of the calculated correlation value Cols have one or more correlation value Cols having the predetermined sensitivity value Sen (= 70%) or more, the control unit 31 of the manuscript. It is determined that the image data is the same as or similar to one or more stored image data indicating a correlation value Col having a sensitivity value of Sen (= 70%) or more among a plurality of stored image data (YES in S309). , The image data of the manuscript is overwritten and stored, and the one or more stored image data is deleted (S311).

Specifically, as shown in FIG. 22, the correlation value Col between the image D3 and the image D32 in the folder named "invitation" is the sensitivity value Sen (= 70%) or more, and the image D3 and the "invitation". The correlation value Col with the image D33 in the folder named "like" is the sensitivity value Sen (= 70%) or more, and the correlation value Col with the image D3 and the image D34 in the folder named "invitation" is the sensitivity. It is assumed that the value is less than Sen (= 70%). That is, it is assumed that the image D3 is similar to the image D32 and the image D33, and the image D3 and the image D34 are dissimilar. The control unit 31 overwrites and saves the image data of the image D3 in the folder named "invitation", and deletes the image data of the image D32 and the image D33 (S311). Since the image D3 and the image D34 are dissimilar, the image data of the image D34 is stored as it is.

As a result, when there is one or more image data of a manuscript that is the same as or similar to the image data of the read document in the destination folder corresponding to the specified classification (tag name), the one or more similar storages. It is possible to delete the completed image data and overwrite and save the image data of the read document, and it is possible to accurately manage the image data in the destination folder.

Next, the information processing system 100 of the second embodiment will be described with reference to the drawings. FIG. 23 is a block diagram showing a main internal configuration of the information processing system of the second embodiment.

In the first embodiment, the classification of the document read by the image reading unit 11 is determined, and the document read by the image reading unit 11 is placed in the folder with the classification name in the shared folder in the storage unit 28 of the image processing device 10. Image data is saved. On the other hand, as shown in FIG. 23, the information processing system 100 of the second embodiment includes an image processing device 10 and an information processing device 40 (for example, a computer) that can be connected via the network N, and performs image processing. The point of storing the image data of the original document read by the image reading unit 11 in the folder of the classification name in the shared folder in the storage unit 42 in the information processing device 40 connected to the device 10 is the above-mentioned first embodiment. Is different.

The information processing apparatus 40 includes a storage unit 42 having a shared folder as a predetermined storage destination folder.

FIG. 24 is a flowchart showing an example of the manuscript image management process of the second embodiment. The second embodiment is different from the first embodiment in that the process of S330 shown in FIG. 24 is added as compared with the process shown in FIG. 15 of the first embodiment. Therefore, S330 shown in FIG. 24 will be described in detail.

When a file sharing protocol (for example, SMB (Server Message Block)) program is installed in the image processing device 10 and the information processing device 40, and the connection to the information processing device 40 is established in the image processing device 10. (S330), the image processing device 10 can access the shared folder of the storage unit 42 of the information processing device 40.

Specifically, the control unit 31 of the image processing device 10 correctly inputs information for establishing a connection to the operation unit 22 or the touch panel 23 (for example, a host name, a path, a user name, and a password). The connection state to the information processing apparatus 40 is established (S330). When the connection state to the information processing device 40 is established (S330), the control unit 31 becomes accessible to the shared folder of the storage unit 42 of the information processing device 40 (S305).

Then, the control unit 31 confirms whether or not there is a destination folder having a name matching the determined classification (tag name) in the shared folder of the storage unit 42 (S306). Then, when the destination folder with the matching name does not exist (NO in S306), the control unit 31 causes the destination folder with the matching name to be created (S313), and causes the image data of the original to be transmitted to the destination folder. And memorize (S312).

Further, when the control unit 31 has a destination folder with the matching name (YES in S306) and there is image data of the original in the destination folder (YES in S307), the control unit 31 executes pixel matching (S308). It is determined whether or not the correlation value Col ≧ sensitivity value Sen (= 70%) (S309). When the control unit 31 determines that the correlation value Col ≧ sensitivity value Sen (= 70%) (YES in S309), the image of the manuscript read by the image reading unit 11 is the stored manuscript in the destination folder. Therefore, when the instruction to overwrite and save is received (YES in S310), the image data of the original is transmitted to the destination folder and stored by overwriting (S311).

FIG. 25 shows an example of a transmission setting screen displayed on the display unit 21 by the control unit 31. The control unit 31 adds information indicating a tag name for the image of the manuscript to the image data of the manuscript, stores it together with the image data of the manuscript, and processes the image data of the manuscript connected to the image processing device 10. When displaying the setting screen G5 for setting transmission when transmitting to the destination folder in the shared folder of the storage unit 42 of the device 40 on the display unit 21, the tag name is displayed in the preview image display area FR2 on the setting screen G5. It is displayed in the item Q5 of the tag name in. That is, the tag name is displayed as information about the manuscript when the preview is displayed. Then, when the control unit 31 has image data indicating an image similar to the destination folder, the message M1 "There is similar image data in the invitation folder" and "Overwrite save or separate save". The message M2 and "?" Are displayed on the setting screen G5. When the key K61 indicating overwrite saving shown in FIG. 25 is touch-operated by the user (YES in S310), the control unit 31 points the manuscript toward the destination folder in the shared folder of the storage unit 42 of the information processing apparatus 40. The image data of the above is transmitted and overwritten and stored (S311). When the key K62 indicating another storage shown in FIG. 25 is touch-operated by the user (NO in S310), the control unit 31 points toward the destination folder in the shared folder of the storage unit 42 of the information processing apparatus 40. The image data of the manuscript is transmitted and stored separately (S312).

According to the second embodiment described above, the image processing device 10 can access the shared folder of the storage unit 42 of the information processing device 40 connected by the network N, and the specified classification (tag) in the shared folder is possible. If there is image data of the original document that is the same as or similar to the image data of the scanned document in the destination folder corresponding to (name), the image data of the scanned document is overwritten and saved, so that the image data of the scanned document is saved. It is possible to prevent multiple image data of the same or similar originals from being stored in duplicate.

In the second embodiment, the computer is taken as an example of the information processing apparatus 40, but it may be a server. In this case, a file sharing protocol (for example, FTP (File Transfer Protocol) program) is installed in the image processing device 10 and the information processing device 40, and the image processing device 10 establishes a connection to the information processing device 40. Then (S330), the image processing device 10 can access the shared folder of the storage unit 42 of the information processing device 40.

In the above embodiment, when the user gives an instruction for overwrite saving (YES in S310), the overwrite save is executed, but if they are the same or similar, the overwrite save is executed without waiting for the above instruction. May be.

In the above embodiment, one embodiment of the present invention is described using an image processing device (multifunction device), but this is only an example, and other images such as a copier, a printer, and a facsimile machine. It may be a processing device.

Further, the configuration and processing of the above-described embodiment described with reference to FIGS. 1 to 25 are merely examples of the present invention, and the present invention is not intended to be limited to the configuration and processing.

10 Image processing device 11 Image reading unit (image acquisition unit)
12 Image forming unit 21 Display unit 22 Operation unit 23 Touch panel 28 Storage unit 29 Control unit 31 Control unit 40 Information processing device 42 Storage unit 91 Network interface unit (image acquisition unit)
100 Information processing system

Claims (9)

The image acquisition unit that acquires the image data of the manuscript, and
The operation unit where settings and instructions are input by user operation,
A control unit for setting a training mode or a classification mode according to an instruction input to the operation unit is provided.
The control unit
When the training mode is set by operating the operation unit, the characteristics of the image of the manuscript whose image data has been acquired by the image acquisition unit are extracted, and the data indicating the characteristics of the image of the manuscript is obtained. Collected as learning data to determine the classification of the manuscript,
When the classification mode is set by the operation to the operation unit, the characteristics of the image of the manuscript whose image data has been acquired by the image acquisition unit are extracted, and the characteristics of the image of the manuscript and the learning data are used. Based on this, the classification of the manuscript is discriminated, information indicating the discriminated classification is added to the image data of the manuscript, and the presence or absence of a destination folder with a name matching the discriminated classification is found in the predetermined storage destination folder. An image processing device that checks and if there is no destination folder with the matching name, creates a destination folder with the matching name and stores the image data of the manuscript in the destination folder. When the control unit has a destination folder with the matching name in the predetermined storage destination folder, the control unit has an image indicated by the stored image data in the destination folder and an image indicated by the image data of the manuscript. A correlation value indicating a correlation with is calculated, and (i) the image data of the manuscript is dissimilar to the stored image data when the calculated correlation value is less than a predetermined sensitivity value. The image data of the manuscript is stored in the destination folder, and (ii) the image data of the manuscript is the stored image when the calculated correlation value is equal to or higher than the predetermined sensitivity value. The image processing apparatus according to claim 1, wherein it is determined that the data is the same as or similar to the data, and the image data of the manuscript is overwritten and stored. The control unit determines that (ii) the image data of the manuscript is the same as or similar to the stored image data when the calculated correlation value is equal to or higher than the predetermined sensitivity value, and the operation unit. When an overwrite instruction is given by the operation to, the image data of the manuscript is overwritten and stored, and when another storage instruction is given by the operation to the operation unit, the image data of the manuscript is combined with the stored image data. Is the image processing apparatus according to claim 2, which is stored separately. When the control unit has a destination folder having the matching name in the predetermined storage destination folder and has a plurality of stored image data in the destination folder, the control unit has the plurality of stored images. Correlation values indicating the correlation between the image shown by the data and the image shown by the image data of the manuscript are calculated, and (i) when all the calculated correlation values are less than the predetermined sensitivity value. It is determined that the image data of the manuscript is dissimilar to the plurality of stored image data, and the image data of the manuscript is stored in the destination folder. When there is one or more correlation values that are equal to or higher than the predetermined sensitivity value, the correlation value in which the image data of the manuscript is equal to or higher than the predetermined sensitivity value among the plurality of stored image data is shown. The image processing apparatus according to claim 1, wherein it is determined that the image data is the same as or similar to one or more stored image data, the image data of the original is overwritten and stored, and the one or more stored image data is deleted. The control unit performs pixel matching between the saved image indicated by the stored image data in the destination folder and the source image indicated by the image data of the manuscript, thereby correlating the saved image with the source image. The correlation value indicating the above is calculated as any value from "-1" to "+1".
The predetermined sensitivity value is one of claims 2 to 4, which is a value having a positive correlation of "+1" or less and is closer to "+1" than "0". The image processing device described. When a different classification of a document is input by an operation to the operation unit after the classification of the original is determined in the classification mode, the control unit indicates the input classification instead of the determined classification. The invention according to any one of claims 1 to 5, wherein the information is added to the image data of the manuscript, and the learning data used for the discrimination is stored in association with the classification of the input different manuscript. Image processing device. The image processing device according to any one of claims 1 to 6, wherein the control unit extracts the hue, contrast, number of colors, or density of the image data as a feature of the image of the document. With more display
The image according to any one of claims 1 to 7, wherein the control unit associates the classification of the manuscript determined in the classification mode with the image data of the acquired manuscript and displays it on the display unit. Processing equipment. An information processing system comprising the image processing device according to any one of claims 1 to 4 and an information processing device, and the image processing device and the information processing device can be connected to each other via a network. hand,
The information processing apparatus includes a storage unit having a shared folder as a predetermined storage destination folder.
The control unit of the image processing device establishes a connection state to the information processing device by inputting information for establishing a connection to the operation unit, and accesses the shared folder of the storage unit of the information processing device. Check if there is a destination folder with a name that matches the identified classification in the shared folder, and if there is no destination folder with the matching name, create a destination folder with the matching name. , An information processing system that stores the image data of the manuscript in the destination folder.

Images