Image reading device and image forming device

Abstract

To provide an image reading device capable of recognizing image information which cannot be recognized when the image information is overlapped.SOLUTION: An image reading device 100 includes a reading unit 10, a recognition unit 32, and a separation unit 33. The reading unit 32 reads an original M and outputs read data SC. The recognition unit 32 identifies the layer on which each piece of image information is arranged on the basis of a plurality of pieces of image information included in the read data SC. The separation unit 33 separates the plurality of layers identified by the recognition unit 32 and deletes all layers except the specific layer.SELECTED DRAWING: Figure 2

Description

[Technical Field]

[0001] The present invention relates to an image reading device and an image forming device. [Background technology]

[0002] A conventional image reading device is described in Patent Document 1. The image reading device described in Patent Document 1 extracts text information in an area specified by the user from an image read from a sheet, and makes the text information available for later use. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Korean Patent Publication No. 10-2016-0097394 Summary of the Invention [Problem to be solved by the invention]

[0004] However, when the image reading device described in Patent Document 1 extracts text information from an image read from a sheet, for example, if part of the text information overlaps with graphic information, the part of the text information overlapping the graphic information may be missing, making it impossible to confirm the extracted text information.

[0005] The present invention has been made in consideration of the above-mentioned problems, and aims to provide an image reading device and an image forming device that can recognize image information that cannot be recognized when the image information is overlapped. [Means for solving the problem]

[0006] According to one aspect of the present invention, an image reading device includes a reading unit, an identification unit, and a separation unit. The reading unit reads an original document and outputs read data. The identification unit identifies a layer on which each piece of image information is arranged based on a plurality of pieces of image information included in the read data. The separation unit separates the plurality of layers identified by the identification unit and deletes all layers except for the specific layer.

[0007] According to another aspect of the present invention, an image forming apparatus includes the image reading device described above and an image forming section that forms an image on a recording medium. [Effects of the Invention]

[0008] According to the image reading device and image forming device of the present invention, it is possible to recognize image information that could not be recognized when the image information was overlapped. [Brief explanation of the drawings]

[0009] [Figure 1] 1 is a schematic diagram illustrating an image forming apparatus according to an embodiment of the present invention. [Figure 2] 1 is a block diagram showing a configuration of an image forming apparatus according to an embodiment of the present invention; [Figure 3] 2 is a diagram illustrating an example of the configuration of a recognition unit included in a control unit of an image forming apparatus according to the present embodiment. FIG. [Figure 4] 3 is a diagram illustrating an example of the configuration of a filling unit included in a control unit of the image forming apparatus according to the present embodiment. FIG. [Figure 5] FIG. 10 is a diagram showing image data from which the overlay of the scanned data has been deleted. [Figure 6] FIG. 10 is another diagram showing image data from which the overlay of the scanned data has been removed. [Figure 7] 10 is a flowchart of an overlay deletion process executed by a control unit of the image forming apparatus according to the present embodiment. [Figure 8] FIG. 10 is a flowchart illustrating an overlay deletion process executed by an image forming apparatus according to a modified example. DETAILED DESCRIPTION OF THE INVENTION

[0010] Hereinafter, embodiments of the present invention will be described with reference to the drawings. In the drawings, the same or corresponding parts are designated by the same reference characters and description thereof will not be repeated.

[0011] An image forming apparatus 300 including an image reading device 100 according to an embodiment of the present invention will be described with reference to Fig. 1. Fig. 1 is a schematic diagram showing the image forming apparatus 300 according to an embodiment of the present invention. The image forming apparatus 300 is, for example, a copier, a printer, or a multifunction peripheral. In the following, as an example, a case will be described in which the image forming apparatus 300 is a monochrome multifunction peripheral having a printer function, a copier function, and a facsimile function.

[0012] 1, image forming apparatus 300 includes image reading device 100 and image forming unit 200. Image reading device 100 includes reading section 10, control section 30, storage section 40, document table 12, document transport section 110, and operation display section 120. Image forming unit 200 includes image forming section 220, paper feed cassette 230, paper transport section 240, and paper discharge section 270.

[0013] The reading unit 10 reads the original document M and outputs read data SC. Specifically, the reading unit 10 reads an image formed on the original document M and generates read data SC. In more detail, the reading unit 10 reads an image formed on the original document M transported by the original document transport unit 110 or an image formed on the original document M placed on the document table 12 and generates read data SC.

[0014] The image forming unit 220 forms an image on the recording medium P based on the read data SC read by the reading unit 10.

[0015] The control unit 30 is a hardware circuit configured by a processor such as a CPU (Central Processing Unit), an ASIC (Application Specific Integrated Circuit), etc. The control unit 30 controls the operation of each part of the image forming apparatus 300 by the processor reading and executing a control program stored in the storage unit 40.

[0016] The storage unit 40 is, for example, an HDD or SSD. The storage unit 40 may include a RAM (Random Access Memory) and a ROM (Read Only Memory). The storage unit 40 stores various data and a control program for controlling the operation of each unit of the image forming apparatus 300. The control program may be an image reading program for controlling the reading of an image. The control program is executed by the control unit 30. Furthermore, the read data SC read by the reading unit 10 is written to a predetermined data area in the storage unit 40.

[0017] The operation display unit 120 is used to allow a user to operate the image forming apparatus 300. The operation display unit 120 includes a display unit 121 and an operation unit 122. The display unit 121 is configured with a display such as an LCD (Liquid Crystal Display) or ELD (Electro Luminescence Display) having a touch panel function. User operations are input through the operation unit 122. The operation unit 122 corresponds to an example of a "reception unit." In an embodiment in which a touch panel functions as the operation display unit 120, the display unit 121 and a part of the operation unit 122 may be integrated. The operation unit 122 of the operation display unit 120 may also be physical buttons. The operation unit 122 of the operation display unit 120 may have a touch panel and physical buttons.

[0018] A recording medium P for printing is stored in paper feed cassette 230. When printing, recording medium P in paper feed cassette 230 is transported by paper transport unit 240 so as to pass through image forming unit 220 and be discharged from paper discharge unit 270.

[0019] The image forming apparatus 300 will be further described with reference to Figures 1 to 6. Figure 2 is a block diagram showing the configuration of the image forming apparatus 300 according to the embodiment of the present invention.

[0020] 2, the control unit 30 of the image forming apparatus 300 has an extraction unit 31, a recognition unit 32, and a separation unit 33. Specifically, the control unit 30 functions as the extraction unit 31, the recognition unit 32, and the separation unit 33 by executing a computer program stored in the storage unit 40.

[0021] The extraction unit 31 extracts image information from the read data SC. Specifically, the extraction unit 31 extracts a plurality of pieces of image information from the read data SC stored in the storage unit 40. One of the plurality of pieces of image information is image information representing text. Furthermore, another of the plurality of pieces of image information is image information representing an object. The image information representing an object is image information including any of an image representing a natural object, an image representing an artificial object, and an image representing a living thing. The image information extracted by the extraction unit 31 is stored in the storage unit 40.

[0022] The identification unit 32 identifies the layer. Specifically, the identification unit 32 identifies the layer on which each piece of image information is arranged for each piece of image information. More specifically, the identification unit 32 identifies the layer on which each piece of image information is arranged for each piece of image information based on the plurality of pieces of image information included in the read data SC.

[0023] The separating unit 33 separates the multiple layers identified by the identifying unit 32. Furthermore, the separating unit 33 deletes all layers except for a specific layer from the multiple layers.

[0024] Therefore, the separation unit 33 can generate image data in which only the specific layer remains. In other words, the image information contained in each layer is deleted along with the other layers except for the image information contained in the specific layer. This allows the separation unit 33 to generate image data containing only the image information contained in the specific layer. As a result, it becomes possible to recognize image information that could not be recognized when the image information was overlapped.

[0025] In addition, the identification unit 32 of this embodiment identifies the first layer LY1 and the second layer LY2 included in the read data SC. The identification unit 32 also identifies multiple pieces of image information included in the read data SC. The multiple pieces of image information include first image information and second image information different from the first image information. The first image information is arranged on the first layer LY1. The second image information is arranged on the second layer LY2.

[0026] The separation unit 33 separates the first layer LY1 and the second layer LY2. The second layer LY2 overlies the first layer LY1. Furthermore, the separation unit 33 deletes the second layer LY2 from the read data SC. Therefore, the second layer LY2 overlaying the first layer LY1 is deleted. In other words, the second image information of the second layer LY2 overlaying the first image information included in the first layer LY1 is deleted. This allows the separation unit 33 to generate image data of only the first layer LY1. As a result, it becomes easier to recognize the image information of the first layer LY1.

[0027] Next, the configuration of the recognition unit 32 will be specifically described with reference to Figures 2 and 3. Figure 3 is a diagram showing an example of the configuration of the recognition unit 32 included in the control unit 30 of the image forming apparatus 300 according to this embodiment.

[0028] The identification unit 32 identifies whether the form of each piece of image information included in each layer is image information representing text. If the form of image information included in a layer is not image information representing text, the layer identification unit 321 identifies it as image information representing an object.

[0029] 3, the classification unit 4 includes a layer classification unit 321, a text classification unit 322, and an arrangement pattern classification unit 323. The layer classification unit 321 identifies layers included in the read data SC. The layer classification unit 321 includes, for example, a depth detection processing unit 321a, an overlap analysis processing unit 321b, and a region classification unit 321c.

[0030] The depth detection processing unit 321a performs a depth detection process to detect the overlapping manner of image information in the read data SC as depth information using information about the shape of the image information. For example, the depth detection processing unit 321a estimates the position in a virtual depth direction for each piece of image information included in the read data SC. Then, a predetermined reference position in the depth direction of the read data SC is set, and the distance from the reference position to each piece of image information is estimated as depth information (an example of approximation) for each piece of image information. The depth detection processing unit 321a may be configured to perform the depth detection process on the read data SC using a neural network model obtained in advance by machine learning.

[0031] That is, the classification unit 32 classifies the first layer LY1 and the second layer LY2 using a neural network model obtained in advance through machine learning. The neural network model detects how the first image information and the second image information overlap as depth information.

[0032] A typical neural network used is a CNN (Convolutional Neural Network). CNN is a type of neural network that is widely used in fields such as image recognition. CNN is one method for realizing deep learning. In this embodiment, CNN extracts features of each piece of image information from information about the form of the image information. By using CNN, the depth detection processing unit 321a can recognize the image information as the same even if the position of the image information in the read data SC has changed, the image information has been rotated, or the image information has been inverted.

[0033] When a neural network model (CNN model) using CNN as the neural network is constructed, machine learning is performed in advance. For example, a set of input / output relationship data is made up of an image as input and information about the shape and depth information of each item in the image as output. The CNN model is obtained as a neural network model that can estimate the depth information and shape type of each image information from any read data SC by performing machine learning in advance using multiple sets of different input / output relationship data as training data.

[0034] The depth detection processing unit 321a estimates depth information from information about the form of the image information using a CNN model, and estimates how the image information overlaps. Note that by performing machine learning in advance using combinations of read data SC with different form types (text or figures) and depth information as training data, the CNN model can estimate whether each piece of image information is image information representing text or image information representing an object. The CNN model is stored in the storage unit 40 as a first trained model 41, as shown in FIG. 2.

[0035] The overlap analysis processing unit 321b identifies an overlapping portion of the read data SC between pieces of image information based on the depth information estimated by the depth detection processing unit 321a. The overlap analysis processing unit 321b may be configured to perform overlap analysis processing on the read data SC using a neural network model (CNN model) obtained in advance by machine learning.

[0036] The region identification unit 321c analyzes the cross section of the read data SC and identifies the differences between each layer. Furthermore, the layer identification unit 321 estimates that image information with no missing elements in the overlapping portion of image information is the surface layer. Furthermore, the layer identification unit 321 estimates that a portion with missing elements in the overlapping portion of image information is a layer below the surface layer. Furthermore, the region identification unit 321c identifies, based on the depth information estimated by the depth detection processing unit 321a, image information that is determined to overlap similarly in the read data SC as belonging to the same layer. The region identification unit 321c may be configured to perform region identification processing on the read data SC using a neural network model (CNN model) obtained in advance by machine learning.

[0037] The text identification unit 322 identifies text image information from the read data SC. The text identification unit 322 includes, for example, a typography analysis processing unit 322a and a character detail analysis processing unit 322b.

[0038] The typography analysis processor 322a performs typography analysis processing to calculate the degree of similarity in the form, composition, etc. of the image information of each piece of text in a layer by analyzing elements including the shape, size, layout, font, color, and background of the text included in the read data SC. The typography analysis processor 322a may be configured to perform this processing in combination with OCR (Optical Character Recognition) processing. By combining OCR processing, such as edge detection and shape recognition, with typography analysis processing, the text identification unit 322 can analyze the shape of characters more accurately.

[0039] The character detail analysis processing unit 322b calculates the degree of similarity regarding the font style of the text image information by identifying the font family, font style, and the like.

[0040] The text identification unit 322 identifies, as a group of image information, image information of one or more texts that are determined to have similar forms to each other based on the similarity of the image information of the text obtained by the typography analysis processing unit 322a and the character detail analysis processing unit 322b.

[0041] The arrangement pattern identification unit 323 identifies the arrangement pattern of image information in the read data SC. The arrangement pattern identification unit 323 has, for example, an object processing unit 323a and a median filtering processing unit 323b. The object processing unit 323a and the median filtering processing unit 323b perform an arrangement pattern analysis process to analyze the arrangement pattern of image information in the read data SC. The object processing unit 323a calculates a degree of approximation as to whether the image information is similar to an arrangement pattern locally arranged in a part of the read data SC. The median filtering processing unit 323b calculates a degree of approximation as to whether the image information is similar to an arrangement pattern arranged over the entire read data SC.

[0042] The object processing unit 323a and the median filtering processing unit 323b may be configured to perform processing using the CNN model described above. The CNN model is machine-learned in advance using multiple training data sets with different arrangement patterns, thereby enabling it to calculate the degree of similarity between the arrangement patterns. The arrangement pattern identification unit 323 classifies, as a group of image information, one or more pieces of image information whose arrangement patterns are determined to be similar to each other, based on the degree of similarity between the arrangement patterns calculated by the object processing unit 323a and the median filtering processing unit 323b.

[0043] Next, the control unit 30 of the image forming apparatus 300 of this embodiment will be described in more detail with reference to FIGS. 2 to 6. The control unit 30 of this embodiment is capable of predicting missing portions of image information of a layer that will remain after the overlay has been deleted and filling in the missing portions. FIG. 4 is a diagram showing an example of the configuration of the filling unit 35 included in the control unit 30 of the image forming apparatus 300 of this embodiment. FIG. 5 is a diagram showing image data from which the overlay of the read data SC has been deleted. FIG. 6 is another diagram showing image data from which the overlay of the read data SC has been deleted.

[0044] 2, the control unit 30 of the image forming apparatus 300 further includes a filling unit 35. Specifically, the control unit 30 functions as the filling unit 35 by executing a computer program stored in the storage unit 40.

[0045] As shown in FIG. 5, the first scanned data SC1 of this embodiment has a first layer LY1 and a second layer LY2 superimposed on it. In FIG. 5, for ease of explanation, the area of ​​the first layer LY1 is smaller than the area of ​​the second layer LY2. The area of ​​the first layer LY1 may be larger than or the same as the area of ​​the second layer LY2. In this embodiment, it is sufficient that the second image information DT2 included in the second layer LY2 superimposes on the first image information DT1 included in the first layer LY1. When the second image information DT2 overlaps a portion of the first image information DT1, elements of the first image information DT1 are partially missing. Therefore, the first image information DT1 includes a first portion DT11 in which elements are partially missing and a second portion DT12 in which elements are not missing.

[0046] The filling unit 35 fills the missing elements into the first portion DT11. Therefore, the missing elements are filled into the first portion DT11 from which elements were missing. As a result, when the second layer LY2 is deleted, the first image information DT1 can be easily recognized.

[0047] As shown in FIG. 4, the filling section 35 includes a text filling section 351 and a color filling section 352.

[0048] The text filling unit 351 includes a typography analysis processing unit 351a, a character detail analysis processing unit 351b, and a character recognition unit 351c.

[0049] The typography analysis processing unit 351a analyzes elements of the text included in the read data SC, including the shape, size, arrangement, font, color, background, etc. The typography analysis processing unit 351a analyzes the form of the text in the second part DT12 and predicts the form of the text in the first part DT11. The typography analysis processing unit 351a may be configured to perform this in combination with OCR processing. By combining OCR processing and typography analysis processing, the typography analysis processing unit 351a can analyze the shape of the text more accurately.

[0050] The character detail analysis processing unit 351b calculates the degree of similarity regarding the font style of the text image information by identifying the font family, font style, and the like.

[0051] The character recognition unit 351c reads the image information in which the missing elements have been filled in the first part DT11, and determines whether it is a correct word.

[0052] Furthermore, the text filling unit 351 may perform natural language processing. The natural language processing is performed using a natural language processing model (NLP model) obtained in advance by performing machine learning. The NLP model is a mathematical model that can perform tasks such as generating, classifying, translating, and summarizing text. The NLP model is obtained in advance by learning statistical features of text through machine learning. There are various types of NLP models, such as statistical models and neural network models. By using the NLP model, the natural language processing can predict the content contained in the first part DT11 from the features contained in the second part DT12 of the image information indicating text.

[0053] The color filling unit 352 fills in the missing elements of the first portion DT11 to generate an image. The color filling unit 352 has a pixel filling unit 352a and an image processing unit 352b.

[0054] The pixel filling unit 352a fills in missing elements. Specifically, the pixel filling unit 352a performs, for example, a color matching process. The color matching process compares and analyzes the colors of the first portion DT11 and the second portion DT12 of the image information, and predicts the color of the first portion DT11. The pixel filling unit 352a also performs a gradation identification process. The gradation identification process analyzes the color gradation of the second portion DT12 of the image information, and predicts the color gradation of the first portion DT11. The pixel filling unit 352a fills in missing elements in the first portion DT11 based on the color and gradation predicted by the color matching process and the gradation identification process. The color matching process and the gradation identification process may be configured using a CNN model. By using the CNN model, the color and gradation of the first portion DT11 can be predicted by extracting the features of the second portion DT12. The CNN model is stored in the storage unit 40 as a second trained model 42, as shown in FIG. 2.

[0055] The image processing unit 352b predicts the shape of the missing portion and generates the missing portion. Specifically, the image processing unit 352b analyzes the second part DT12 and generates the shape of the missing portion. The image processing unit 352b executes, for example, a shape fitting process and a shape matching process. The shape fitting process is a process for predicting the shape of the first part DT11 from the shape of the second part DT12. The shape fitting process may be configured using a CNN model. By using the CNN model, the shape fitting process can predict the shape of the first part DT11 by extracting local shape features of the second part DT12.

[0056] The shape matching process is a process for analyzing the similarity between the two shapes of the first part DT11 and the second part DT12. The shape matching process analyzes the similarity between the shapes and extracts shape features such as the contours of the shape to predict the shape of the first part DT11.

[0057] The image processing unit 352b can fill in the missing elements based on the shape of the first part DT11 predicted by the shape fitting process and the shape matching process.

[0058] Continuing with reference to FIGS. 2 to 6, the processing of the control unit 30 will be described in detail. FIG. 5 shows first read data SC1 and first image data OP1. The first read data SC1 includes first image information DT1 and second image information DT2. The first image information DT1 is image information showing text. The first image information DT1 is shown as "AAAAAA...". "AAAAAA..." is an example and shows a sentence. The second image information DT2 is image information showing text. The second image information DT2 is shown as "SAMPLE DOCUMENT". The description in the second image information DT2 is an example and may be other description.

[0059] As shown in FIG. 5, the second image information DT2 is arranged to overlap the first image information DT1. As a result, a portion of the first image information DT1 cannot be recognized due to the second image information DT2. In other words, the overlap of the second image information DT2 on the first image information DT1 causes elements to be partially missing from the first image information DT1. Therefore, the first image information DT1 includes a first portion DT11 in which elements are partially missing, and a second portion DT12 in which no elements are missing. For example, the first portion DT11 overlaps with the "S" of the second image information DT2, and part of the "A" is missing. Furthermore, for example, the second portion DT12 does not have any missing elements because the second image information DT2 does not overlap. The first image data OP1 includes the first image information DT1.

[0060] As shown in FIG. 5, the control unit 30 performs an overlay deletion process on the first scanned data SC1 to generate first image data OP1. The overlay deletion process includes an analysis stage, a selection and separation stage, and a filling stage. The analysis stage is a stage in which the scanned data SC is analyzed and image information and layers containing image information are identified from the scanned data SC. The selection and separation stage is a stage in which the identified multiple layers are grouped and the overlay is deleted. The filling stage is a stage in which missing image information resulting from the deletion of the overlay is filled.

[0061] In the analysis stage, the extraction unit 31 and the recognition unit 32 mainly perform the processing. Specifically, the reading unit 10 reads the original M and outputs first read data SC1. When the reading unit 10 reads the first read data SC1, the extraction unit 31 extracts image information from the first read data SC1. In other words, the extraction unit 31 extracts first image information DT1 and second image information DT2 from the first read data SC1.

[0062] The recognition unit 4 recognizes the layers included in the first read data SC1. Specifically, the recognition unit 4 recognizes the first layer LY1 including the first image information DT1 and the second layer LY2 including the second image information DT2 using the layer recognition unit 321, the text recognition unit 322, and the layout pattern recognition unit 323.

[0063] 5, in a first portion DT11 in which image information indicating "SAMPLE DOCUMENT" and image information indicating "AAAAAA..." overlap, the layer identification unit 321 estimates that the second layer LY2 containing image information indicating "SAMPLE DOCUMENT" is overlapped above the first layer LY1 containing image information indicating "AAAAAA...". Based on the overlapping state of the first layer LY1 and the second layer LY2, the layer identification unit 321 estimates that the first portion DT11 in which an element is missing is formed in the image information indicating the text included in the first layer LY1.

[0064] Next, in the separation stage, the separation unit 33 mainly performs the processing. The separation unit 33 separates the second layer LY2 from the first layer LY1 and deletes the second layer LY2 from the first read data SC1. Furthermore, if there are multiple second layers LY2, the separation unit 33 groups the multiple second layers LY2. Then, the separation unit 33 deletes the second layer LY2 from the first read data SC1. As a result, only the first layer LY1 remains in the first read data SC1. Therefore, the overlay of the first read data SC1 is released.

[0065] Next, in the filling stage, the filling unit 35 mainly performs processing. The filling unit 35 fills the first portion DT11 with elements that are missing when the second image information DT2 of the second layer LY2 overlaps the first image information DT1. Specifically, the filling unit 35 fills the missing elements in the first portion DT11 using a text filling unit 351 and a color filling unit 352.

[0066] In this way, the first image data OP1 is generated. As a result, the overlay is deleted, there is no loss of the first image information DT1 due to the overlay, and the user can easily understand the first image information DT1.

[0067] Next, with reference to FIG. 6, a process of deleting the overlay from the second scanned data SC2 that includes image information representing an object and image information representing text will be described.

[0068] FIG. 6 shows second read data SC2 and second image data OP2. The second read data SC2 includes third image information DT3 and fourth image information DT4. The third image information DT3 is image information showing an object. The third image information DT3 is, for example, an image showing a "mountain." "Mountain" is an example, and the third image information DT3 may include any of natural objects, artificial objects, and living things. The fourth image information DT4 is image information showing text. The fourth image information DT4 is shown as "AAAAAA...". "AAAAAA..." is an example, and shows a sentence. The description in the fourth image information DT4 is an example, and other descriptions may be used.

[0069] As shown in FIG. 6, the fourth image information DT4 is arranged to overlap the third image information DT3. As a result, a portion of the third image information DT3 cannot be recognized due to the fourth image information DT4. In other words, the fourth image information DT4 overlaps the third image information DT3, causing elements to be partially missing from the third image information DT3. Therefore, the third image information DT3 includes a third portion DT31 with partially missing elements and a fourth portion DT32 with no missing elements. For example, the third portion DT31 overlaps with the "A" in the fourth image information DT4, and part of the "mountain" is missing. Furthermore, for example, the fourth portion DT32 is not missing because the fourth image information DT4 does not overlap it. The second image data OP2 includes the third image information DT3.

[0070] As shown in Fig. 6, the control unit 30 performs an overlay deletion process on the second scanned data SC2 to generate second image data OP2. The overlay deletion process is the same as the process described with reference to Fig. 5, so a detailed description will be omitted. The second image data OP2 is generated by the overlay deletion process. As a result, the overlay is deleted, and there is no loss of the third image information DT3 due to the overlay, making it easier for the user to understand the third image information DT3.

[0071] Next, the overlay deletion process executed by the control unit 30 will be described in more detail with reference to Figures 2 to 7. Figure 7 is a flowchart of the overlay deletion process executed by the control unit 30. The overlay deletion process shown in Figure 7 includes steps S1 to S14.

[0072] In step S1, the control unit 30 controls the reading unit 10 so that the reading unit 10 reads the document M, generates read data SC, and outputs the data to the control unit 30. The process proceeds to step S2.

[0073] In step S2, the control unit 30 executes the trimming instruction received by the operation unit 122. The process proceeds to steps S4 to S7.

[0074] After step S2, step S5, step S6, or step S8, in step S4, the layer identification unit 321 of the identification unit 32 identifies the layer included in the read data SC. The process proceeds to step S9.

[0075] After step S2, step S4, step S6, or step S8, in step S5, the text identifying section 322 of the identifying section 32 identifies text image information from the read data SC. The process proceeds to step S9.

[0076] After step S2, step S4, step S5, or step S8, in step S6, the arrangement pattern identification section 323 of the identification unit 32 identifies the arrangement pattern of the image information in the read data SC. The process proceeds to step S9.

[0077] After step S2, after step S4, after step S5, or after step S6, in step S7, the recognition unit 32 corrects the color of the read data SC. The process proceeds to step S8.

[0078] In step S8, the identifying unit 32 identifies the color of the read data SC, and the process proceeds to step S9.

[0079] After step S4, step S5, step S6, or step S8, the separation unit 33 groups the layers in step S9, and the process proceeds to step S10.

[0080] In step S10, the separation unit 33 separates the second layer LY2 from the first layer LY1 and deletes the second layer LY2. The process proceeds to step S11.

[0081] After step S10 or step S13, in step S11, the language is identified by the typography analysis processing unit 351a, the character detail analysis processing unit 351b, and the character recognition unit 351c of the filling unit 35. The process proceeds to step S12.

[0082] In step S12, the filling unit 35 fills the gaps in the first portion DT11. The process proceeds to step S14.

[0083] After step S10 or after step S12, in step S13, the filling unit 35 fills in the missing color in the first portion DT11. The process proceeds to step S14.

[0084] After step S12 or step S13, in step S14, the control unit 30 determines whether or not there is an overlay in the read data SC. If there is an overlay in the read data SC, the process returns to steps S4 to S7. If there is no overlay in the read data SC, the process ends.

[0085] Steps S4 to S8 are processes in the analysis stage. Steps S9 and S10 are processes in the separation stage. Steps S11 to S13 are processes in the filling stage. Steps S4 to S7 may be performed simultaneously. Furthermore, steps S4 to S7 may be performed starting from any step among steps S4 to S9. Steps S11 to S13 may be performed simultaneously. Furthermore, steps S11 to S13 may be performed starting from any step among steps S11 to S13.

[0086] [Variations] Next, a modified example of image forming apparatus 300 of this embodiment will be described with reference to Figures 1 to 6 and 8. The modified example executes an overlay deletion process that is different from the overlay deletion process executed by image forming apparatus 300 of this embodiment. Below, parts that are the same as in this embodiment will be omitted, and only the parts that are different will be described.

[0087] Fig. 8 is a flowchart of the overlay deletion process executed by the image forming apparatus 300 of the modified example. The overlay deletion process shown in Fig. 8 includes steps S31 to S45. Steps S31 and S32 shown in Fig. 8 are the same as steps S1 and S2 shown in Fig. 7. Steps S34 to S44 shown in Fig. 8 are the same as steps S4 to S14 shown in Fig. 7.

[0088] After step S32, in step S33, the control unit 30 acquires an instruction to select a specific area from the read data SC accepted by the operation unit 122. The process proceeds to step S33.

[0089] When a specific region is selected, the overlay removal process is performed in the specific region. That is, the identification unit 32 identifies the first layer LY1 in the specific region and the second layer LY2 in the specific region. Then, the separation unit 33 separates the first layer LY1 in the specific region from the second layer LY2 in the specific region and removes the second layer LY2 from the specific region. Therefore, the overlay in the specific region is removed. As a result, it becomes possible to recognize image information that would not have been recognized if image information had been overlapped in the specific region.

[0090] After step S44, in step S45, the control unit 30 determines whether the termination condition is satisfied. If the termination condition is satisfied (Yes in step S45), the process ends. If the termination condition is not satisfied (No in step S45), the process returns to step S33.

[0091] The embodiments of the present invention have been described above with reference to the drawings. However, the present invention is not limited to the above embodiments and can be embodied in various forms without departing from the spirit of the present invention. The drawings mainly show each component in a schematic manner to facilitate understanding, and the thickness, length, number, spacing, etc. of each component shown in the drawings may differ from the actual ones due to the convenience of creating the drawings. Furthermore, the speed, material, shape, dimensions, etc. of each component shown in the above embodiments are merely examples and are not particularly limited, and various modifications are possible within a range that does not substantially deviate from the configuration of the present invention. [Industrial Applicability]

[0092] The present invention provides an image reading device and an image forming device, and has industrial applicability. [Explanation of symbols]

[0093] 4: Identification unit 10: Reading unit 32: Identification unit 33: Separation section 35: Filling section 41: Model 42: Model 100: Image reader 220: Image forming unit 300: Image forming device DT1: First image information DT11: 1st part DT12: 2nd part DT2: Second image information LY1: First layer LY2: Second layer M:Manuscript P: Recording medium SC: Read data

Claims

1. a reading unit that reads an original and outputs read data; an identification unit that identifies, for each of a plurality of pieces of image information included in the read data, a layer on which the image information is arranged; a separation unit that separates the plurality of layers identified by the identification unit and deletes layers other than the specific layers; An image reading device comprising:

2. the plurality of pieces of image information included in the read data include first image information and second image information different from the first image information, the identification unit identifies a first layer on which the first image information is arranged and a second layer on which the second image information is arranged, The image reading device according to claim 1 , wherein the separating unit separates the first layer from the second layer and deletes the second layer from the read data.

3. a receiving unit that receives an instruction to select a specific area from the read data; the identification unit identifies a first layer in the specific region and a second layer in the specific region; The image reading device according to claim 2 , wherein the separating unit separates the first layer in the specific region from the second layer in the specific region, and deletes the second layer from the specific region.

4. the identification unit identifies the first layer and the second layer using a neural network model obtained in advance by machine learning; The image reading device according to claim 3 , wherein the neural network model detects how the first image information and the second image information overlap as depth information.

5. the first image information includes a first portion in which elements are partially missing and a second portion in which the elements are not missing, a filling section that fills the missing element into the first section; The image reading device according to claim 4 , further comprising:

6. the filling unit predicts the missing elements using a neural network model obtained in advance by performing machine learning; The image reading device according to claim 5 , wherein the neural network model analyzes a pattern of missing elements based on the first portion.

7. The image reading device according to any one of claims 1 to 6, an image forming unit that forms an image on a recording medium; An image forming apparatus comprising:

Images