JP2019161623A - Image processing device and computer program - Google Patents

Abstract

To specify character pixels with high accuracy even when an image includes dots.SOLUTION: An image processing device, using target image data, determines for each pixel whether each of a plurality of pixels in the target image is a character pixel candidate or not, thereby determining a plurality of candidate pixels, executing pre-processing including expansion processing on the target image data, generating pre-processed image data, and determining whether each of a plurality of target areas arranged on the image corresponding to the pre-processed image data satisfies a specific determination condition or not. Thereby, a non-character area in the target image is specified, and among the plurality of pixels in the target image, pixels that are determined as candidate pixels and in an area different from the non-character area are specified as the character pixels.SELECTED DRAWING: Figure 2

Description

  The present specification relates to image processing on image data, and more particularly to image processing for specifying character pixels in an image.

  In image data generated by reading a printed material using an image sensor, halftone dots included in the printed material appear in the image indicated by the image data. The pixels constituting such a halftone dot are easily specified by mistake as character pixels when specifying character pixels in the image.

  The image processing apparatus disclosed in Patent Literature 1 performs edge determination for determining whether each pixel is an edge and halftone dot determination for determining whether each pixel is a halftone dot. The image processing apparatus specifies a pixel that is an edge and not a halftone dot as a pixel indicating a character.

JP-A-6-164928 JP 2003-189090 A JP 2017-135609 A

  In this way, even for an image including halftone dots, for example, there is a technology that can accurately specify character pixels by suppressing erroneously specifying character pixels in an area without characters due to the halftone dots. It was sought after.

  The present specification discloses a technique that can accurately specify character pixels in a target image even in an image including halftone dots.

  The technology disclosed in the present specification has been made to solve at least a part of the above-described problems, and can be realized as the following application examples.

Application Example 1 In an image processing apparatus, an image acquisition unit that acquires target image data indicating a target image, and each of a plurality of pixels in the target image is a character pixel using the target image data. A candidate determining unit that determines a plurality of candidate pixels by performing a preprocessing including a dilation process on the target image data by determining whether or not each of the target image data is a preprocessed A pre-processing unit that generates image data; and determining whether each of a plurality of target regions arranged on an image corresponding to the pre-processed image data satisfies a specific determination condition. A non-character area specifying unit that specifies a non-character area in an image, and a pixel that is determined to be the candidate pixel among a plurality of pixels in the target image and is in a different area from the non-character area The character pixel And and a character identifying unit that identifies, the image processing apparatus.

  According to the above configuration, the non-character region in the target image is specified using the preprocessed image data that has been subjected to the preprocessing including the expansion process. In the preprocessed image data, the pixels constituting the halftone dot are likely to appear as a lump object by the expansion process. As a result, the area including the pixels constituting the halftone dots in the target image can be accurately identified as the non-character area. Then, a pixel in an area different from the non-character area is specified as a character pixel. As a result, for example, even if the pixel constituting the halftone dot is erroneously determined to be a candidate pixel by the judgment for each pixel, the pixel constituting the halftone dot is included by the judgment using the preprocessed image data. By determining that the region is a non-character region, it is possible to prevent the pixels constituting the halftone dots from being erroneously specified as character pixels. Therefore, even in an image including halftone dots, character pixels in the target image can be specified with high accuracy.

  According to the above configuration, different image processing is executed on the value of the character pixel and the value of the pixel different from the character pixel, so that appropriate image processing on the target image data can be realized.

  The technology disclosed in the present specification can be realized in various forms. For example, a multifunction device, a scanner, a printer, an image processing method, a function of these devices, or a computer for realizing the above method The present invention can be realized in the form of a program, a recording medium on which the computer program is recorded, and the like.

2 is a block diagram illustrating a configuration of a multifunction peripheral 200 that is an example of an image processing apparatus. It is a flowchart of an image process. It is a 1st figure which shows an example of the image used by image processing. It is a figure which shows an example of the image used by a character specific process. It is a flowchart of the 1st binary image data generation process. It is a flowchart of an edge emphasis process. It is explanatory drawing of R component image data after smoothing, and R component image data after edge enhancement. It is a figure which shows an example of the tone curve for a level correction process. It is a flowchart of the 2nd binary image data generation process. It is explanatory drawing of the minimum component value and the maximum component value of scan data. It is a 2nd figure which shows an example of the image used for an image process. It is a flowchart of a non-character area | region determination process. It is a figure which shows an example of the image used by a non-character area | region determination process. It is a 1st figure explaining an expansion process and a contraction process. It is the 2nd figure explaining expansion processing and contraction processing.

A. Example:
A-1: Configuration of MFP 200 The embodiment will be described based on examples. FIG. 1 is a block diagram illustrating a configuration of a multifunction peripheral 200 that is an example of an image processing apparatus. The multifunction device 200 includes a CPU 210 that is a processor that controls the image processing apparatus, a volatile storage device 220 such as a DRAM, a nonvolatile storage device 230 such as a flash memory and a hard disk drive, a display unit 240 such as a liquid crystal display, An operation unit 250 including a touch panel and buttons superimposed on a liquid crystal display, an interface (communication IF) 270 for communicating with an external device such as a user terminal device 100, a print execution unit 280, and a read execution unit 290 It is equipped with.

  The reading execution unit 290 generates scan data by optically reading a document using a one-dimensional image sensor under the control of the CPU 210. Under the control of the CPU 210, the print execution unit 280 uses a plurality of types of toners, specifically, cyan (C), magenta (M), yellow (Y), and black (K) toners as color materials. An image is printed on a printing medium such as paper by a laser method. Specifically, the print execution unit 280 exposes the photosensitive drum to form an electrostatic latent image, and attaches toner to the electrostatic latent image to form a toner image. The print execution unit 280 transfers the toner image formed on the photosensitive drum to a sheet. In the modification, the print execution unit 280 may be an ink jet type print execution unit that discharges ink as a color material to form an image on a sheet.

  The volatile storage device 220 provides a buffer area for temporarily storing various intermediate data generated when the CPU 210 performs processing. The non-volatile storage device 230 stores a computer program PG. The computer program PG is a control program that causes the CPU 210 to control the multifunction device 200. In this embodiment, the computer program PG is provided in a form stored in advance in the non-volatile storage device 230 at the time of manufacture of the multifunction device 200. Instead of this, the computer program PG may be provided in a form downloaded from a server, or may be provided in a form stored in a DVD-ROM or the like. The CPU 210 can execute image processing to be described later by executing the computer program PG.

A-2: Image Processing FIG. 2 is a flowchart of image processing. This image processing is executed, for example, when the user places a document on the document table of the reading execution unit 290 and inputs a copy execution instruction. In this image processing, scan data generated by reading an original using the reading execution unit 290 is acquired, and print data indicating the original is generated using the scan data, so that a so-called copy of the original is generated. This is a process to be realized.

  In S <b> 10, the CPU 210 generates scan data as target image data by reading a document placed on a document table by a user using the reading execution unit 290. The document is, for example, a printed material on which an image is printed by the multifunction device 200 or a printer (not shown). The generated scan data is stored in the buffer area of the volatile storage device 220 (FIG. 1). The scan data includes a plurality of pixel values, and each of the plurality of pixel values represents the color of the pixel by an RGB color system color value (also referred to as an RGB value). That is, the scan data is RGB image data. The RGB value of one pixel includes, for example, three color component values of red (R), green (G), and blue (B) (hereinafter also referred to as R value, G value, and B value). It is out. In this embodiment, the number of gradations of each component value is 256 gradations.

  The scan data that is RGB image data includes three component image data (R component image data, G component image data, and B component image data) corresponding to the three color components constituting the RGB color system. Can be said. Each component image data is image data in which a value of one type of color component is used as a pixel value.

  In S15, the CPU 210 performs an enlargement process for enlarging the scan image indicated by the scan data at the enlargement ratio Lr on the scan data, and generates the enlarged scan data. The enlargement ratio Lr is input together with a copy execution instruction by the user, for example. The enlargement ratio Lr is, for example, a value in the range of 0.1 to 6 (10% to 600%), and the enlargement process at an enlargement ratio Lr of less than 1 is a process for reducing the size of the scanned image. The enlargement process at a larger enlargement ratio Lr is a process for increasing the size of the scan image. Specifically, the size of the scanned image is defined by the number of pixels in the vertical direction and the horizontal direction. Note that scan data before enlargement processing is also called original image data, and scan data that has already been enlarged is also called target image data. Hereinafter, when simply referred to as scan data, it means scan data that has been enlarged.

  FIG. 3 is a first diagram illustrating an example of an image used in image processing. FIG. 3A shows an example of the scan image SI indicated by the scan data. Scanned image SI includes a plurality of pixels. The plurality of pixels are arranged in a matrix along a first direction D1 and a second direction D2 orthogonal to the first direction D1.

  The scanned image SI in FIG. 3A includes a white background Bg1 indicating the background color of the original paper, objects Ob1 and Ob2 that are different from the two characters, and four characters Ob3 to Ob6. Yes. In this embodiment, the object different from the character is a photograph.

  In S20, the CPU 210 performs a character specifying process on the scan data. The character specifying process is a process of specifying a character pixel by classifying a plurality of pixels in the scanned image SI into a plurality of character pixels indicating characters and a plurality of non-character pixels not indicating characters. It is.

  By the character specifying process, for example, binary image data (also referred to as character specifying data) in which the value of the character pixel is “1” and the value of the non-character pixel is “0” is generated. FIG. 3B shows an example of the character specifying image TI indicated by the character specifying data. In the character specifying image TI, a plurality of pixels constituting the edges of the four characters Ob3 to Ob6 in the scan image SI are specified as character pixels Tp3 to Tp7. For relatively large characters, the pixels constituting the edge of the character are identified as character pixels, and for relatively small characters, the entire pixels constituting the character are identified as character pixels. Details of the character identification process will be described later.

  In S30, the CPU 210 performs halftone dot smoothing processing on the scan data to generate smoothed image data indicating a smoothed image. Specifically, the CPU 210 performs a smoothing process using a smoothing filter such as a Gaussian filter on each of the values of the plurality of non-character pixels included in the scan data, and the smoothing process has been performed. The values of a plurality of non-character pixels are calculated. The non-character pixel to be subjected to the smoothing process is specified with reference to the character specifying data generated by the character specifying process of S20. The CPU 210 generates smoothed image data including a plurality of character pixel values included in the scan data and a plurality of smoothed non-character pixel values.

  FIG. 3C shows a smoothed image GI indicated by the smoothed image data. The smoothed image GI includes a white background Bg1g and objects Ob1g to Ob6g obtained by smoothing the objects Ob1 to Ob6 in the scan image SI. Of these objects Ob1g to Ob6g and background Bg2g, portions other than the characters Ob3g to Ob6g (also referred to as non-character portions) are smoothed compared to the scanned image SI.

  In S40, the CPU 210 performs a character sharpening process on the smoothed image data to generate processed image data. Specifically, the CPU 210 performs a sharpening process such as an unsharp mask process or a process of applying a sharpening filter on each of the plurality of character pixel values included in the smoothed image data, The values of a plurality of character pixels that have been sharpened are calculated. The character pixel to be subjected to the sharpening process is specified with reference to the character specifying data generated by the character specifying process of S20. The CPU 210 then includes a plurality of non-character pixel values (smoothed non-character pixel values) included in the smoothed image data, and a plurality of sharpened character pixel values. To generate processed image data. Since the values of the plurality of character pixels included in the smoothed image data are not subject to smoothing processing, they are the same as the values of the plurality of character pixels included in the scan data. Therefore, it can be said that the character sharpening process in this step is performed on the values of a plurality of character pixels included in the scan data.

  FIG. 3D shows a processed image FI indicated by the processed image data. The processed image FI includes a white background Bg1f and objects Ob1f to Ob6f corresponding to the objects Ob1 to Ob6 in the scan image SI. Among these objects Ob1f to Ob6f, the edges of the characters Ob3f to Ob6f are sharpened as compared with the characters Ob3 to Ob6 in the scan image SI and the characters Ob3g to Ob6g in the smoothed image GI. Further, the edges of the photographs Ob1f and Ob2f are not sharpened.

  As can be understood from the above description, the objects Ob1f to Ob6f and the background Bg2f in the processed image FI include sharpened characters and smoothed non-characters.

  In S50, the CPU 210 executes a print data generation process for generating print data using the processed image data. Specifically, a color having a color component (C, M, Y, K components) corresponding to a color material used for printing is performed by performing color conversion processing on processed image data that is RGB image data. CMYK image data indicating a color for each pixel with a CMYK value that is a value is generated. The color conversion process is executed with reference to a known lookup table, for example. Halftone processing is performed on the CMYK value image data, and dot data indicating a dot formation state is generated for each color material used for printing and for each pixel. The dot formation state can take, for example, two types of states with and without dots, and four types of states with large dots, medium dots, small dots, and no dots. The halftone process is executed according to, for example, a dither method or an error diffusion method. The dot data is rearranged in the order used in printing, and print data is generated by adding a print command to the dot data.

  In S60, the CPU 210 executes the printing process and ends the image processing. Specifically, the CPU 210 supplies print data to the print execution unit 280 and causes the print execution unit 280 to print the processed image.

  According to the image processing described above, first image processing (specifically, edge sharpening processing) is performed on the values of a plurality of specified character pixels in the scan data (S40), A second image process (specifically, a halftone dot smoothing process) different from the first image process is performed on a plurality of non-character pixel values (S30), and processed image data is generated. The As a result, different image processing is executed for the character pixel value and the pixel value different from the character pixel, so that appropriate image processing for the scan data can be realized. In the modification, the character sharpening process of S40 may be executed first, and then the halftone dot smoothing process of S30 may be executed.

  More specifically, processed image data including a plurality of character pixel values that have undergone sharpening processing and a plurality of non-character pixel values that have undergone smoothing processing are generated (S30, S40). . As a result, processed image data indicating the processed image FI having a good appearance can be generated.

  For example, as shown in the processed image FI in FIG. 3D, in the processed image data, the value after the sharpening process is used as the value of the character pixel. As a result, since the characters of the processed image FI look sharp, for example, the appearance of the processed image FI to be printed can be improved.

  In the processed image data, a smoothed value is used as the value of a non-character pixel constituting an object different from a character such as a photograph in the processed image FI. As a result, for example, it is possible to suppress the appearance of halftone dots that cause moire in a portion different from the characters of the processed image FI, thereby suppressing the occurrence of defects such as moire in the processed image FI to be printed. it can. As a result, the appearance of the processed image FI to be printed can be improved. Moreover, since it is suppressed that the edge in a photograph is emphasized too much, the appearance of the image FI can be further improved.

  For example, a document used for generating scan data is a printed material on which an image is printed. For this reason, for example, a portion having a color different from white in the original forms halftone dots when viewed at the dot level for forming an image. A halftone dot includes a plurality of dots and a portion where no dot is arranged (a portion indicating the background color of the document). For this reason, halftone dots are shown at the pixel level in a portion having a color different from white in the original in the scanned image SI. The dots in the halftone dots are arranged with periodicity due to the influence of a dither matrix used at the time of printing an original. For this purpose, when printing is performed using scan data, the dot components of halftone dots existing in the original image (scanned image SI) before halftone processing and the dots of dots constituting the print image are displayed. Moire is likely to appear due to interference with periodic components. In the processed image FI of the present embodiment, the periodic component of the dot in the portion different from the edge in the original image (scanned image SI) is reduced by the smoothing process. As a result, when the processed image FI is printed using the processed image data, for example, it is possible to suppress the occurrence of moire in the processed image FI to be printed.

  In particular, in the image processing, print data is generated using the processed image data (S50). For example, appropriate print data that can suppress moire that is likely to occur in the processed image FI to be printed is generated. can do.

A-3: Character Identification Process The character identification process in S20 of FIG. 2 will be described. In S22, the CPU 210 executes a first binary image data generation process using the scan data, and generates first binary image data. The first binary image data is binary data indicating edge pixels and non-edge pixels. Here, the edge pixel indicated by the first binary image data is also referred to as a first edge pixel, and the non-edge pixel indicated by the first binary image data is also referred to as a first non-edge pixel. Details of the first binary image data generation processing will be described later.

  In S24, the CPU 210 executes the second binary image data generation process using the scan data, and generates the second binary image data. Similar to the first binary image data, the second binary image data is binary data indicating edge pixels and non-edge pixels. The second binary image data is generated by a process different from the first binary image data, and is different from the first binary image data. Here, the edge pixel indicated by the second binary image data is also referred to as a second edge pixel, and the non-edge pixel indicated by the second binary image data is also referred to as a second non-edge pixel. Details of the second binary image data generation processing will be described later.

  In S26, the CPU 210 executes a logical sum synthesis process for synthesizing the first binary image data generated in S22 and the second binary image data generated in S24. Binary image data (also referred to as edge specifying data) indicating edge pixels and non-edge pixels specified in (1) is generated. Specifically, the CPU 210 generates binary image data as edge specifying data by calculating the logical sum of each pixel of the first binary image data and the second binary image data. In other words, the CPU 210 includes a plurality of first edge pixels specified by the first binary image data and a plurality of second edge pixels specified by the second binary image data. A pixel group that includes a pixel group that does not include a pixel different from the first edge pixel and the second edge pixel is finally specified as a plurality of edge pixels. As a result, it is possible to accurately determine whether or not the pixel in the target image is an edge pixel using the first binary image data and the second binary image data. For example, it is possible to effectively reduce specific omission of edge pixels in the scan image SI.

  The edge pixels specified by the edge specifying data are pixels that are candidates for character pixels, and are also referred to as character candidate pixels. For example, in the edge specifying data, the value of a character candidate pixel (edge pixel in this embodiment) is “1”, and the value of a pixel that is not a character candidate pixel (non-edge pixel in this embodiment) is “0”. Binary image data.

  FIG. 4 is a diagram illustrating an example of an image used in the character identification process. FIG. 4A shows an example of the edge specifying image EI indicated by the edge specifying data. In the edge specifying image EI, a plurality of edge pixels constituting the edges Eg1 to Eg6 of the objects Ob1 to Ob6 in the scan image SI are specified as character candidate pixels. Thus, the edge indicated by the character candidate pixel includes the edge of the character. In addition, the edge includes an edge such as a thin line included in an object (for example, a photograph) different from the character.

  In S27, it is determined whether or not the enlargement ratio Lr in the enlargement process in S15 is less than the threshold value Lth. For example, the threshold value Lth of the enlargement ratio is 3 (300%), for example. When the enlargement ratio Lr is less than the threshold Lth (S27: YES), the CPU 210 advances the process to S28.

  In S28, the CPU 210 performs a non-character area determination process on the scan data to identify a non-character area. That is, binary image data (also referred to as non-character area data) indicating a plurality of pixels constituting a non-character area and a plurality of pixels not constituting a non-character area is generated. Although details will be described later, in the non-character area determination process, the CPU 210 uses the scan data and each of the plurality of target areas arranged in the scan image SI is an object (for example, a photograph) different from the character. Is determined for each target area. The CPU 210 generates non-character area data based on the determination result for each target area. The non-character area data is, for example, binary image data in which the value of a pixel constituting the non-character area is “1” and the value of a pixel not constituting the non-character area is “0”.

  FIG. 4B shows an example of the non-character area image AI indicated by the non-character area data. In this non-character area image AI, non-character areas AA1 and AA2 are specified in areas corresponding to photographs Ob1 and Ob2 in the scan image SI.

  In S29, the CPU 210 performs a synthesis process for synthesizing the edge specifying data generated in S26 and the non-character area data generated in S28 to show the character pixel and the non-character pixel as described above. Character specifying data (see FIG. 3B) is generated. Specifically, the CPU 210 sets, as character pixels, pixels that are not located in the non-character areas AA1 and AA2 specified by the non-character area data among the plurality of character candidate pixels specified by the edge specifying data. Identify. CPU210 specifies the pixel located in non-character area AA1 and AA2 and the pixel which is not a character candidate pixel among a plurality of character candidate pixels specified by edge specific data as a non-character pixel. In other words, among a plurality of pixels in the scanned image SI, pixels that are determined to be character candidate pixels in S22 to S26 and that are in a region different from the non-character region determined in S28 are: Identified as a character pixel. When the character specifying data is generated, the character specifying process is terminated.

  When the enlargement ratio Lr is equal to or greater than the threshold value Lth (S27: NO), the CPU 210 skips S28 and S29 and ends the character specifying process. In this case, the edge specifying data generated by the logical sum synthesis process in S26 is the final character specifying data. That is, in this case, the CPU 210 identifies the pixels determined to be character candidate pixels in S22 to S26 as character pixels. As a modification, the CPU 210 may execute the determination of S27 after executing the non-character area determination process of S28. That is, after S28, it may be determined whether or not the enlargement ratio Lr is less than the threshold value Lth. Also in this case, when the enlargement ratio Lr is less than the threshold value Lth, S29 is executed and character specifying data is generated. If the enlargement ratio Lr is greater than or equal to the threshold value Lth, S29 is skipped, and the edge specifying data generated by the logical sum synthesis process of S26 is used as final character specifying data.

A-4: First Binary Image Data Generation Processing The first binary image data generation processing in S22 of FIG. 2 will be described. FIG. 5 is a flowchart of the first binary image data generation process. In S100, the CPU 210 executes a smoothing process on each of the three component image data included in the scan data, that is, R component image data, G component image data, and B component image data. As a result, three smoothed component image data, that is, smoothed R component image data, smoothed G component image data, and smoothed B component image data are generated.

  The smoothing process is a process for smoothing the component image indicated by the component image data to be processed. The smoothing process of the present embodiment is a process of calculating the value of each smoothed pixel by applying a predetermined smoothing filter to the value of each pixel of the component image data to be processed. As the smoothing filter, for example, a Gaussian filter having a size of 7 vertical pixels × 7 horizontal pixels is used.

  In S110, edge enhancement processing is performed on each of the three smoothed component image data to obtain three edge-enhanced component image data, that is, edge-enhanced R-component image data, edge Emphasized G component image data and edge enhanced B component image data are generated.

  FIG. 6 is a flowchart of edge enhancement processing. Here, it is assumed that the smoothed R component image data is a processing target. The same processing is performed on the smoothed G component image data and the smoothed B component image data.

  In S200, the CPU 210 prepares canvas data for generating edge-enhanced R component image data in a memory (specifically, a buffer area of the volatile storage device 220). The canvas (initial image) indicated by the canvas data is an image having the same size as the scan image SI, that is, an image having the same number of pixels. The value of each pixel of the canvas data is a predetermined initial value (for example, 0).

  In S205, the CPU 210 selects one pixel of interest from a plurality of pixels in the smoothed R component image indicated by the smoothed R component image data.

  In S210, the CPU 210 calculates a mask value MV corresponding to the target pixel. The mask value MV smoothes the value TV of the target pixel using the value TV of the target pixel and the values of a predetermined number of surrounding pixels including four pixels adjacent to the target pixel in the vertical and horizontal directions. Calculated by processing. For this reason, the mask value MV is also called a smooth value. Specifically, an average value of the values of 100 pixels within a rectangular range of 10 pixels vertically × 10 pixels horizontally centered on the target pixel is calculated as a mask value MV corresponding to the target pixel.

  In S220, the CPU 210 calculates a difference ΔV between the value TV of the target pixel and the mask value MV corresponding to the target pixel (ΔV = (TV−MV)).

  In S230, the CPU 210 determines whether or not the difference ΔV is greater than or equal to a reference. Specifically, it is determined whether or not the difference ΔV is equal to or greater than a predetermined threshold value TH. The threshold value TH is a value of about 20 to 30 when the component value is a value of 256 gradations in the range of 0 to 255, for example.

  When the difference ΔV is greater than or equal to the reference (S230: YES), in S240, the CPU 210 calculates the sum of the target pixel value TV and the difference ΔV corresponding to the target pixel (TV + ΔV) as a processed value. Calculate as When the difference ΔV is less than the reference (S230: NO), the CPU 210 skips S240.

  In S245, the CPU 210 records the value of the target pixel in the canvas data prepared in S200. When S240 is executed, the sum of the target pixel value TV calculated in S240 and the difference ΔV corresponding to the target pixel is recorded in the canvas data as a processed value. When S240 is skipped, the value of the target pixel of the smoothed R component image data is recorded in the canvas data as it is.

  In S250, the CPU 210 determines whether or not all the pixels in the R component image have been processed as the target pixel. When there is an unprocessed pixel (S250: NO), the CPU 210 returns to S205 and selects an unprocessed pixel as a target pixel. When all the pixels have been processed (S250: YES), the CPU 210 ends the edge enhancement processing. At this time, edge-enhanced R component image data is generated.

  FIG. 7 is an explanatory diagram of smoothed R component image data and edge-enhanced R component image data. FIG. 7A shows a graph conceptually showing the R component image data before the smoothing process in S100 of FIG. FIGS. 7B and 7C show graphs conceptually showing smoothed R component image data and edge-enhanced R component image data, respectively. In each graph, the left portion conceptually shows a halftone dot region indicating a halftone dot, and the right portion conceptually shows an edge region indicating an edge of an object such as a character. The vertical axis of each graph indicates the value of the R component, and the horizontal axis indicates the position in a predetermined direction (for example, the first direction D1 in FIG. 3).

  R component image data before smoothing processing includes, for example, a plurality of valley portions C1 to C3 corresponding to a plurality of halftone dots and a plurality of gaps between the halftone dots in a halftone dot region, A plurality of peaks P1 and P2 appear (FIG. 7A). If the difference in R component values between the valleys C1 to C3 and the plurality of peaks P1 and P2 remains large, in the binarization process of S150 described later, the R Due to the difference in the component values, an edge pixel indicating a halftone dot is likely to be specified. A halftone dot area includes halftone dots when viewed from a pixel-level viewpoint (a micro viewpoint that can recognize halftone dots), but from an observer's viewpoint (a macro viewpoint that does not recognize halftone dots). It is a uniform area. For this reason, in this embodiment, the edge pixel due to the halftone dot should not be specified in the halftone dot region. This is because the halftone dot region is preferably smoothed in S30 of FIG. 2, and should not be sharpened in S40. This is because if the halftone dot edge is sharpened, the periodicity of the halftone dot becomes conspicuous, and therefore, moire becomes conspicuous when the image is printed. For example, edge pixels should not be identified in a uniform part in an object such as a photograph in the scanned image SI.

  In the smoothed R component image data, for example, in the halftone dot region, the difference in R component values between the plurality of valleys C1a to C3a and the plurality of peaks P1a and P2a is obtained by the smoothing process. It is sufficiently smaller than the R component image data before the smoothing process (FIG. 7B).

  Here, in the edge enhancement processing of the present embodiment, the effect of edge enhancement increases as the difference ΔV between the value TV of the target pixel and the mask value MV corresponding to the target pixel increases. For this reason, the effect of edge emphasis is reduced in a region where the difference in R component values is relatively small and flat, such as the halftone dot region in FIG. 7B. Further, in the edge enhancement processing of this embodiment, when the difference ΔV is less than the reference, the edge enhancement is not performed and the pixel value of the smoothed R component image data is used as it is (see FIG. 6). S230). As a result, in the R component image data that has undergone edge enhancement, for example, in the halftone dot region, a plurality of valleys C1b to C3b and a plurality of peaks P1b, The difference between the R component values of P2b and the R component image data subjected to the smoothing process is not large (FIG. 7C). That is, similarly to the smoothed R component image data, the edge-enhanced R component image data has R component values of a plurality of valleys C1b to C3b and a plurality of peaks P1b and P2b. The difference is sufficiently smaller than the R component image data before smoothing (FIG. 7A) (FIG. 7C).

  In the R component image data before the smoothing process, for example, in an edge region indicating an edge of an object such as a character, a fluctuation part E1 in which the value of the R component changes rapidly corresponding to the edge appears (FIG. 7 ( A)). In such a variation part E1, the larger the change in the value, the easier it is to identify the edge pixel indicating the edge of the object due to the difference in the value of the R component in the binarization process of S150 described later.

  In the smoothed R component image data, for example, in the edge region, the change in the value in the changing portion E1a is smaller than the R component image data before the smoothing processing due to the smoothing processing (slowly). (FIG. 7B).

  However, since the change in the value in the fluctuation part E1a corresponding to the edge of the object such as a character is sufficiently larger than the change in the value in the halftone dot region, it is again returned to a rapid change by the edge emphasis process. As a result, in the edge component-enhanced R component image data, the change in the value of the R component of the changing portion E1b in the edge region is larger than that in the smoothed R component image data (FIG. 7 ( C)). For this reason, in the edge component-enhanced R component image data, in the edge region, the change in the value in the variation portion E1b is comparable or abrupt compared to the R component image data before the smoothing process. (FIG. 7C).

  As can be seen from the above description, in this embodiment, the smoothing process (S100) and the edge enhancement process (S110) are executed in this order for each component image data. It can suppress that the edge pixel to show is specified, and can promote specifying the edge pixel which shows the edge of objects, such as a character. As a result, it is possible to appropriately specify a plurality of edge pixels in the scan image SI.

  When three enhanced component image data corresponding to the three color components of R, G, and B are generated, 120 of FIG. 5 uses the three enhanced component image data. Then, luminance image data is generated. The luminance image data is data indicating the luminance of a plurality of pixels in the enhanced image indicated by the three enhanced component image data. Specifically, the CPU 210 calculates the luminance Y of each pixel using the R value, G value, and B value of each pixel acquired from the three emphasized component image data. The luminance Y is, for example, a weighted average of the above three components, and specifically can be calculated using an equation of Y = 0.299 × R + 0.587 × G + 0.114 × B. The luminance image data is single component image data composed of one kind of component value (value indicating luminance). The luminance component data includes the luminance Y based on the value (RGB value) of the corresponding pixel of the scan data for each pixel.

  In S <b> 130, the CPU 210 generates edge extraction data by executing edge extraction processing for extracting edges in the luminance image indicated by the luminance image data with respect to the generated luminance image data. Specifically, the CPU 210 calculates the edge strength of each pixel by applying a known edge extraction filter, for example, a Sobel filter, to the value of each pixel of the luminance image data. The CPU 210 generates edge extraction data having these edge strengths as values of a plurality of pixels.

  In S140, the CPU 210 performs level correction processing on the edge extraction data to generate corrected edge extraction data. The level correction process is a correction process for enlarging a specific range within a range of gradation values that can be taken by the pixel value of the edge extraction data (in this embodiment, a range of 0 to 255).

  FIG. 8 is a diagram illustrating an example of a tone curve for level correction processing. Specifically, the CPU 210 applies the tone curve of FIG. 8 to each pixel of the edge extraction data. As a result, all values above the threshold Vb (for example, 245) are converted to the maximum value (255), and all values below the threshold Va (for example, 10) are converted to the minimum value (0). A range larger than the threshold value Va and less than the threshold value Vb is expanded to a range of 0 to 255. As described above, since the range including the binarization threshold (the range greater than the threshold Va in FIG. 8 and less than the threshold Vb) is expanded before the binarization processing in S150 described later, the binarization accuracy is increased. Can be improved.

  In S150, the CPU 210 performs binarization processing on the corrected edge extraction data to generate binary image data. For example, in the edge image data, the CPU 210 classifies pixels having a pixel value (that is, edge strength) equal to or higher than a threshold value (for example, 128) as edge pixels, and sets pixels whose pixel value is less than the threshold value to non-pixel values. Classify into edge pixels. In the binary image data, as described above, the value of the edge pixel is “1”, and the value of the non-edge pixel is “0”.

  According to the first binary image data generation process described above, as described with reference to FIG. 7, the scan image SI is obtained by performing the smoothing process on each of the plurality of component image data. It is possible to reduce the characteristics of halftone dots appearing in the image. Further, as described with reference to FIG. 7, the edge enhancement process is performed on each of the plurality of smoothed component image data, thereby the scan image SI smoothed by the smoothing process. Can be properly emphasized. As a result, it is possible to appropriately specify the edge pixels in the scan image SI while suppressing the specification of the edge pixels due to the halftone dots.

  Furthermore, since luminance image data is used as the single component image data, a plurality of edge pixels in the scan image SI can be more appropriately identified. For example, a halftone dot often has C, M, and Y primary colors used for printing, but such a difference between a plurality of primary colors is relatively large in each of R, G, and B component image data. However, the luminance image data is relatively small. For this reason, it can suppress appropriately that the edge pixel resulting from a halftone dot is specified by using luminance image data. In addition, for the sake of readability of characters, there is often a relatively large difference in luminance between the character color and the background color. For this reason, by using the luminance image data, it is possible to appropriately specify the edge pixel indicating the edge of an object such as a character.

  Further, in the edge enhancement processing of FIG. 6, calculation of a mask value (also referred to as a smooth value) corresponding to the target pixel (S210), and calculation of a difference ΔV between the value TV of the target pixel and the mask value corresponding to the target pixel ( A so-called unsharp mask process including S220) and calculation of the sum (TV + ΔV) of the value TV of the target pixel and the corresponding difference ΔV (S240) is executed. As a result, since the edge of the scan image SI can be appropriately emphasized, the specific omission of the edge pixel to be specified can be suppressed. As a result, the edge pixels in the scan image SI can be specified more appropriately.

  Furthermore, in the edge enhancement processing of FIG. 6, among the plurality of pixels in the scan image SI, pixels having a corresponding difference ΔV that is greater than or equal to the reference are subjected to unsharp mask processing, and the difference ΔV is less than the reference. Are not subjected to unsharp mask processing (S230, 240). As a result, as described with reference to FIG. 7, it is possible to further suppress the enhancement of the value difference between the pixels due to the halftone dots of the scan image SI. Further suppression is possible. The edges of objects such as characters can be emphasized appropriately. Therefore, the edge pixel in the scan image SI can be specified more appropriately.

A-5: Second Binary Image Data Generation Processing The second binary image data generation processing in S24 of FIG. 2 will be described. FIG. 9 is a flowchart of the second binary image data generation process. In S300, the CPU 210 generates minimum component data using the scan data. Specifically, the CPU 210 acquires the minimum component value Vmin from each of a plurality of pixel values (RGB values) included in the scan data. The minimum component value Vmin is a minimum value among a plurality of component values (R value, G value, B value) included in the RGB value. The CPU 210 generates image data having these minimum component values Vmin as values of a plurality of pixels as minimum component data. The minimum component data is image data indicating an image having the same size as the scanned image SI. Each of the values of the plurality of pixels included in the minimum component data is a minimum component value Vmin of the corresponding pixel value (RGB value) of the scan data.

  FIG. 10 is an explanatory diagram of the minimum component value and the maximum component value of the scan data. 10A to 10E, as an example of RGB values, RGB values of cyan (C), magenta (M), yellow (Y), black (K), and white (W) are shown as bar graphs. It is shown in the figure. As shown in FIG. 10, RGB values (R, G, B) of C, M, Y, K, and W are respectively (0, 255, 255), (255, 0, 255) (255, 255, 0), (0, 0, 0), (255, 255, 255).

  As described above, the luminance Y of these RGB values can be calculated using, for example, the equation Y = 0.299 × R + 0.587 × G + 0.114 × B. The luminances of C, M, Y, K, and W (represented by values from 0 to 255) are about 186, 113, 226, 0, and 255, and are different values (FIG. 10). On the other hand, the minimum component values Vmin of C, M, Y, K, and W are 0, 0, 0, 0, and 255 as shown in FIG. 10, and are the same values except for white (W). .

  FIG. 11 is a second diagram illustrating an example of an image used for image processing. FIG. 11A is an enlarged view of the above-described halftone dot region in the scan image SI. For example, in the example of FIG. 11A, the halftone dot region in the scan image SI includes a plurality of M dots MD and a plurality of Y dots YD. Here, for the sake of explanation, it is assumed that an image showing M dots MD is a uniform image having a magenta primary color, and an image showing Y dots YD is a uniform image having a yellow primary color.

  FIG. 11B shows an example of the minimum component image MNI indicated by the minimum component data. This minimum component image MNI corresponds to the scan image SI of FIG. In the minimum component image MNI, the value of the pixel in the region MDb corresponding to the Y dot MD of the scan image SI and the value of the pixel in the region YDb corresponding to the Y dot YD are the same. FIG. 11C shows a luminance image YI indicated by luminance image data indicating the luminance of each pixel as a comparative example. This luminance image YI corresponds to the scanned image SI in FIG. In the luminance image YI, unlike the minimum component image MNI, the value of the pixel in the region MDd corresponding to the M dot MD of the scan image SI is different from the value of the pixel in the region YDd corresponding to the Y dot YD. .

  As can be understood from the above description, in the minimum component image MNI, the difference between the values of the plurality of pixels corresponding to the portion where the C, M, Y, and K dots in the document are formed in the scanned image SI is as follows. It becomes smaller than the luminance image YI. In the minimum component image MNI, in the scanned image SI, the pixel value of the ground color area corresponding to the area indicating the ground color (white of the paper) in the document is the pixel value corresponding to the portion where the dot is formed. Bigger than.

  In S310, the CPU 210 executes a smoothing process for smoothing the minimum component image MNI indicated by the minimum component data on the generated minimum component data, and generates smoothed minimum component data. Specifically, the CPU 210 applies a predetermined smoothing filter to the value of each pixel of the minimum component data, in this embodiment, a Gaussian filter of 5 vertical pixels × 5 horizontal pixels, so that each smoothed data The pixel value is calculated. The smoothed minimum component data includes, for each pixel, the smoothed value generated by the above-described processing based on the corresponding pixel value (RGB value) of the scan data.

  In S320, the CPU 210 executes edge extraction processing for extracting the edge in the smoothed minimum component image MNI indicated by the smoothed minimum component data with respect to the smoothed minimum component data, Edge extraction data is generated. Specifically, the CPU 210 calculates the edge strength by applying the same Sobel filter as the processing of S130 of FIG. 5 to the value of each pixel of the smoothed minimum component data. The CPU 210 generates edge extraction data having these edge strengths as values of a plurality of pixels.

  In S330, the CPU 210 performs level correction processing on the edge extraction data to generate corrected edge extraction data. The level correction process is the same as the process of S140 in FIG. In S340, the CPU 210 performs binarization processing similar to the processing in S150 of FIG. 5 on the edge extraction data that has been subjected to correction processing, and generates binary image data. In the binary image data, as described above, the value of the edge pixel is “1”, and the value of the non-edge pixel is “0”.

  According to the second binary image data generation process described above, the edge extraction process is executed on the minimum component data to generate edge extraction data (S320). A plurality of edge pixels of the scan image SI are specified by executing an edge pixel specifying process including a process of binarizing the edge extraction data (S340) (S330, S340, S26 in FIG. 2). . In the minimum component data, as described with reference to FIG. 11, the difference in values between pixels can be suppressed in the halftone dot region. Therefore, when the edge pixel is subsequently specified, the edge pixel due to the halftone dot is determined. Can be specified. Therefore, it is possible to appropriately specify the edge pixel in the scan image SI.

  More specifically, there are five types of elements that constitute the halftone dot region: C, M, Y, and K dots, and the ground color (white) of the paper. In the minimum component data, it is possible to suppress a difference in values between pixels indicating four types of elements among these elements. As a result, when the minimum component data is used, it is possible to prevent the edge pixel indicating the edge of the halftone dot from being specified.

  On the other hand, in many cases, one of the character color and the background color has a dark color and the other has a light color. For this reason, one of the characters and the background includes a relatively large portion indicating the background color (white) of the paper, and the other includes a relatively large portion indicating C, M, Y, and K dots. There are many. As shown in FIG. 10, in the minimum component data, between the value of the pixel indicating the C, M, Y, and K dots and the value of the pixel indicating the ground color (white) of the paper, There is a big difference. For this reason, if edge pixels are specified using the minimum component data, there is a high possibility that the edge pixels constituting the edge of the character can be specified appropriately. In particular, yellow (Y) has a lower density (higher brightness) than C, M, and K. For this reason, if there is a yellow character on the background of the paper background (white), even if the luminance image data is binarized, the edge pixels constituting the edge of the yellow character are appropriately specified. There are cases where it is impossible. In this embodiment, even in such a case, the edge pixels constituting the edge of the yellow character can be appropriately specified. For this reason, in addition to specifying edge pixels using luminance image data, edge pixels such as characters that cannot be specified by luminance image data alone are specified by specifying edge pixels using minimum component data. obtain. As a result, it is possible to improve the accuracy of specifying edge pixels in the scan image SI.

  Further, a smoothing process is executed on the minimum component data before the edge extraction process (S310). As a result, the smoothing process can further suppress the difference in value between the pixels in the halftone dot region in the minimum component image MNI. For example, in a halftone dot region in the scanned image SI, a portion indicating a dot is not necessarily C, M, Y, due to overlapping of C, M, Y, K dots or blurring at the time of reading by the reading execution unit 290. , K does not have the primary color. For this reason, in the minimum component image MNI, values between a plurality of pixels indicating C, M, Y, and K dots are small but not zero. By the smoothing process, the value difference between the pixels can be further reduced. As a result, it can further be suppressed that the edge pixel caused by the halftone dot is specified. Also, in the second binary image data generation process, as in the first binary image data generation process, the level correction process (S330) is executed, so that the edge pixel specifying accuracy in the scan image SI is improved. Can be improved.

  As described above, in the above embodiment, an edge pixel is finally specified using two types of single component image data, that is, luminance image data and minimum component data (S22 to S22 in FIG. 2). S26). Since a plurality of edge pixels in the scan image SI are specified using two types of single component image data generated by using different processes in this way, a plurality of edge pixels in the scan image SI are identified. Specific leakage can be suppressed. For example, when there is a yellow character on a white background, the difference in luminance between white and yellow is relatively small, so it is difficult to identify edge pixels that constitute the edge of the character using luminance image data. It is. On the other hand, as can be seen from FIGS. 10C and 10E, in the minimum component data, a large difference between white and yellow appears. Therefore, when there is a yellow character on the white background, the minimum component Using data, it is easy to specify edge pixels that constitute the edge of the character. Also, for example, when there is a yellow character in the background of magenta, the minimum component data does not show a difference between magenta and yellow, so the edge pixel that forms the edge of the character using the minimum component data It is difficult to specify. On the other hand, when there is a yellow character on the magenta background, the brightness difference between magenta and yellow is relatively large, so the edge pixels that make up the edge of the character are identified using the brightness image data. It's easy to do.

  Further, since the single component image data used in addition to the luminance image data is the minimum component data, as described above, it is possible to suppress the specification of the edge pixels due to the halftone dots in the scan image SI.

A-6: Non-Character Area Determination Process The non-character area determination process in S28 of FIG. 2 will be described. FIG. 12 is a flowchart of the non-character area determination process. FIG. 13 is a diagram illustrating an example of an image used in the non-character area determination process.

  The CPU 210 performs preprocessing of S405 to S420 on the scan data, and generates preprocessed image data.

  In S405, the CPU 210 generates minimum component data using the scan data, as in S300 of FIG. 9 described above. When the minimum component data generated in S300 of FIG. 9 is stored in the volatile storage device 220 or the nonvolatile storage device 230, the minimum component data may be used.

  In S410, the CPU 210 binarizes the minimum component data to generate binary image data. For example, in the minimum component data, the CPU 210 classifies pixels whose pixel value is a threshold value (for example, 128) or more as a specific target pixel, and classifies pixels whose pixel value is a threshold value or more as a non-specific target pixel. To do. In the binary image data, for example, the value of the specific target pixel is “1”, and the value of the non-specific target pixel is “0”. As described above, in the minimum component data, as described with reference to FIGS. 10 and 11, a plurality of scan images SI corresponding to portions where C, M, Y, and K dots are formed in the document. Is smaller than the pixel value of the ground color area corresponding to the area indicating the ground color (paper white) in the document. Therefore, in the binary image data generated in this step, a plurality of pixels corresponding to a portion where dots are formed in the document are classified, and a plurality of pixels in the ground color area are classified as non-specific target pixels. In this way, by binarizing the minimum component data, it is possible to accurately specify object pixels that constitute an object (image formed by dots) printed on a document.

  Here, since the binary image data generated in this step is image data before execution of the expansion processing of S415 described later, it is also referred to as pre-expansion image data, and the image indicated by the pre-expansion image data. Also referred to as pre-expansion image OI. FIG. 13A shows an example of the pre-expansion image OI. In the pre-expansion image OI, the identification target pixels Op1 to Op6 corresponding to the objects Ob1 to Ob6 in the scan image SI are identified. Since the characters of the document are often relatively dark, the specific target pixels Op3 to Op6 are specified with almost no gap in the area corresponding to the characters in the pre-expansion image OI. Since dots are formed in the entire photograph area of the document, the specific target pixels Op1 and Op2 are identified in the entire area corresponding to the photograph in the pre-expansion image OI. However, since the photograph can include both dark and light colors, the region corresponding to the photograph in the pre-expansion image OI includes a white portion in the pre-expansion image OI of the non-specific target pixel (FIG. 13A). ) Are also distributed throughout.

  In S415 and S420, the CPU 210 performs expansion processing and contraction processing on the pre-expansion image data to generate pre-processed image data. 14 and 15 are diagrams illustrating the expansion process and the contraction process.

  In S415, the CPU 210 performs one expansion process on the pre-expansion image data. FIG. 14A shows a partial image PI1 including thin lines corresponding to the characters Ob3 to Ob6 in the scan image SI in the pre-expansion image OI. FIG. 15A shows a partial image PI2 that is a part of an area corresponding to the photographs Ob1 and Ob2 in the scan image SI in the pre-expansion image OI. The hatched portion in FIG. 14A indicates the specific target pixel Opt corresponding to the character, and the hatched portion in FIG. 15A indicates the specific target pixel Op corresponding to the photograph.

  The expansion processing is executed using, for example, a filter having a predetermined size, and in the example of FIG. 14A, a filter FI1 having a size of 3 vertical pixels × 3 horizontal pixels. Specifically, the CPU 210 applies the filter FI1 to the pre-expansion image data to generate binary image data that has undergone expansion processing. That is, the CPU 210 arranges the filter FI1 on the pre-expansion image OI so that the center position CC1 (see FIG. 14A) of the filter FI1 overlaps the target pixel. If even one specific target pixel exists within the range of the filter FI1, the CPU 210 sets a pixel corresponding to the target pixel in the binary image to be generated as the specific target pixel. When there is no specific target pixel in the range of the filter FI1, that is, when nine pixels in the range of the filter FI1 are non-specific target pixels, the CPU 210 A pixel corresponding to the target pixel is set as a non-specific target pixel. The CPU 210 sets all the pixels of the binary image BI as the target pixel and sets the corresponding pixel in the binary image to be generated as one of the specific target pixel and the non-specific target pixel, so that the expansion processing has been completed. Binary image data is generated.

  FIG. 14B shows a partial image PI1b that has been subjected to expansion processing, corresponding to the partial image PI1 of FIG. FIG. 15B shows an expanded partial image PI2b corresponding to the partial image PI2 in FIG. As illustrated in FIG. 14B, in the binary image that has been subjected to the expansion process (for example, the partial image PI1b), the line indicated by the specific target pixel Opt corresponding to the character is thicker than before the expansion process. . As shown in FIG. 15B, in the binary image after the expansion process (for example, the partial image PI2b), the non-specific target pixel VA (FIG. 15A) mixed in the region corresponding to the photograph before the expansion process. )) Is changed to the specific target pixel Op, and the entire region corresponding to the photograph is configured by the specific target pixel Op.

  In S420, the CPU 210 executes three contraction processes on the binary image data that has been subjected to the expansion process.

  The contraction process is executed on the binary image data that has been subjected to the expansion process using, for example, a filter of a predetermined size, and in the example of FIG. 14B, a filter FI2 having a size of 3 pixels vertically × 3 pixels horizontally. . Specifically, the CPU 210 applies the filter FI2 to the binary image data to generate binary image data that has been subjected to shrinkage processing. That is, the CPU 210 arranges the filter FI2 in the binary image (for example, partial images PI1b and PI2b) that has been subjected to the expansion process so that the center position CC2 (see FIG. 14B) of the filter FI2 overlaps the target pixel. . When at least one non-specific target pixel exists within the range of the filter FI2, the CPU 210 sets a pixel corresponding to the target pixel in the binary image to be generated as the non-specific target pixel. Then, if there is no non-specific target pixel in the range of the filter FI2, the CPU 210 sets a pixel corresponding to the target pixel in the binary image to be generated as the specific target pixel. The CPU 210 sets all of the pixels of the dilated binary image as the target pixel, and sets the corresponding pixel in the binary image to be generated as one of the specific target pixel and the non-specific target pixel. Processed binary image data is generated.

  FIG. 14C shows a contracted partial image PI1c corresponding to the partial image PI1b of FIG. 14B. FIG. 15C shows a contracted partial image PI2c corresponding to the partial image PI2b of FIG. As shown in FIG. 14C, in the binary image that has been subjected to the expansion process (for example, the partial image PI1c), the line indicated by the specific target pixel Opt corresponding to the character is thinner than before the contraction process. . Since the non-specific target pixel VA is not mixed in the region corresponding to the photograph before the contraction process (FIG. 15B), as illustrated in FIG. 15C, a binary image that has been subjected to the contraction process (for example, In the partial image PI2c), the entire area corresponding to the photograph is maintained as the specific target pixel Op.

  The CPU 210 performs similar contraction processing on the binary image data that has been subjected to the first contraction processing, generates binary image data that has been subjected to the second contraction processing, and the binary image data that has been subjected to the second contraction processing. A similar contraction process is performed on the value image data to generate binary image data subjected to the third contraction process. The binary image data that has undergone the third contraction process is also referred to as preprocessed image data, and the image indicated by the preprocessed image data is also referred to as preprocessed image OIb.

  FIG. 13B shows an example of the preprocessed image OIb. In the preprocessed image OIb, specific target pixels Op1b to Op6b corresponding to the objects Ob1 to Ob6 in the scan image SI are specified.

  In the preprocessed image OIb, the lines indicated by the specific target pixels Op3b to Op6b corresponding to the characters are thinner than the pre-expansion image OI. This is because the number of times of contraction processing (in this embodiment, three times) is larger than the number of times of expansion processing (in this example, one time), the degree of contraction of the contraction processing is the degree of expansion of the expansion processing. It is because it is larger than. In addition, when the lines indicated by the specific target pixels Op3 to Op6 corresponding to the characters in the pre-expansion image OI are relatively thin, all or part of the specific target pixels Op3 to Op6 corresponding to the characters by the contraction processing. However, it may disappear. In the non-character area determination process, it is sufficient that the non-character area can be specified. Therefore, the specification target pixels Op3 to Op6 corresponding to the characters may disappear.

  In the preprocessed image OIb, the entire area corresponding to the photograph is configured by the specific target pixels Op1b and Op2b. This is because once the entire area corresponding to the photograph becomes the target pixel by the expansion process, even if the degree of contraction of the contraction process is greater than the degree of expansion of the expansion process, This is because the entire corresponding region is maintained as the specific target pixel. Since the degree of contraction of the contraction process is larger than the degree of expansion of the expansion process, in the preprocessed image OIb, the entire region where the plurality of specific target pixels Op1b and Op2b corresponding to the photograph are located is the expansion process. Although it is slightly smaller than the entire area where the plurality of specific target pixels Op1 and Op2 corresponding to the photograph in the previous image OI are located, the difference is not a problem.

  In S425, the CPU 210 determines a plurality of target areas by performing a labeling process on the preprocessed image data. Specifically, the CPU 210 assigns one identifier, with one area composed of one or more consecutive specific target pixels as one target area. Then, the CPU 210 assigns different identifiers to a plurality of target areas separated from each other. A plurality of target areas are specified by such a labeling process. In the example of FIG. 13B, four target areas TA3 to TA6 corresponding to the four characters Ob3b to Ob6 in the scan image SI and 2 corresponding to the two photographs Ob1 and Ob2 in the scan image SI. The target areas TA1 and TA2 are determined. Since the plurality of pixels of the scan image SI and the plurality of pixels of the pre-expansion image OI correspond one-to-one, determining the target area on the pre-expansion image OI means that the scan image SI This is synonymous with determining the target area above.

  In S430, the CPU 210 prepares canvas data for generating non-character area data in a memory (specifically, a buffer area of the volatile storage device 220). The canvas (initial image) indicated by the canvas data is a binary image having the same size as the scanned image SI, that is, a binary image having the same number of pixels. The value of each pixel of the canvas data is a predetermined initial value (for example, 0) indicating that the pixel is in a region different from the non-character region.

  In S435, the CPU 210 selects one target region of interest from among a plurality of target regions that have been determined for the preprocessed image OIb.

  In S440, the CPU 210 calculates the number of pixels PN of the specific target pixels constituting the target region of interest. In S445, the CPU 210 determines whether or not the pixel number PN is equal to or greater than the threshold value TH1. The threshold value TH1 is a value corresponding to the size of the document and the resolution of the scan image SI, and is empirically determined in advance using scan data based on the document including characters and photographs. When the pixel number PN is equal to or greater than the threshold value TH1 (S445: YES), the CPU 210 advances the process to S450.

  In S450, the CPU 210 calculates the occupation rate PR of the target area for the circumscribed rectangle OS circumscribing the target area. For example, when the target area TA5 in FIG. 13B is the target area of interest, the pixel number PN of the target area TA5 with respect to the pixel number SN of the circumscribed rectangle OS5 indicated by the broken line is calculated as the occupation rate PR. (PR = SN / PN). In S455, the CPU 210 determines whether or not the occupation rate PR is equal to or greater than the threshold value TH2. The threshold value TH2 is empirically determined in advance using scan data based on a document including characters and photographs. The threshold value TH2 is, for example, 80% to 90%. If the occupation rate PR is equal to or greater than the threshold value TH2 (S455: YES), the CPU 210 advances the process to S460.

  In S460, the CPU 210 sets the target area as a non-character area. Specifically, the CPU 210 sets the value of all the pixels in the region corresponding to the target region of interest in canvas data to a value “1” indicating a non-character.

  When the pixel number PN is less than the threshold value TH1 (S445: NO), and when the occupation rate PR is less than the threshold value TH2 (S455: NO), the CPU 210 skips S460 and performs the process in S470. To proceed. That is, in this case, the attention target area is set to an area different from the non-character area. Accordingly, in the canvas data, the values of all the pixels in the region corresponding to the target region of interest are maintained at the initial value “0”.

  In S470, the CPU 210 determines whether or not all target areas in the preprocessed image OIb have been processed as the target area of interest. If there is an unprocessed target area (S470: NO), the CPU 210 returns to S435 and selects the unprocessed target area as the target area of interest. When all the target areas have been processed (S470: YES), the CPU 210 ends the non-character area determination process. At this point, the non-character area data shown in FIG. 4B is generated.

  According to the present embodiment described above, in S22 to S26 of the character specifying process (S20 in FIG. 2), the CPU 210 uses the scan data so that each of the plurality of pixels in the scanned image SI is a character pixel. A plurality of character candidate pixels are determined by determining for each pixel whether or not it is a candidate for (S22 to S26 in FIG. 2). In the non-character area determination process (S28 in FIG. 2, FIG. 12), the CPU 210 performs pre-processing (S405 to S420 in FIG. 12) including scan processing (S415 in FIG. 12) on the scan data, Pre-processed image data is generated. The CPU 210 determines whether each of the plurality of target areas arranged on the image corresponding to the preprocessed image data (preprocessed image OIb) satisfies a specific determination condition, thereby determining the scan image SI. The non-character area is specified (S435 to S470 in FIG. 12). Among the plurality of pixels in the scan image SI, pixels that are determined to be character candidate pixels in S26 of FIG. 2 and that are in a region different from the non-character region specified in S28 of FIG. It is specified as a character pixel (S29 in FIG. 2). In the preprocessed image data, the specific target pixels constituting the halftone dots of the photograph are likely to appear as a group of pixels by the expansion process (FIG. 13B). As a result, in the scanned image SI, an area including pixels constituting a halftone dot such as a photograph can be accurately identified as a non-character area. Then, a pixel in an area different from the non-character area is specified as a character pixel. Therefore, for example, even if it is erroneously determined that the pixel constituting the halftone dot is a character candidate pixel in S26 of FIG. 2, the halftone dot is determined by the determination using the preprocessed image data in S28 of FIG. By determining that the area including the constituent pixels is a non-character area, it is possible to suppress erroneously specifying that the pixels constituting the halftone dots are character pixels. Therefore, even in an image including halftone dots, character pixels in the scan image SI can be specified with high accuracy.

  Further, in the non-character area determination process of the present embodiment, in the preprocess, the contraction process (S420) is executed after the expansion process (S415) to generate preprocessed image data. As a result, in the preprocessed image OIb, it is possible to suppress the specific target pixels corresponding to the characters from being a group of pixels. For example, even if a thin line constituting a character is partially connected to another adjacent line by the expansion process, it can be separated again by the contraction process. As a result, it is possible to prevent an area corresponding to the character in the scan image SI from being erroneously specified as a non-character area.

  Furthermore, in this embodiment, the specific determination condition for determining the non-character region includes that the ratio of the specific target pixel (occupancy ratio PR) in the circumscribed rectangle OS is equal to or higher than the reference ratio (threshold value TH2). (S455 in FIG. 12). A photograph is likely to have a rectangular shape, whereas characters are composed of thin lines. For this reason, as shown in FIG. 13B, the target area corresponding to the photograph (for example, the target area TA1) has a higher occupation rate PR than the target area corresponding to the characters (for example, the target area TA3). Conceivable. By using the occupation rate PR, the non-character region can be appropriately specified.

  Further, in the present embodiment, the specific determination condition for determining the non-character region includes that the number of specific target pixels (pixel number PN) in the target region is equal to or greater than the reference number (threshold value TH1) ( S445 in FIG. 12). A photograph assumed as a non-character in this embodiment is composed of a group of specific target pixels as shown in FIG. 13B and is likely to occupy a certain area on the scan image SI. Since the characters are composed of thin lines, the area of the characters is likely to be relatively small. For this reason, as shown in FIG. 13B, if the target area corresponding to the photograph (for example, the target area TA1) has a larger number of pixels PN than the target area corresponding to the characters (for example, the target area TA3). Conceivable. By using the number of pixels PN, the non-character region can be appropriately specified.

  Furthermore, in this embodiment, the degree of contraction in the contraction process in S420 is greater than the degree of expansion in the expansion process in S415. As a result, the thin line connected by the expansion process can be more reliably separated by the contraction process. Therefore, it can suppress more effectively that the specific object pixel corresponding to a character becomes a lump pixel group.

  Further, in this embodiment, the expansion process of S415 is performed once, and the expansion process of S420 is performed three times. That is, in this embodiment, the expansion process includes one unit expansion process, and the contraction process includes three unit contraction processes. As a result, the degree of contraction in the contraction process in S420 can be easily made larger than the degree of expansion in the expansion process in S415.

  Further, in the non-character area determination process according to the present embodiment, the CPU 210 uses, as pre-processed image data, an area configured by a plurality of continuous specific target pixels in the pre-processed image OIb as a target area. Determine (S425 in FIG. 12). As a result, it is possible to determine an appropriate target area for determining whether or not it is a non-character area.

  Furthermore, in the non-character area determination process of the present embodiment, in the canvas data prepared in S430 of FIG. 12, the initial value of each pixel is a value indicating that the pixel is in a different area from the non-character area (for example, , 0). And only the area | region judged to be a non-character area among object areas is determined as a non-character area (S460). Therefore, in the preprocessed image OIb, an area (for example, a white background area) configured by a pixel (non-specific target pixel) different from the specific target pixel is specified as a different area from the non-character area (see FIG. 4 (B)). As a result, for example, there is no problem even if the specific target pixel corresponding to the character disappears by the contraction process of S420. Therefore, character pixels can be specified without omission.

  In this embodiment, the CPU 210 generates minimum component data using the scan data (S405 in FIG. 12), generates pre-expansion image data using the minimum component data (S410 in FIG. 12), and performs expansion processing. Processing (S415, S420) including expansion processing is executed on the pre-image data to generate pre-processed image data. As a result, by using the minimum component data, the pixels constituting the halftone dots are accurately identified as the identification target pixels. Therefore, the non-character area is specified with high accuracy.

  Further, according to the above embodiment, when the enlargement ratio Lr of the enlargement process in S15 is less than the threshold Lth (S27: YES), the CPU 210 performs the non-character area determination process in S28 and the composition process in S29. Execute. That is, in this case, among a plurality of pixels in the scan image SI, pixels that are determined as character candidate pixels in S22 to S26 and that are not determined as non-character pixels in S28 are characters. Identified as a pixel. When the enlargement ratio Lr is equal to or greater than the threshold Lth (S27: NO), the processes of S28 and S29 are not executed. That is, in this case, a pixel determined to be a character candidate pixel among a plurality of pixels in the scan image SI is specified as a character pixel. As a result, an appropriate character pixel is specified in accordance with the enlargement ratio Lr when the scan data is generated. More specifically, when the scan image SI is enlarged and printed (printed in this embodiment) with an enlargement ratio equal to or greater than the threshold value Lth, the halftone dots are relatively large in the output image, so the halftone dots are conspicuous. However, it often does not look unnatural. For this reason, when the enlargement ratio Lr is equal to or greater than the threshold value Lth, it is often more natural to see the large halftone dot clearly in the output image than to blur it. In addition, since the enlarged copy should output an image showing the original as it is, it is desirable that the halftone dot is also enlarged and output in the enlarged copy. Furthermore, when the halftone dots are enlarged beyond a certain size, moire due to interference with the periodic component of dots during printing does not occur. For this reason, when the enlargement ratio Lr is greater than or equal to the threshold value Lth, the edge of the halftone dot is preferably the target of edge enhancement processing, as is the case of the character edge, than the target of smoothing processing. it is conceivable that. For this reason, when the enlargement ratio Lr is equal to or greater than the threshold value Lth, a character candidate pixel that can include a pixel that forms a halftone dot edge together with the character pixel among the plurality of pixels in the scan image SI is a character pixel. It is considered preferable to be specified as

  As can be understood from the above description, the minimum component data in the present embodiment is an example of the first image data, the specific target pixel is an example of a pixel that satisfies the specific condition, and the pre-expansion image data is intermediate. It is an example of image data.

B. Variations:

(1) In the non-character area determination process of FIG. 12 in the above-described embodiment, the preprocessing includes the contraction process of S420, but the contraction process of S420 may be omitted. Even in this case, in the non-character area determination process, the areas corresponding to the photographs Ob1 and Ob2 can be specified as non-character areas. In this case, for example, since the occupancy rate PR of the character area is larger than that in the above embodiment in the determination of S455, the threshold value such as TH2 is set larger than that in the above embodiment.

(2) In the non-character area determination process of FIG. 12 of the above embodiment, the degree of contraction in the contraction process in S420 is greater than the degree of expansion in the expansion process in S412. Instead, the degree of contraction in the contraction process in S420 may be the same as the degree of expansion in the expansion process in S412. Further, the degree of contraction in the contraction process in S420 may be smaller than the degree of expansion in the expansion process in S412.

(3) In the non-character area determination process of FIG. 12 in the above embodiment, the number of expansion processes in S415 is one, but is not limited to one. In general, the expansion processing in S415 may include unit expansion processing N times (N is an integer of 1 or more). Similarly, the number of contraction processes in S420 is three, but is not limited to three. In general, the contraction process in S420 may include unit contraction processes M times (M is an integer of 1 or more). In this case, it is preferable to satisfy M> N in order to make the degree of contraction of the contraction process larger than the degree of expansion of the expansion process.

(4) In the non-character area determination process of FIG. 12 in the above embodiment, the number of contraction processes in S420 is made larger than the number of expansion processes in S415, so that the degree of contraction in the contraction process in S420 is the same as in S412. It is larger than the degree of expansion in the expansion process. Instead, the filter FI2 used in the contraction process of S420 is made larger than the filter FI1 used in the expansion process of S415, so that the degree of contraction of the contraction process is larger than the degree of expansion of the expansion process. May be.

(5) In the non-character area determination process of FIG. 12 in the above embodiment, a plurality of consecutive specific target pixels are determined as one target area TA1 to TA6 (S425, FIG. 13B). Instead, for example, a plurality of rectangular target areas may be determined on the preprocessed image OIb by dividing the preprocessed image OIb into a grid shape. In this case, the target region may include both the specific target pixel and the non-specific target pixel. In this case, if the number of specific target pixels included in the target area is greater than or equal to the threshold, the target area may be determined to be a non-character area, or the specific target occupying the total number of pixels in the target area When the ratio of the number of pixels is equal to or greater than the threshold value, the target area may be determined to be a non-character area.

(6) In the non-character area determination process of the above embodiment, the specific determination condition for determining that the target area is a non-character area is that the number of pixels PN is equal to or greater than the threshold value TH1 and the occupation ratio PR is That is, the threshold value TH2 or more. The specific judgment condition is not limited to this. For example, the specific determination condition may be that the number of pixels PN is equal to or greater than the threshold value TH1, or the occupation rate PR is equal to or greater than the threshold value TH2. Further, the specific determination condition may be that only the number of pixels PN is equal to or greater than the threshold value TH1, or that the occupation rate PR is equal to or greater than the threshold value TH2. Further, the specific determination condition may include other conditions, for example, that the aspect ratio of the target area is within a predetermined aspect ratio assumed as a photograph area.

(7) In the image processing of FIG. 2 in the above embodiment, the second binary image data generation processing in S24 and the synthesis processing in S26 may be omitted. That is, the plurality of edge pixels specified in the first binary image data generation process may be final edge specifying data.

(8) In the image processing of FIG. 2 in the above embodiment, S27 of FIG. 2 may be omitted. In other words, the CPU 210 may always execute S28 and S29 regardless of the enlargement ratio Lr when the scan data is generated.

(9) In the first binary image data generation process (FIG. 5) in FIG. 5, luminance image data is used as the single component image data (S120). Instead of this, average component value image data having an average value of three component values (R value, G value, B value) included in the RGB value of the corresponding pixel of the scan data as the value of each pixel is used. May be.

(10) In the second binary image data generation process (FIG. 9) of the above embodiment, the minimum component data is used as the single component image data (S300). Instead of this, maximum component data or inverted minimum component data may be used.

  The maximum component data includes a plurality of values corresponding to a plurality of pixels included in the scan data, and each of the plurality of values is a maximum component value Vmax of the corresponding pixel of the scan data. The maximum component value Vmax is the maximum value among a plurality of component values (R value, G value, B value) included in the RGB value of the corresponding pixel of the scan data.

  The inversion minimum component data is acquired as follows. First, for each of a plurality of pixel values (RGB values) included in the scan data, an inverted color value is generated by inverting a plurality of component values (R value, G value, B value). If the RGB values before inversion are (Rin, Gin, Bin), the inverted RGB values (Rout, Gout, Bout) are expressed by the following equations (1) to (3).

Rout = Rmax−Rin (1)
Gout = Gmax−Gin (2)
Bout = Bmax−Bin (3)

  Here, Rmax, Gmax, and Bmax are the maximum values that the R value, G value, and B value can take, respectively, and in this embodiment, Rmax = Gmax = Bmax = 255. Image data having these inverted RGB values as values of a plurality of pixels is generated as inverted image data. Then, the inverted minimum component data is generated using the inverted image data. Specifically, the inverted minimum component value VRmin is obtained from each of a plurality of inverted RGB values included in the inverted image data. The inversion minimum component value VRmin is the minimum value among a plurality of component values (R value, G value, B value) included in the inverted RGB value. The inversion minimum component data is image data in which these inversion minimum component values VRmin are values of a plurality of pixels.

  The inversion minimum component value VRmin is an inversion value of the maximum component value, and the relationship VRmin = (255−Vmax) is established. For this reason, both the maximum component data and the inverted minimum component data are both values based on the maximum value among the plurality of component values included in the value of each pixel of the scan data (the inverted value of the maximum value, or It can be said that the maximum value itself is image data having pixel values. It can be said that the minimum component data of the embodiment is image data in which a value based on the minimum value among a plurality of component values included in each pixel value of the scan data is a pixel value. .

  As shown in FIG. 10, the maximum component values Vmax of C, M, Y, K, and W are 255, 255, 255, 0, and 255, which are the same values except for black (K). Therefore, in the maximum component data and the inverted minimum component data, among the five types of elements constituting the halftone dot region, that is, each dot of C, M, Y, K, and the ground color (white) of the paper Differences in values between pixels indicating four types of elements (C, M, and Y dots and paper ground color (white)) are suppressed. As a result, when the maximum component data and the inverted minimum component data are used, it is possible to suppress the specification of the edge pixels due to the halftone dots, as in the case of using the minimum component data.

(11) In each of the above embodiments, the character sharpening process is performed on the character pixels (S40 in FIG. 2), and the halftone dot smoothing process is performed on the non-character pixels (S30 in FIG. 2). ). Instead, an antialiasing process for improving the appearance of the character may be performed on the character pixel. For non-character pixels, for example, a process of skipping colors (a process of converting to white) may be performed in order to reduce the amount of color material used during printing. In general, it is preferable that different image processing is performed on character pixels and non-character pixels. Alternatively, the specific image processing may be performed on one of the character pixel and the non-character pixel, and the specific image processing may not be performed on the other.

(12) In the above embodiment, a Sobel filter is used in the edge extraction process of S130 of FIG. 5 or S320 of FIG. Instead, other edge extraction filters such as a Roberts filter and a Laplacian filter may be used in these edge extraction processes.

(13) In the above embodiment, the target image data is scan data, but is not limited thereto. The target image data may be generated by reading a printed matter with a digital camera including a two-dimensional image sensor.

(14) In the above embodiment, the edge specifying data is generated by taking the logical sum of the first binary image data and the second binary image data (S26 in FIG. 2). Instead of this, the edge specifying data may be generated by taking the logical sum of the first binary image data, the second binary image data, and the third binary image data. As the third binary image data, for example, binary image data generated using the maximum component data described above may be used. As a result, it is possible to further suppress specific omission of edges such as characters.

(15) The first binary image data generation process (FIG. 5) and the second binary image data generation process (FIG. 9) of the above embodiment can be changed as appropriate. For example, all or part of the processing of S110, S110, and S140 in FIG. 5 can be omitted. Further, all or part of S310 and S330 in FIG. 9 may be omitted.

(16) The image processing apparatus that realizes the image processing of FIG. 2 is not limited to the multifunction machine 200, and may be various apparatuses. For example, the scanner or digital camera may execute the image processing of FIG. 2 in order to generate print data to be supplied to a printer using image data generated by itself. Further, for example, a connected terminal device (for example, the terminal device 100) or server (not shown) that can communicate with the scanner or printer executes the image processing of FIG. 2 using the scan data acquired from the scanner. The print data may be generated and the print data may be supplied to the printer. In addition, a plurality of computers (for example, cloud servers) that can communicate with each other via a network may share the functions required for image processing, and execute image processing as a whole. In this case, the entirety of the plurality of computers is an example of the image processing apparatus.

(17) In each of the above embodiments, a part of the configuration realized by hardware may be replaced with software. Conversely, part or all of the configuration realized by software is replaced with hardware. You may do it. For example, the process of calculating the character probability Txr using the machine learning model in S410 of FIG. 12 may be executed by dedicated hardware such as an ASIC.

  As mentioned above, although this invention was demonstrated based on the Example and the modification, Embodiment mentioned above is for making an understanding of this invention easy, and does not limit this invention. The present invention can be changed and improved without departing from the spirit and scope of the claims, and equivalents thereof are included in the present invention.

  DESCRIPTION OF SYMBOLS 100 ... Terminal device, 200 ... MFP, 210 ... CPU, 220 ... Volatile storage device, 230 ... Nonvolatile storage device, 240 ... Display part, 250 ... Operation part, 270 ... Communication IF, 280 ... Print execution part, 290 ... Read execution unit, PG ... Computer program, SI ... Scanned image, TI ... Character specific image, GI ... Smoothed image, FI ... Processed image, EI ... Edge specific image, BI ... Block determination image, YI ... Luminance image, SI: Scanned image, DT: Halftone dot, SB: Sub-block

Claims (11)

An image processing apparatus,
An image acquisition unit for acquiring target image data indicating the target image;
A candidate determination unit that determines a plurality of candidate pixels by determining, for each pixel, whether or not each of the plurality of pixels in the target image is a character pixel candidate using the target image data. When,
A preprocessing unit that performs preprocessing including expansion processing on the target image data to generate preprocessed image data;
A non-character region in the target image is identified by determining whether each of the plurality of target regions arranged on the image corresponding to the preprocessed image data satisfies a specific determination condition. A character area identification part;
A character identifying unit that identifies, as the character pixel, a pixel that is determined to be the candidate pixel among a plurality of pixels in the target image and that is in a region different from the non-character region;
An image processing apparatus comprising: The image processing apparatus according to claim 1,
The image processing apparatus, wherein the preprocessing unit executes a contraction process after the expansion process to generate the preprocessed image data. The image processing apparatus according to claim 2,
The image processing apparatus, wherein a degree of contraction of the contraction process included in the preprocess is larger than a degree of expansion of the expansion process included in the preprocess. The image processing apparatus according to claim 3,
The expansion processing included in the preprocessing includes unit expansion processing N times (N is an integer of 1 or more),
The image processing apparatus, wherein the contraction process included in the preprocessing includes unit contraction process M times (M is an integer satisfying M> N). The image processing apparatus according to claim 1, further comprising:
An image processing apparatus that determines, using the preprocessed image data, an area composed of a plurality of continuous pixels that satisfy a specific condition in an image as the target area. The image processing apparatus according to claim 5,
An image processing apparatus, wherein an area constituted by a plurality of pixels different from a plurality of pixels satisfying a specific condition in the image is specified as an area different from the non-character area. The image processing apparatus according to claim 1,
The image processing apparatus, wherein the specific determination condition includes a ratio of pixels satisfying a specific condition in a determination area based on the target area being a reference ratio or more. The image processing apparatus according to claim 1,
The image processing apparatus, wherein the specific determination condition includes that the number of pixels satisfying the specific condition in the target region is equal to or greater than a reference number. The image processing apparatus according to claim 1,
Each of the plurality of pixel values included in the target image data is a color value including a plurality of component values,
The pre-processing unit is
First image data including a plurality of first values corresponding to the values of the plurality of pixels included in the target image data, and a plurality of corresponding to the values of the plurality of pixels included in the target image data Generating at least one of second image data including the second values,
Generating intermediate image data using at least one of the first image data and the second image data;
The intermediate image data is subjected to processing including the expansion processing to generate the preprocessed image data,
Each of the plurality of first values included in the first image data is a value based on a minimum value among the plurality of component values included in the value of the corresponding pixel,
Each of the plurality of second values included in the second image data is a value based on a maximum value among the plurality of component values included in the corresponding pixel value. The image processing apparatus according to claim 1, further comprising:
A first image process is executed on the specified character pixel value of the target image data, and a second value different from the first image process is applied to a pixel value different from the character pixel. An image processing apparatus comprising: an image processing unit that executes the image processing to generate the target image data that has undergone image processing. A computer program,
An image acquisition function for acquiring target image data indicating the target image;
Candidate determination function for determining a plurality of candidate pixels by determining for each pixel whether or not each of the plurality of pixels in the target image is a character pixel candidate using the target image data When,
A preprocessing function for generating preprocessed image data by executing preprocessing including expansion processing on the target image data;
A non-character region in the target image is identified by determining whether each of the plurality of target regions arranged on the image corresponding to the preprocessed image data satisfies a specific determination condition. Character area identification function,
A character specifying function for specifying, as the character pixel, a pixel that is determined to be the candidate pixel among a plurality of pixels in the target image and is in a region different from the non-character region;
A computer program that causes a computer to realize

Images