JP5870826B2 - Image processing apparatus, image processing method, and computer program - Google Patents

Description

  The present invention relates to a technique for binarizing a so-called multilevel image.

  A technique for converting an image represented by a plurality of gradations including a line drawing such as a character or a figure into a binary image is known. The technology for converting to a binary image by replacing pixels in the image whose brightness is greater than or equal to a predetermined brightness with white pixels and replacing pixels whose brightness is less than the predetermined brightness with black pixels is well known. It has been.

  However, when this image is read by a scanner, this technique converts the boundary between the line drawing portion and the background (background) portion to a binary image even though it is a straight line or a smooth curve. If converted, it may be rattling. This is because there may be a difference between the brightness of each pixel detected by the scanner and the actual brightness. That is, it is caused by so-called reading error.

  The following techniques have been proposed as a method for suppressing the occurrence of edge rattling. The input multi-value image data is compared with threshold data, a binarization processing unit that performs binarization processing, a binarization data and template data are compared, a template matching processing unit that determines whether smoothing is possible, and a smoothing target A smoothing processing unit that interpolates and corrects a pixel determined to be present based on smoothing data and converts it to multi-value data, a register in which threshold data, template data, and smoothing data are set, and smoothing processing. A data selector that receives multi-value pixel data and input multi-value image data, selects and outputs multi-value pixel data after smoothing processing when it is a smoothing target, and selects and outputs original data when it is not a smoothing target Are provided in an image forming apparatus (Patent Document 1).

Alternatively, when pixel data is input for a dot matrix image, if image data is input, the pixel of interest is scanned over the entire image, and the edge exists in the 45 ° direction or 135 ° direction with respect to the pixel of interest. In the case of edges having the same angle, smoothing is performed in the 45 ° direction or 135 ° direction, and then pixel interpolation is performed regardless of the presence or absence of the edge (Patent Document 2).
JP 2000-92327 A JP 2000-123162 A

  However, according to the technique described in Patent Document 1, in order to deal with various shakiness of edges, templates of various patterns must be prepared.

  Further, according to the technique described in Patent Document 2, it is not possible to suppress edge shakiness caused by a reading error.

  In view of such problems, an object of the present invention is to binarize edges of a multi-valued image by suppressing edge shakiness caused by a reading error of a scanner more easily than in the past.

  An image processing apparatus according to an aspect of the present invention is an image processing apparatus that binarizes a multi-valued image that is reproduced with three or more gradations and includes a line drawing portion representing a line drawing and a ground portion representing a background. Threshold determination means for determining one threshold and a second threshold smaller than the first threshold; and gradation of a boundary pixel at a boundary between the line drawing portion and the background portion of the pixels of the multi-valued image Is between the first threshold value and the second threshold value, the binary value of the boundary pixel is the same value as the binary value already determined for the specific position pixel at a specific position in the vicinity of the boundary pixel. And binary determination means.

  Preferably, the threshold value determination unit determines the first threshold value to be a value smaller than the maximum value among the gradations of each of the plurality of surrounding pixels around the boundary pixel, and sets the second threshold value. , A value larger than the minimum value of the gradations is determined.

Alternatively, the first threshold value may be any one of an average value, a median value, and a value obtained by dividing the sum of the maximum value and the minimum value by 2, for each of the plurality of surrounding pixels, The sum of one specific value and the second threshold value is determined by subtracting the first specific value from any one of the values. The multi-valued image is the image when a read image read by the device, the threshold value determining means, the use of a larger value as the reading error of the image reading apparatus is large and the first specific value To determine the first threshold and the second threshold.

  Alternatively, the gradation of the multi-valued image is lightness, and the threshold value determination unit is configured to detect each of the boundary pixel and one or more neighboring tangential pixels arranged on the tangent line to the boundary at the boundary pixel. The first threshold value is determined by using, as the first specific value, the maximum absolute value of the absolute values of the differences between the lightness and the lightness of pixels adjacent to the base portion side. The minimum absolute value of the absolute values of the difference between the brightness of each of the boundary pixel and the one or more tangential pixels and the brightness of the pixel adjacent to the line drawing portion side is the first absolute value. The second threshold is determined by using it as a second specific value.

Alternatively, the threshold determination means determines the first threshold to a value obtained by subtracting a second specific value from the maximum value, and sets the second threshold as the minimum value and the second specific value . Determine the sum with the value . The multi-valued image is the image when an image read by the reading device, the threshold value determining means, the use of a smaller value read error is large in the image reading apparatus and said second specific value To determine the first threshold and the second threshold.

  Alternatively, the specific position pixel is a pixel whose gradation is between the first threshold value and the second threshold value. Alternatively, the pixel having the smallest absolute value of the difference in gradation from the boundary pixel among the pixels adjacent to the boundary pixel. Or it is a pixel in which the predetermined binary value is the same and a predetermined number or more are arranged in the same direction around the boundary pixel.

  Or a smoothing processing unit that performs a smoothing process on the multi-valued image, and the threshold value determining unit includes the first threshold value based on the multi-valued image after the smoothing process is performed. The second threshold value is determined, and the binary determination unit determines a binary value of the boundary pixel based on the multi-valued image after the smoothing process is performed. For example, the smoothing processing unit selects, for each pixel of the multi-valued image, a gradation closest to the gradation of the pixel from gradations obtained by performing smoothing in each of a plurality of directions. The smoothing process is performed. Alternatively, the smoothing process is performed by smoothing the edge direction for each pixel of the multi-valued image.

  According to the present invention, it is possible to binarize an edge of a multi-valued image by suppressing the rattling of the edge due to the reading error of the scanner more easily than in the past.

1 is a diagram illustrating an example of an entire image processing system including an image forming apparatus. 2 is a diagram illustrating an example of a hardware configuration of an image forming apparatus. FIG. It is a figure which shows the example of a multi-value image. It is a figure which shows the example of a structure of an image editing circuit. It is a figure which shows the example of a smooth process part. It is a figure for demonstrating the example of the relationship between an attention pixel and a surrounding pixel, and the example of calculation of average brightness. It is a figure for demonstrating the example of the method of determination of the minimum direction. It is a figure which shows the example of a prescription | regulation matrix. It is a flowchart explaining the example of the flow of an inside / outside determination process. It is a figure for demonstrating the modification of the method of the determination of an upper limit threshold value and a lower limit threshold value. It is a figure for demonstrating the modification of the method of the determination of an upper limit threshold value and a lower limit threshold value. It is a figure which shows the modification of a smooth process part. It is a figure for demonstrating the modification of the method of determining whether an attention pixel is a pixel inside or outside a character. 6 is a flowchart illustrating an example of the overall processing flow of the image forming apparatus. 10 is a flowchart illustrating a modification of the overall processing flow of the image forming apparatus.

  FIG. 1 is a diagram illustrating an example of the entire image processing system including the image forming apparatus 1. FIG. 2 is a diagram illustrating an example of a hardware configuration of the image forming apparatus 1. FIG. 3 is a diagram illustrating an example of the multi-valued image 5.

  An image forming apparatus 1 shown in FIG. 1 is an apparatus generally called a multi-function peripheral or MFP (Multi Function Peripherals), and is an apparatus in which functions such as copying, network printing (PC printing), fax, and scanner are integrated. .

  The image forming apparatus 1 can exchange image data with an apparatus such as a personal computer 4 via a communication line NW such as a LAN (Local Area Network), a public line, or the Internet.

  As shown in FIG. 2, the image forming apparatus 1 includes a CPU (Central Processing Unit) 10a, a RAM (Random Access Memory) 10b, a ROM (Read Only Memory) 10c, a mass storage device 10d, a scan unit 10e, a print unit 10f, A network interface card (NIC) 10g, an operation panel 10h, a modem 10i, an image editing circuit 10j, and a control circuit are included.

  The NIC 10g is a device for communicating with the personal computer 4 or the like by using a protocol such as TCP / IP (Transmission Control Protocol / Internet Protocol) via the communication line NW.

  The modem 10i is a device for exchanging image data with a fax terminal or the like using a protocol such as G3 via a fixed telephone network.

  The operation panel 10h is configured by a touch panel display, a key group, and the like.

  The touch panel display displays a screen for giving a message to the user, a screen showing a result of processing, a screen for the user to input an instruction to the image forming apparatus 1, and the like. The touch panel display detects a touched (pressed) position and notifies the CPU 10a of the position.

  The key group includes keys such as a numeric keypad, a start key, and a stop key.

  The user can give commands or input data to the image forming apparatus 1 by operating the operation panel.

  The scan unit 10e is an apparatus that generates image data by reading an image such as a photograph, a character, a picture, or a chart on a sheet.

  The image editing circuit 10j performs processing for converting an image read by the scan unit 10e or an image transmitted from the personal computer 4 or the like into a binary image.

  The printing device 10f prints an image read by the scan unit 10e or an image transmitted from the personal computer 4 or the like on a sheet. It is also possible to print an image that has been processed by the image editing circuit 10j.

  The ROM 10c and the mass storage device 10d store software such as an operating system, middleware, and applications for realizing the above-described functions. Modules constituting each software are loaded into the RAM 10b as necessary and executed by the CPU 10a. A hard disk drive, SSD (Solid State Drive), or the like is used as the mass storage device 10d.

  Hereinafter, a case where the multi-valued image 5 as shown in FIG. 3A read by the scan unit 10e is converted into a binary image will be described as an example. The multivalued image 5 is an image having a gradation of 3 or more. For example, when the multi-value image 5 is an 8-bit image, the multi-value image 5 has 256 gradations. Further, as shown in FIG. 3B, it is constituted by a character area 51 which is an area representing a character and a background area 52 which is an area representing a character background (background).

  FIG. 4 is a diagram illustrating an example of the configuration of the image editing circuit 10j. FIG. 5 is a diagram illustrating an example of the smoothing processing unit 101. FIG. 6 is a diagram for explaining an example of the relationship between the target pixel 6C and the surrounding pixel 6S and an example of calculating the average brightness. FIG. 7 is a diagram for explaining an example of how to determine the minimum direction. FIG. 8 is a diagram illustrating an example of the definition matrix MT. FIG. 9 is a flowchart illustrating an example of the flow of the inside / outside determination process.

  As shown in FIG. 4, the image editing circuit 10j includes a smoothing processing unit 101, a binarization threshold value determination unit 102, an inside / outside determination unit 103, a pixel binarization processing unit 104, a determination result data storage unit 105, and the like. The

  When the image data of the multi-valued image 5 is input from the scan unit 10e, the smoothing processing unit 101 rattles the outline of characters in the multi-valued image 5 (that is, the boundary between the character region 51 and the background region 52). For example, the following process is executed to suppress the sticking.

  As shown in FIG. 5, the smoothing processing unit 101 includes a directivity smoothing processing unit 121, a difference calculation unit 122, a selector 123, and the like. Then, the pixel value for each pixel of the multi-valued image 5 is corrected based on the pixel values of the surrounding pixels. Hereinafter, a case where brightness is used as the pixel value will be described. In addition, among the pixels constituting the multi-valued image 5, the processing target is described as “target pixel 6C”, and the pixels around the target pixel 6C are described as “surrounding pixel 6S”.

When the image data of the multi-valued image 5 is input, the directivity smoothing processing unit 121 selects one pixel from the multi-valued image 5 as shown in FIG. And Then, the average brightness of the brightness of the target pixel 6C shown in FIG. 6B and the brightness of each of the upper and lower surrounding pixels 6S adjacent thereto, that is, (D _{m, n} + D _{m, n-1} + D _{m, n + 1} ) / 3. Hereinafter, this average brightness is referred to as “average brightness Dv1”.

Similarly, the directivity smoothing processing unit 121 obtains the average brightness of the brightness of the target pixel 6C shown in FIG. 6C and the brightness of the left and right surrounding pixels 6S adjacent thereto, that is, (D _{m, n} + D _{m−1, n} + D _{m + 1, n} ) / 3. Hereinafter, this average brightness is referred to as “average brightness Dv2”. The average of the brightness of the pixel of interest 6C shown in FIG. 6D and the brightness of the surrounding pixels 6S adjacent to the upper right and lower left adjacent thereto, that is, (D _{m, n} + D _{m-1, n + 1} + D _{m + 1, n-1} ) / 3 is calculated. Hereinafter, this average brightness is referred to as “average brightness Dv3”. The brightness of the target pixel 6C shown in FIG. 6E and the brightness of the surrounding pixels 6S adjacent to the upper left and lower right adjacent thereto, that is, (D _{m, n} + D _{m-1, n-1} + D _{m + 1). , n + 1} ) / 3. Hereinafter, this average brightness is referred to as “average brightness Dv4”.

  The directivity smoothing processing unit 121 then sends the average brightness data 3A indicating the average brightness Dv1 to Dv4 to the difference calculation unit 122 and the selector 123. In the examples of FIGS. 6B to 6E, the average brightness value of three pixels is calculated. However, the average brightness value of more than three pixels (for example, seven pixels) is calculated. May be.

  The difference calculation unit 122 calculates the total value of the absolute values of the differences between the brightness calculated by the directivity smoothing processing unit 121, the brightness of the target pixel 6C, and the brightness of the surrounding pixels 6S in a predetermined direction. Specifically, the following four values are calculated.

  For the up and down direction, the difference calculation unit 122 calculates the absolute value of the difference between the average brightness Dv1 and the brightness of each of the target pixel 6C and a predetermined number of neighboring pixels 6S in the up and down direction, and the total value is calculated. calculate. The predetermined number is preferably the same as the size (number of pixels) of one side of a prescribed matrix MT (see FIG. 8), which will be described later, used by the binarization threshold value determination unit 102, but may be another value. .

For example, when the predetermined number is “7”, the difference calculation unit 122 calculates the average brightness Dv1 and the target pixel 6C shown in FIG. 7A, the immediately above three surrounding pixels 6S, and the immediately below three surrounding pixels. The absolute value of the difference from the brightness of each 6S is calculated. As a result, seven absolute values are obtained. Then, the total value of the seven absolute values is calculated. That is, | _{Dm, n-3−} Dv1 | + | _{Dm, n−2−} Dv1 | + | _{Dm, n−1−} Dv1 | + | _{Dm, n−} Dv1 | + | _{Dm, n + 1−} Dv1 | + | _{Dm, n + 2−} Dv1 | + | _{Dm, n + 3−} Dv1 | Hereinafter, this total value is referred to as “difference value Ds1”.

  Similarly in the left-right direction, the direction connecting the upper right and the lower left, and the direction connecting the upper left and the lower right, the average pixels Dv1 to Dv3 and the target pixels shown in FIGS. 7B, 7C, and 7D, respectively. The difference value Ds2, the difference value Ds3, and the difference value Ds4 are calculated by calculating and summing the absolute value of the difference between the brightness of each of the 6C and the six neighboring pixels 6S immediately adjacent to each direction. To do.

That is, as the difference value Ds2, | _{Dm-3, n-} Dv2 | + | _{Dm-2, n-} Dv2 | + | _{Dm-1, n-} Dv2 | + | _{Dm, n-} Dv2 | + | _{Dm + 1, n-} Dv2 | + | _{Dm + 2, n-} Dv2 | + | _{Dm + 3, n-} Dv2 |

As the difference value Ds3, | D _{m−3, n + 3} −Dv3 | + | D _{m−2, n + 2} −Dv3 | + | D _{m−1, n + 1} −Dv3 | + | D _{m, n} − Dv3 | + | _{Dm + 1, n-1−} Dv3 | + | _{Dm + 2, n−2−} Dv3 | + | _{Dm + 3, n−3−} Dv3 |

As the difference value Ds4, | D _{m−3, n−3−} Dv3 | + | D _{m−2, n−2−} Dv3 | + | D _{m−1, n−1−} Dv3 | + | D _{m, n} − Dv3 | + | _{Dm + 1, n + 1-} Dv3 | + | _{Dm + 2, n + 2-} Dv3 | + | _{Dm + 3, n + 3-} Dv3 |

  Then, the difference calculation unit 122 sends the minimum direction data 3B indicating the direction related to the smallest difference value among the difference values Ds1 to Ds4 to the selector 123.

  The selector 123 selects the average brightness related to the direction indicated by the minimum direction data 3B input from the difference calculation unit 122 among the average brightness Dv1 to Dv4 indicated by the average brightness data 3A input from the directivity smoothing processing unit 121. select. For example, when the vertical direction is indicated by the minimum direction data 3B, the average brightness Dv1 is selected.

  The selected average brightness is a new value corrected by smoothing (smoothing) of the target pixel 6C.

  Then, the selector 123 sends the correction value data 3C indicating the selected average lightness as the smoothing lightness Dh to the binarization threshold value determination unit 102.

  The directivity smoothing processing unit 121 or the selector 123 sequentially selects each pixel of the multi-valued image 5 as the target pixel 6C, and executes the above-described processing. Thereby, the whole multi-valued image 5 is smoothed.

  Returning to FIG. 4, the binarization threshold value determination unit 102 or the pixel binarization processing unit 104 sets the target pixel 6 </ b> C so as to sequentially scan the pixels in the character outline portion of the smoothed multilevel image 5 one by one. The selected pixel 6C is replaced with one of black and white pixels. That is, binarization is performed. The “character outline portion” is a boundary between the character region 51 and the background region 52 and a portion in the vicinity thereof. Hereinafter, the processing of each unit will be described by taking as an example a case where the direction from left to right is the main scanning direction and the direction from top to bottom is the sub-scanning direction.

  The binarization threshold value determination unit 102 determines a threshold value for determining the target pixel 6C as either black or white as follows.

  The binarization threshold value determination unit 102 superimposes the defined matrix MT on the multi-valued image 5 so that the center cell thereof matches the target pixel 6C. One cell overlaps one pixel. Hereinafter, a case where a prescribed matrix MT including 7 × 7 cells as shown in FIG. 8 is used will be described as an example.

  The binarization threshold value determination unit 102 calculates the average value of the brightness of the pixels that overlap each cell of the defined matrix MT as the average brightness Dg. That is, the brightness of the pixel of interest 6C and the brightness of each of the 48 surrounding pixels 6S immediately surrounding it, that is, the average value of the brightness of 49 pixels are calculated as the average brightness Dg.

Based on the following equations (1) to (3), the central threshold value REFc, the upper limit threshold value REFu, and the lower limit threshold value REFb are calculated.
REFc = Dg (1)
REFu = Dg + DR / β (2)
REFb = Dg−DR / β (3)

However, in Equations (2) and (3), DR is the amount of change from the lowest value Dmin to the highest value Dmax of the brightness of the pixels with which the prescribed matrix MT overlaps. That is, Dmax−Dmin.

  Β is a value of 1 or more and is determined by the accuracy of reading by the scan unit 10e. The higher the accuracy, that is, the smaller the so-called reading error, the larger value is used as β, and the larger the reading error, the smaller the value is used. A table or function representing the relationship between accuracy (reading error) and β is prepared in advance, and β may be determined based on this table or function. The reading error may be obtained in advance by causing the scan unit 10e to read a sheet on which a predetermined image is written, and comparing the result of reading with the image on the sheet.

  The inside / outside determination unit 103 determines whether the pixel of interest 6C should be a pixel on the character side (that is, inside the character) or a pixel on the ground side (that is, outside the character) by the method shown in FIG.

  When the determination result data 3D is not stored in the determination result data storage unit 105 (No in # 801 in FIG. 9), the inside / outside determination unit 103 binarizes the smoothing lightness Dh calculated by the smoothing processing unit 101. If it is larger than the central threshold value REFc determined by the threshold value determination unit 102 (Yes in # 802), it is determined that the target pixel 6C should be a pixel outside the character (# 803). On the other hand, if the smoothing brightness Dh is smaller than the central threshold REFc (No in # 802), it is determined that the pixel should be inside the character (# 804).

  When the smoothing lightness Dh and the central threshold value REFc are equal, it may be determined that the pixel should be outside the character, or may be determined as the pixel inside the character. Similarly, in the comparison of the calculated value and the threshold value, which will be described later, how to determine (determine) in the case of the same value may be arbitrarily determined in advance.

  Then, data indicating the result of determination by the inner / outer determination unit 103 (whether it should be a pixel outside the character or a pixel inside the character) is stored in the determination result data storage unit 105 as determination result data 3D. (# 806).

  However, if the smoothing lightness Dh is not between the lower limit threshold value REFb and the upper limit threshold value REFu (No in # 805), it is not stored.

  On the other hand, when the determination result data storage unit 105 stores the determination result data 3D (Yes in # 801), the inside / outside determination unit 103 has the smoothing lightness Dh between the lower limit threshold value REFb and the upper limit threshold value REFu. If it is (Yes in # 807), it is determined that it should be as shown in the determination result data 3D (# 808). That is, if the determination result data 3D indicates that the pixel should be outside the character, it is determined that the pixel should be outside the character and should be the pixel inside the character. If so, it is determined that the pixel should be inside the character. The determination result data 3D is continuously stored in the determination result data storage unit 105 as it is.

  If not included (No in # 807), if the smoothing lightness Dh is larger than the central threshold REFc (Yes in # 809), it is determined that the pixel should be outside the character (# 810). If it is smaller (No in # 809), it is determined that the pixel should be inside the character (# 811). Then, the determination result data 3D stored in the inside / outside determination unit 103 is deleted (# 812).

  The binarization threshold value determination unit 102 and the inside / outside determination unit 103 perform the above-described processing using each pixel in the character outline portion of the multi-valued image 5 as the target pixel 6C. Thereby, it is determined whether each pixel of the character outline portion of the multi-valued image 5 should be a white pixel or a black pixel.

  The pixel binarization processing unit 104 binarizes the target pixel 6C as follows, for example. When the inside / outside determination unit 103 determines that the character should be outside the character, the pixel at the same position as the target pixel 6C in the binarized multi-valued image 5 is a value representing white (for example, “0”). On the other hand, when it is determined that the character should be inside the character, a value representing black (for example, “1”) is determined. And the data of the binary image 7 are produced | generated by producing | generating the data showing the position of each pixel, and the determined value.

  Note that the above-described processing of the binarization threshold value determination unit 102 and the inside / outside determination unit 103 may be performed also on portions other than the character outline portion. Alternatively, without performing the above-described processing, each pixel in the portion other than the character outline portion in the character region 51 is binarized to a value representing black, and the portion other than the character outline portion in the background region 52 is You may binarize to the value showing white.

  According to the present embodiment, for a pixel having a certain range of lightness, the binary value is determined to be the same value as the binary value already determined for other neighboring pixels. Therefore, it is possible to more easily prevent the character boundary (edge) rattling caused by the reading error of the scan unit 10e.

  FIG. 10 is a diagram for explaining a modification of how to determine the upper limit threshold REFu and the lower limit threshold REFb. FIG. 11 is a diagram for explaining a modification of how to determine the upper limit threshold REFu and the lower limit threshold REFb. FIG. 12 is a diagram illustrating a modification of the smoothing processing unit 101. FIG. 13 is a diagram for explaining a modification of the method for determining whether the target pixel 6C is a pixel inside or outside the character. FIG. 14 is a flowchart illustrating an example of the overall processing flow of the image forming apparatus 1.

  In the above-described embodiment, the center threshold value REFc, the upper limit threshold value REFu, and the lower limit threshold value REFb are calculated using the average value of the brightness of the target pixel 6C and the surrounding pixels 6S surrounding it, that is, the average brightness Dg. ), The lightness median value Dc, which is the median value of these pixels, may be used. In this case, the median brightness value Dc may be applied to the equations (1) to (3) instead of the average brightness Dg. The lightness median value Dc is a value obtained by dividing the sum of the minimum value Dmin and the maximum value Dmax by 2. Alternatively, the lightness median value Dc is the lightness in the middle of the lightness of the target pixel 6C and the surrounding pixels 6S surrounding it.

  Alternatively, as shown in FIG. 10B, the minimum value Dmin + DR / γ may be used as the lower limit threshold REFb, and the maximum value Dmax−DR / γ may be used as the upper limit threshold REFu. Note that γ is a value of 1 or more and is determined by the reading error of the scan unit 10e. A smaller value is used as γ as the reading error is smaller, and a larger value is used as the reading error is larger.

Alternatively, the upper limit threshold value REFu and the lower limit threshold value REFb may be calculated using the following formulas (4) and (5).
Upper threshold REFu = Dg + DEmax (4)
Lower threshold REFb = Dg−DMmax (5)

However, DEmax is the maximum value among the differences ΔD11, ΔD21, ΔD31, ΔD41, ΔD51, ΔD61, and ΔD71. DMmax is the maximum value among the differences ΔD12, ΔD22, ΔD32, ΔD42, ΔD52, ΔD62, and ΔD72.

  As shown in FIG. 11, the difference ΔD41 is an absolute value of the difference between the brightness of the target pixel 6C and the brightness of the surrounding pixel 6S41 adjacent to the lighter side of the change direction of the light and shade (that is, the edge direction) in the target pixel 6C. . The difference ΔD42 is an absolute value of the difference between the brightness of the target pixel 6C and the brightness of the surrounding pixel 6S42 adjacent to the dark side.

  The surrounding pixels 6S10 to 6S30 and 6S50 to 6S70 are the surrounding pixels 6S on the normal line L in the edge direction of the target pixel 6C, and are arranged so as to sandwich the target pixel 6C.

  The surrounding pixels 6S11 to 6S31 and 6S51 to 6S71 are the surrounding pixels 6S adjacent to the surrounding pixels 6S10 to 6S30 and 6S50 to 6S70 on the light side in the edge direction, respectively. The surrounding pixels 6S12 to 6S32 and 6S52 to 6S72 are surrounding pixels 6S adjacent to the surrounding pixels 6S10 to 6S30 and 6S50 to 6S70 on the dark side in the edge direction, respectively.

  Differences Δ11 to Δ31 and Δ51 to Δ71 are absolute values of differences between the brightness of the surrounding pixels 6S10 to 6S30 and 6S50 to 6S70 and the brightness of the surrounding pixels 6S11 to 6S31 and 6S51 to 6S71, respectively. Differences Δ12 to Δ32 and Δ52 to Δ72 are absolute values of differences between the brightness of the surrounding pixels 6S10 to 6S30 and 6S50 to 6S70 and the brightness of the surrounding pixels 6S12 to 6S32 and 6S52 to 6S72, respectively.

  In the example of FIG. 11, the maximum value DEmax is selected from among the seven differences. However, the maximum value DEmax may be selected from a larger difference or a smaller difference. For example, you may select from five differences (DELTA) 21, (DELTA) 31, (DELTA) 51, and (DELTA) 61. The number of differences to be selected may be matched to the size of the prescribed matrix MT. That is, when the defined matrix MT is composed of 7 × 7 cells as shown in FIG. 8, it is only necessary to select from seven differences, and when it is composed of 5 × 5 cells, it is only necessary to select from five differences. Further, the maximum value DEmax may be calculated using the lightness median value Dc instead of the average lightness Dg. The same applies to the maximum value DMmax.

  As shown in FIG. 12, the smoothing processing unit 101 includes a directivity smoothing processing unit 131, an edge direction determination unit 132, and a selector 133, and the brightness correction for each pixel in the character outline portion of the multilevel image 5 is performed as follows. You may go to

  The directivity smoothing processing unit 131, like the directivity smoothing processing unit 121 (see FIG. 5), average brightness in the vertical direction, the horizontal direction, the direction connecting the upper right and the lower left, and the direction connecting the upper left and the lower right. Dv1 to Dv4 are calculated.

  The edge direction determination unit 132 determines the edge direction in the target pixel 6C. The edge direction is the direction of change in shading, and can be determined by a known method.

  The selector 133 is an average related to the direction having the smallest angle with the edge direction determined by the edge direction determination unit 132 among the vertical direction, the horizontal direction, the direction connecting the upper right and the lower left, and the direction connecting the upper left and the lower right. Select brightness. For example, when the angle between the vertical direction and the edge direction is the smallest, the average brightness Dv1 is selected. Then, similarly to the selector 123, the correction value data 3C indicating the selected average brightness as the smoothing brightness Dh is sent to the binarization threshold value determination unit 102.

  In the embodiment described above, the inside / outside determination unit 103 refers to only the determination result for the immediately preceding pixel when the brightness of the target pixel 6C is between the lower limit threshold value REFb and the upper limit threshold value REFu. You may refer the result of a some determination.

  For example, the determination results for the four surrounding pixels 6S shown in FIG. 13A that have already been determined among the surrounding pixels 6S adjacent to the target pixel 6C are referred to as follows.

  The inside / outside determination unit 103 compares the brightness of the target pixel 6C and the brightness of each of the four surrounding pixels 6S. Of these four surrounding pixels 6S, the determination result for the surrounding pixel 6S having the brightness closest to the brightness of the target pixel 6C is taken as the determination result for the target pixel 6C.

  Alternatively, the determination results for the plurality of surrounding pixels 6S arranged in the respective directions indicated by the dotted lines in FIG. 13B that have already been determined among the surrounding pixels 6S within the predetermined distance from the target pixel 6C are as follows. Refer to.

  The inside / outside determination unit 103 selects a direction in which the surrounding pixels 6S having the same determination result are continuously arranged by a predetermined number (for example, two) or more. The determination results for these surrounding pixels 6S are the determination results for the target pixel 6C. When there are a plurality of such directions, a direction having the largest number of consecutive determinations may be selected. Alternatively, a plurality of appearance patterns in a direction in which a predetermined number or more of the same determination results are continuously arranged are prepared in advance, and for each pattern, the target pixel 6C should be the pixel inside the character or the pixel outside the character. Decide what to do. Then, it is only necessary to perform matching between the selected direction and each pattern, and determine that the pixel that is predetermined for the pattern having the highest degree of coincidence should be used.

  In the above-described embodiment, the image editing circuit 10j performs the process of converting the multi-valued image 5 into the binary image 7. However, it may be performed by causing the CPU 10a to execute a computer program. In this case, a program module having the processing procedure of the smoothing processing unit 101 to the pixel binarization processing unit 104 shown in FIG. 4 as a main routine, and a processing procedure of each part of the smoothing processing unit 101 shown in FIG. 5 or FIG. What is necessary is just to prepare the computer program which has the program module which uses as a subroutine. The computer program may be stored in the ROM 10c or the mass storage device 10d and executed by the CPU 10a.

  The main routine is, for example, as shown in FIG. That is, the following processing is performed for each pixel of the multi-valued image 5 read by the scan unit 10e. Smoothing processing is performed on the multi-valued image 5 (# 11). One pixel is selected in a predetermined order, and is designated as the target pixel 6C (# 12 or # 16). A plurality of threshold values (center threshold value REFc, upper limit threshold value REFu, and lower limit threshold value REFb) are calculated for the target pixel 6C (# 13). Based on these threshold values and determination result data 3D, it is determined whether the pixel of interest 6C should be a pixel inside or outside the character (# 14). The method of determination is as described above with reference to FIG. Then, the target pixel 6C is binarized into either “0” or “1” according to the determination result (# 15).

  As shown in FIG. 15, a series of processes from smoothing to binarization may be executed for each pixel.

  In addition, the configuration of the entire image forming apparatus 1 or each unit, processing contents, processing order, image configuration, and the like can be appropriately changed in accordance with the spirit of the present invention.

1 Image forming device (image processing device)
10e Scan unit 101 Smoothing processing unit (smoothing processing means)
102 Binarization threshold value determination unit (threshold value determination means)
103 Inside / outside determination unit (binary determination means)
104 Pixel binarization processing unit (binary determination means)
5 Multi-valued image 6C Target pixel 6S Surrounding pixel

Claims (15)

An image processing apparatus that binarizes a multi-value image that is reproduced with three or more gradations and includes a line drawing portion representing a line drawing and a ground portion representing a background,
Threshold determination means for determining a first threshold and a second threshold smaller than the first threshold;
Among the pixels of the multi-valued image, when the gradation of the boundary pixel at the boundary between the line drawing portion and the background portion is between the first threshold value and the second threshold value, Binary determination means for determining a value to be the same value as the binary already determined for a specific position pixel at a specific position in the vicinity of the boundary pixel;
An image processing apparatus comprising: The threshold value determining means determines the first threshold value to a value smaller than the maximum value among the gradations of each of a plurality of surrounding pixels around the boundary pixel, and sets the second threshold value to the floor. To a value greater than the minimum value of the key,
The image processing apparatus according to claim 1. The threshold value determination unit is any one of the average value, the median value, and the value obtained by dividing the sum of the maximum value and the minimum value by 2 for the first threshold value . Determining the sum of the value and the first specific value, and determining the second threshold to be a value obtained by subtracting the first specific value from any one of the values,
The image processing apparatus according to claim 2. The multi-valued image is an image read by an image reading device,
The threshold value determining unit, by using a larger value as the reading error of the image reading apparatus is large and the first specific value, determining said first threshold and said second threshold value,
The image processing apparatus according to claim 3. The gradation of the multi-valued image is brightness,
The threshold value determining means includes the brightness of each of the one or more neighboring tangential pixels arranged on the boundary pixel and the tangent to the boundary of the boundary pixel, and the brightness of the pixel adjacent to the base portion side. The first absolute value of the absolute value of the difference is used as the first specific value to determine the first threshold, and the boundary pixel and the one or more tangential pixels By using, as the first specific value, the minimum absolute value of the absolute values of the differences between the respective brightness values and the brightness values of the pixels adjacent to the line drawing portion side, the second threshold value is used. To decide,
The image processing apparatus according to claim 3. The threshold value determining means determines the first threshold value to be a value obtained by subtracting a second specific value from the maximum value, and sets the second threshold value to the minimum value and the second specific value . Decide on the sum of
The image processing apparatus according to claim 2. The multi-valued image is an image read by an image reading device,
The threshold value determining unit, by using a smaller value read error is large in the image reading apparatus and said second specific value, determining said first threshold and said second threshold value,
The image processing apparatus according to claim 6. The specific position pixel is a pixel whose gradation is between the first threshold and the second threshold.
The image processing apparatus according to claim 1. The specific position pixel is a pixel having the smallest absolute value of a difference in gradation from the boundary pixel among pixels adjacent to the boundary pixel.
The image processing apparatus according to claim 1. The specific position pixels are pixels that have the same predetermined binary value around the boundary pixel and are arranged in a predetermined number or more in the same direction.
The image processing apparatus according to claim 1. Smoothing processing means for performing smoothing processing on the multi-valued image,
The threshold value determining means determines the first threshold value and the second threshold value based on the multi-valued image after the smoothing process is performed,
The binary determination means determines binary of the boundary pixel based on the multi-valued image after the smoothing process is performed.
The image processing apparatus according to claim 1. The smoothing processing unit selects, for each pixel of the multi-valued image, a gradation closest to the gradation of the pixel from gradations obtained by performing smoothing in each of a plurality of directions. Process
The image processing apparatus according to claim 11. The smoothing processing means performs the smoothing process by smoothing the edge direction for each pixel of the multi-valued image.
The image processing apparatus according to claim 11. An image processing method for binarizing a multi-valued image that is reproduced with three or more gradations and includes a line drawing portion representing a line drawing and a ground portion representing a background,
Determining a first threshold and a second threshold less than the first threshold;
Among the pixels of the multi-valued image, when the gradation of the boundary pixel at the boundary between the line drawing portion and the background portion is between the first threshold value and the second threshold value, Determining the value to be the same value as the binary value already determined for a specific position pixel at a specific position near the boundary pixel;
An image processing method. A computer program for use in a computer that binarizes a multi-valued image that is reproduced with three or more gradations and includes a line drawing portion representing a line drawing and a ground portion representing a background,
In the computer,
A process of determining a first threshold and a second threshold smaller than the first threshold,
Among the pixels of the multi-valued image, when the gradation of the boundary pixel at the boundary between the line drawing portion and the background portion is between the first threshold value and the second threshold value, Causing a value to be determined to be the same value as the binary value already determined for a specific position pixel at a specific position near the boundary pixel;
A computer program characterized by the above.

Images