JP3880958B2 - Image processing apparatus and image processing method - Google Patents

Description

  The present invention relates to an image processing apparatus and an image processing method for discriminating an image area in order to perform optimum image processing on input multi-value image data in a digital color copying machine, a color scanner, or the like.

  Conventionally, in an image processing apparatus such as a digital copying machine or a facsimile machine, image processing for improving image quality is performed on an image signal obtained by reading a document with a CCD (charge coupled device) sensor or the like. . That is, the image quality is improved by discriminating the halftone dot area, silver lead photograph area, and character area included in the document, and performing image processing according to the image characteristics of each area. For example, a smoothing process is performed on the halftone dot area to suppress moire, and an emphasis process is performed on the character area to clearly display characters.

  The following methods are known as image region discrimination methods performed in such image processing for improving image quality.

  Patent Document 1 discloses a black character processing method. In this method, an image is divided into predetermined blocks, and each region is discriminated using a feature parameter representing the character of a character or halftone dot image. That is, in a predetermined area including the target pixel, the average value is obtained as a threshold value, the difference between the average value and the target pixel value is obtained, binarized by comparison with the threshold value, and character determination is performed from the value. Further, a black character is discriminated from 8 pixels adjacent to the target pixel, and a difference from a halftone dot is extracted.

  Patent Document 2 discloses a method for discriminating a character / line drawing area, a halftone dot area, and a photographic area from CMY image data according to a difference in color distribution.

Patent Document 3 discloses a method of determining a character area and a photograph area by calculating an average value, a maximum value, and the like in a specific area and comparing them with a threshold value.
JP 9-163166 A (publication date: June 20, 1997) JP-A-8-56287 (Publication date: February 27, 1996) Japanese Unexamined Patent Publication No. 6-54180 (Publication date: February 25, 1994)

  However, the above-described conventional method has a problem that the accuracy is not sufficient for determining the character / line drawing area, the halftone dot area, and the photographic area, and there is a possibility that they are erroneously determined.

  In addition, although the above-described conventional method can distinguish a color image, there is a problem that the color of the edge portion may be different from the original color particularly when edge extraction / enhancement processing is performed on a color character. Occurs.

  The present invention has been made to solve the above-described problems, and an object of the present invention is to determine a black character area and a line drawing area with high accuracy for an area to which pixels of multi-value image data belong. Another object of the present invention is to provide an image processing apparatus and an image processing method.

In order to solve the above-described problem, the image processing apparatus of the present invention provides a black portion for a pixel in the first area including a target pixel in the image data for each color component and a pixel in the vicinity of the target pixel. Color detecting means for detecting color, counting means for counting the number of black and non-black pixels for each pixel in the first area, and detecting whether or not the color of the pixel of interest changes abruptly And a black detection means for detecting whether the pixel of interest belongs to a black character area or a line drawing area based on the output results of the color detection means, the count means, and the edge determination means. It is characterized by having.

  With the above configuration, by determining whether or not the pixel of interest on the image is black with respect to the first area composed of the pixel of interest of the image data for each color component and adjacent pixels adjacent to each other by the color determination unit, Image processing that is optimal for a black region is possible. In addition, by determining whether or not the pixel of interest on the image belongs to the black character edge region with respect to the image data in the first area by the edge determining means, it is possible to perform optimum image processing as an edge portion of a black character or a black line drawing. Is possible. Then, based on these results and the number of black and non-black pixels in the first area obtained by the counting means, the black character area and the line drawing area with respect to the area to which the pixels of the image data belong Can be determined with high accuracy.

  Therefore, it is possible to perform image processing optimal for the region to which the target pixel belongs on the image, and achieve high image quality. Therefore, when edge extraction / enhancement processing is performed on a color character as in the prior art, the color of the edge portion does not differ from the original color.

The image processing apparatus of the present invention, in addition to the above configuration, based on the image data of the first area, provided with a color feature extracting means for extracting a feature amount indicating a characteristic of the first area, the Based on the output result of the color feature amount extraction means, the color detection means and the count means perform processing.

With the above configuration , feature values are extracted from the first area for image data for each color component, and based on the result, optimal image processing can be performed for the region to which the target pixel belongs on the image. Can be achieved.

In the image processing apparatus according to the present invention , in addition to the above configuration, the color feature amount extraction unit includes a maximum value calculation unit that calculates a maximum value and a minimum value that calculates a minimum value for the density level of the pixels in the first area. A value calculating means, a maximum density difference calculating means for obtaining a maximum density difference, a density average value calculating means for obtaining an average value, a sum of absolute values of density level differences between adjacent pixels in the main scanning direction, and a sub scanning direction. It is characterized by comprising density difference sum calculating means for obtaining the sum of absolute values of density level differences between adjacent pixels.

With the above configuration , the color feature amount includes the maximum value, the minimum value, the maximum density difference, the average value, and the absolute value of the density level difference between adjacent pixels in the main scanning direction. The sum of the sum and the sum of absolute values of density level differences between pixels adjacent in the sub-scanning direction can be obtained.

  Therefore, the calculation of the color feature amount is easy, and the black identification accuracy can be further improved.

The image processing apparatus of the present invention, in addition to the above configuration, the edge determination unit, to the pixels of the first area, by convolving a mask filter for edge extraction, edge of the pixel of interest It is characterized by comprising edge detection means for performing detection.

With the above configuration, by or falling edge of edge detection means can extract a high precision black character edge portion enables optimum image processing as black character edge, higher image quality can be achieved.

The image processing apparatus of the present invention, in addition to the above configuration, the edge determination means, a pixel of the second area including the first area is binarized by a predetermined threshold density level, it is binarized Edge threshold selection means for generating a main edge threshold selection signal based on the number of inversions of each pixel in the main scanning direction and sub-scanning direction, and the density of the pixels included in the first area and the second area The maximum value, the minimum value, and the maximum density difference of the level are obtained, and the standard deviation of the density level of the pixel in the first area is obtained. Based on these results and the main edge threshold selection signal, the edge detection means And an edge feature quantity extracting means for generating an edge discrimination signal for verifying the edge detection result.

With the above configuration, by or falling edge of di feature extraction means, the maximum value of the density level of the pixel encompassed in the first area and second area, minimum, maximum density difference and, the density level of the pixel of the first area Based on the standard deviation and the main edge threshold selection signal obtained by the edge threshold selection means, an edge determination signal for determining the black character edge portion can be generated.

  Therefore, since the edge detection result obtained by the edge detection means can be verified by the edge discrimination signal, it is possible to prevent erroneous recognition of the edge portion and improve detection accuracy.

  Therefore, it is possible to extract a black character edge portion with high accuracy and perform optimal image processing on the black character edge portion, so that a higher image quality can be achieved.

In order to solve the above-described problem, the image processing method of the present invention provides a black portion for a pixel in a first area including a target pixel in image data for each color component and a pixel in the vicinity of the target pixel. A color detection step for detecting color, a count step for counting the number of pixels of black and a color other than black for each pixel in the first area, and detecting whether or not the color suddenly changes in the target pixel And a black detection step for detecting whether the pixel of interest belongs to a black character region or a line drawing region based on output results of the color detection step, the counting step, and the edge determination step It is characterized by having.

  With the above configuration, in the color determination step, by determining whether or not the pixel of interest on the image is black with respect to the first area composed of the pixel of interest of the image data for each color component and the neighboring pixels adjacent thereto, Image processing that is optimal for a black region is possible. Also, in the edge discrimination process, it is possible to determine whether the target pixel on the image belongs to the black character edge region or not for the image data in the first area, and thus optimal image processing as an edge portion for black characters, black line drawings, etc. Is possible. Based on these results and the number of black and non-black pixels in the first area obtained in the counting step, the black character region and the line drawing region are compared with the region to which the pixel of the image data belongs. Can be determined with high accuracy.

  Therefore, it is possible to perform image processing optimal for the region to which the target pixel belongs on the image, and achieve high image quality. Therefore, when edge extraction / enhancement processing is performed on a color character as in the prior art, the color of the edge portion does not differ from the original color.

As described above, the image processing apparatus according to the present invention detects the black portion of the pixels in the first area including the target pixel in the image data for each color component and the pixels in the vicinity of the target pixel. Detecting means, counting means for counting the number of pixels of black and non-black colors for the pixels in the first area, and edge discriminating means for detecting whether or not the color of the pixel of interest changes abruptly; And black detection means for detecting whether the pixel of interest belongs to a black character area or a line drawing area based on the output results of the color detection means, the count means, and the edge determination means It is.

  Therefore, the black character area and the line drawing area can be discriminated with high accuracy for the area to which the pixel of the image data belongs. Therefore, it is possible to perform optimal image processing for the region to which the target pixel belongs on the image, and it is possible to achieve high image quality.

The image processing apparatus of the present invention, in addition to the above configuration, based on the image data of the first area, provided with a color feature extracting means for extracting a feature amount indicating a characteristic of the first area, the Based on the output result of the color feature amount extraction means, the color detection means and the counting means perform processing.

Therefore, it is possible to optimum image processing on the area to which the pixel of interest belongs on images, an effect that high image quality can be achieved.

Minimum image processing apparatus of the present invention, in addition to the above configuration, the color feature extraction means is for determining the concentration levels of the pixels of the first area, the maximum value calculating means for determining a maximum value, a minimum value A value calculating means, a maximum density difference calculating means for obtaining a maximum density difference, a density average value calculating means for obtaining an average value, a sum of absolute values of density level differences between adjacent pixels in the main scanning direction, and a sub scanning direction. A density difference sum calculating means for obtaining a sum of absolute values of density level differences between adjacent pixels is provided.

Therefore , it is possible to easily calculate the color feature amount and to further improve the black discrimination accuracy.

The image processing apparatus of the present invention, in addition to the above configuration, the edge determination unit, to the pixels of the first area, by convolving a mask filter for edge extraction, edge of the pixel of interest It is the structure provided with the edge detection means which performs a detection.

Therefore , the black character edge portion can be extracted with high accuracy, and the optimum image processing can be performed as the black character edge portion, so that the image quality can be further improved.

The image processing apparatus of the present invention, in addition to the above configuration, the edge determination means, a pixel of the second area including the first area is binarized by a predetermined threshold density level, it is binarized Edge threshold selection means for generating a main edge threshold selection signal based on the number of inversions of each pixel in the main scanning direction and sub-scanning direction, and the density of the pixels included in the first area and the second area The maximum value, the minimum value, and the maximum density difference of the level are obtained, and the standard deviation of the density level of the pixel in the first area is obtained. Based on these results and the main edge threshold selection signal, the edge detection means An edge feature amount extraction unit that generates an edge determination signal for verifying an edge detection result is provided.

Therefore, the edge detection result obtained by or falling edge of edge detection means, it is possible to verify by the edge determination signal effect to prevent erroneous recognition of the edge portion, it is possible to improve the detection accuracy Play.

  Therefore, it is possible to extract a black character edge portion with high accuracy, and it is possible to perform optimum image processing on the black character edge portion.

As described above, the image processing method of the present invention detects a black portion of a pixel in a first area composed of a pixel of interest in image data for each color component and pixels in the vicinity of the pixel of interest. A detection step, a counting step for counting the number of pixels of black and a color other than black for each pixel in the first area, and an edge determination step for detecting whether the color changes abruptly at the target pixel, And a black detection step for detecting whether the pixel of interest belongs to a black character region or a line drawing region based on the output results of the color detection step, the counting step, and the edge determination step. is there.

  Therefore, the black character area and the line drawing area can be discriminated with high accuracy for the area to which the pixel of the image data belongs. Therefore, it is possible to perform optimal image processing for the region to which the target pixel belongs on the image, and it is possible to achieve high image quality.

[Embodiment 1]
One embodiment of the present invention will be described below with reference to FIGS. Note that the thick signal lines in FIGS. 1 to 17 indicate signal lines that can transmit multiple bits in parallel.

  First, an outline of the image processing apparatus according to the present embodiment will be described. The image processing apparatus according to the present embodiment performs high-precision area determination on an area to which pixels of multi-valued image data belong in order to perform optimal image processing on input multi-valued image data. be able to.

  In particular, it is determined from a plurality of elements obtained by extracting the high density characteristic parameter, the granularity characteristic parameter, and the periodic characteristic parameter, and the determination result is subjected to block correction, so that the character, halftone dot, The photograph area can be discriminated.

  Here, the high density characteristic parameter is used to determine how many color dots exist in a predetermined region. The graininess characteristic parameter is used to determine whether a color dot is a point, a line, or a surface. The periodicity characteristic parameter is used to determine the period in which color dots exist. The block correction is to improve the certainty of the determination by correcting the determination that could not be detected by the three characteristic parameters.

  In the image processing apparatus according to the present embodiment, each pixel on an image belongs from multi-value image data (for example, image data composed of cyan, magenta, and yellow) for each color component obtained by scanning a document. Color image area discrimination processing is performed for discriminating whether the area is a character area or a halftone dot area. The discrimination between the black character area and the line drawing area will be described in the second embodiment.

  First, as shown in FIG. 6, the image processing apparatus includes at least an input unit 1, an image processing unit 2, and a recording unit 4.

  The input unit 1 is an input device that scans a document image and obtains a digital signal of document image data. Specifically, the input unit 1 includes a CCD (charge coupled device) sensor (CCD sensor 1C, CCD sensor 1M, CCD sensor 1Y) for each color component of cyan, magenta, and yellow, and the obtained digital image. The image data C, M, and Y are output to the image processing unit 2 for each color. The multi-value image data input device is not limited to the CCD sensor. In the following, cyan, magenta, and yellow are sometimes referred to as C, M, and Y, respectively.

  The image processing unit 2 includes an image area determination unit 3 that determines an image area as preprocessing for processing a digital image. The image area determination unit 3 includes at least a first image area determination circuit 3a and a second image area determination circuit 3b. In addition, the image processing unit 2 is provided with a processing unit (not shown) that performs post-processing (filter processing) of the digital image in response to the result of the pre-processing in the image region determination unit 3.

  The recording unit 4 is a storage device that records the digital image processed in the image processing unit 2.

  In the present embodiment, the first image area discrimination circuit 3a for discriminating whether an image area is a character area or a halftone dot area will be described. The second image area discriminating circuit 3b for discriminating whether the image area is a black character area or a line drawing area will be described in the second embodiment.

  The outline of the configuration of the first image area discrimination circuit 3a will be described with reference to FIG. The first image area discrimination circuit 3a includes a first inspection circuit 10 (10C, 10M, 10Y) and a second inspection circuit 20 (20C, 20M, 20Y) for each color of cyan, magenta, and yellow, and a feature amount. An extraction circuit 30 and a correction circuit 40 are provided. The first inspection circuit 10, the second inspection circuit 20, the feature amount extraction circuit 30, and the correction circuit 40 are arranged in series in this order.

  The first image region discriminating circuit 3a receives the image data C, M, Y of the digital image obtained by the CCD sensors 1C, 1M, 1Y for each color of the input unit 1 and outputs a halftone discriminating result MESH. .

  Then, it is input to post-processing (filter processing) in the image processing unit 2, smoothing processing is performed on the halftone dot region to suppress moire, and emphasis processing is performed on the character region to display characters clearly. To improve the image quality.

  The first inspection circuit 10 calculates a high density characteristic parameter by detecting the maximum / minimum pixel density for each color, and calculates a granularity characteristic parameter by detecting run length and complexity. .

  The first inspection circuit 10C (FIG. 2A), the first inspection circuit 10M (FIG. 2B), and the first inspection circuit 10Y (FIG. 2C) have the same configuration. Only the first inspection circuit 10C will be described.

  As shown in FIG. 2A, the first inspection circuit 10C includes a line memory 11C, a maximum / minimum pixel calculation circuit (maximum / minimum pixel calculation means, maximum / minimum pixel calculation step) 12C, and a maximum / minimum. And a pixel peripheral information detection circuit (maximum / minimum pixel peripheral information detection means, local maximum / minimum pixel peripheral information detection step) 13C.

  The line memory 11C is a line memory arranged in m lines in parallel. The line memory 11C stores the cyan 8-bit image data C obtained by scanning the original image by the CCD sensor 1C of the input unit 1 in a state synchronized with the system clock, and always stores the image data C of m + 1 lines. Are output in parallel.

  The maximum / minimum pixel calculation circuit 12C receives the pixel density of the image data C output from the line memory 11C, and uses the cyan maximum determination value PDATAC, the minimum determination value DDATAC, and the average density value AVEC as local maximum / minimum pixel peripheral information. While outputting to the detection circuit 13C, the flag signal AVEKC, which is the result of comparing the threshold value of the density average value AVEC, is output to the second inspection circuit 20C. Here, the maximum determination value PDATAC, the minimum determination value DDATAC, and the density average value AVEC correspond to cyan high-density characteristic parameters.

  The contents of processing in the maximum / minimum pixel calculation circuit 12C are as follows. Note that all element regions such as a specific region range, a matrix, and a calculation area described below can be changed, and can correspond to any region. Although the image data C will be described here, the image data M and Y are processed in the same manner.

  Here, all arbitrarily set values (threshold values) used in the following calculations and determination formulas are integers in the range of −255 to +255, and are set as register values by the CPU (not shown) of the image processing unit 2. The These arbitrarily set values can also be changed according to the format of the image data to be used and the output device.

  Each of the above high density characteristic parameters is calculated within a specific determination block of (m + 1) × n. Therefore, it is possible to determine how many color dots exist in the (m + 1) × n specific determination block based on the high density characteristic parameter.

  In the present embodiment, the high-density feature parameter is calculated in the specific region range shown in FIG. 7, that is, a block of 7 pixels × 13 pixels centered on the center pixel P0 (corresponding to 6 stages of line memory). The center column in FIG. 7 is the target pixel P, and the center pixel is the center pixel P0. A cyan density average AVEC of 7 pixels × 9 pixels centered on the central pixel P0 is obtained.

  Further, the maximum and minimum pixels are calculated using the 3 × 3 matrix A and the 5 × 5 matrix B as shown in FIG. At this time, the number and types of matrix pixels that can be taken differ depending on the position of the target pixel P.

  As shown in FIG. 8A, the 3 × 3 matrix A corresponds to one 2 × 3 matrix Aa corresponding to the uppermost pixel among the pixels of interest in which seven pixels are vertically arranged, and to the lowermost pixel. One 2 × 3 matrix Ac and five 3 × 3 matrices Ab corresponding to the pixels in between are included. Further, as shown in FIG. 8B, the 5 × 5 matrix B includes one 3 × 5 matrix Ba corresponding to the uppermost pixel among the pixels of interest in which 7 pixels are vertically arranged, and 2 from the top. One 4 × 5 matrix Bb corresponding to the first pixel, one 3 × 5 matrix Be corresponding to the lowermost pixel, and one 4 × 5 matrix Bd corresponding to the second pixel from the bottom, , Three 5 × 5 matrices Bc corresponding to the pixels in between are included.

  Then, the cyan density difference sum PSC, which is the sum of the absolute differences between the target pixel P and the peripheral pixels Pij in FIGS. 8A and 8B, is obtained. Further, the number of peripheral pixels Pij having the same density value as the target pixel P is counted to obtain the cyan density matching pixel number EQC.

  The cyan density average value AVEC, density difference total PSC, and density matching pixel number EQC obtained here are the density average value AVE and density difference total PS of the following conditional expressions (A), (B), (C). , Density matching pixel numbers EQ are respectively applied to determine whether the target pixel P is maximum or minimum, and to determine the degree of density in the region. As a result of the determination, the maximum / minimum pixel calculation circuit 12C outputs the cyan maximum determination value PDATAC, the minimum determination value DDATAC, and the flag signal AVEKC as 1-bit signals.

Conditional expression (A) for determining the target pixel maximum
(P> AVE-AK1 and P ≧ Pij and PS> C3 and EQ <THEQ3) [3 × 3 matrix] or
(P> AVE-AK1 and P ≧ Pij and PS> C5 and EQ <THEQ3) [3 × 3 matrix] or
(P> AVE-AK1 and P ≧ Pij and PS> C5 and EQ <THEQ5) [5 × 5 matrix]
Conditional expression (B) for determining minimum pixel of interest
(P> AVE + AK2 and P ≦ Pij and PS> C3 and EQ <THEQ3) [3 × 3 matrix] or
(P> AVE + AK2 and P ≦ Pij and PS> C5 and EQ <THEQ3) [3 × 3 matrix] or
(P> AVE + AK2 and P ≦ Pij and PS> C5 and EQ <THEQ5) [5 × 5 matrix]
Conditional expression for determining degree of density in region (C)
AVE> THAVE
However,
AK1: Signed 8bit
AK2: Signed 8bit
C3: TH3M × COUNT3
C5: TH5M × COUNT5
TH3M = THMM3 (AVE> THMM1)
= THMM5 (AVE> THMM2)
= THMM7 (other than above)
TH5M = THMM4 (AVE> THMM1)
= THMM6 (AVE> THMM2)
= THMM8 (other than above)
THMM1 to 9: Threshold (Set value by register: Any)
THMM3 <THMM4, THMM5 <THMM6, THMM7 <THMM8
COUNT3: 5or8 (determined by the number of surrounding pixels in the matrix in Fig. 3)
COUNT5: 10or13or16 (determined by the number of surrounding pixels in the matrix of FIG. 3)
THAVE: Threshold (Set value by register: Arbitrary)
If the conditional expression (A) is satisfied, the target pixel P is determined to be a maximum value, and “1” is output as the cyan maximum determination value PDATAC. On the other hand, if the conditional expression (B) is satisfied, the target pixel P is determined to be a minimum value, and “1” is output as the cyan minimum determination value DDATAC. In other cases, “0” is output to the maximum determination value PDATAC and the minimum determination value DDATAC. If the conditional expression (C) is satisfied, “1” is output as the cyan flag signal AVEKC.

  Next, the maximum / minimum pixel peripheral information detection circuit 13C receives the cyan maximum determination value PDATAC, the minimum determination value DDATAC, and the density average value AVEC output from the maximum / minimum pixel calculation circuit 12C as inputs, and the cyan run length. PRUNC and the degree of congestion PBUSYC are calculated and output to the second inspection circuit 20C. Here, the run length PRUNC and the degree of congestion PBUSYC correspond to the cyan graininess characteristic parameter.

  The contents of processing in the local maximum / minimum pixel peripheral information detection circuit 13C are as follows. Here, the image data C will be described, but the image data M and Y are similarly processed.

  The local maximum / minimum pixel peripheral information detection circuit 13C uses the local maximum determination value PDATAC, the local minimum determination value DDATAC, and the density average value AVEC obtained by the local maximum / minimum pixel calculation circuit 12C, in the specific processing area (FIG. 7). The run length (peripheral information) PRUNC for the pixel of interest P is calculated and output.

  As shown in FIG. 9, the main scanning direction is constant, but the sub-scanning direction is a direction that run length PRUN (cyan run length PRUNC, magenta run length PRUNM, yellow run length PRUNY) can take depending on the position of the target pixel P. And its number changes. There are seven ways of taking run length PRUN in FIGS. 9A to 9G for each target pixel, but the maximum value is 6 (FIGS. 9A and 9G).

  Specifically, the cyan density average value AVEC obtained by the maximum / minimum pixel calculation circuit 12C is applied to the density average value AVE of the following conditional expression (D), and the conditional expression (D) is established. Until then, the number of pixels is counted as RUN1, RUN2, RUN3, and RUN4 in the four directions of the main scanning direction and the sub-scanning direction (direction 1 to direction 4 shown in FIG. 9B), and the maximum value is cyan. Run length (peripheral information) PRUNC is output.

Conditional expression for run length determination (D)
Pij ≦ AVE ± BK1 (when PDATA = 1)
Pij> AVE ± BK2 (when DDATA = 1)
However,
BK1, BK2: Threshold value (Register setting value: Any)
PRUN = MAX (RUN1, RUN2, RUN3, RUN4)
RUN1, RUN2, RUN3, RUN4: Number of pixels in four directions Further, the maximum / minimum pixel peripheral information detection circuit 13C is configured to detect each pixel included in the run length and the target pixel P in four directions of the main scanning direction and the sub-scanning direction. Are calculated as BUSY1, BUSY2, BUSY3, and BUSY4, and the value of the run length PRUNC having the maximum run length is output as the busyness PBUSYC.
Conditional expression for determining the degree of congestion (E)
PBUSY = PBUSY1 (when PRUN = RUN1)
= PBUSY2 (when PRUN = RUN2)
= PBUSY3 (when PRUN = RUN3)
= PBUSY4 (when PRUN = RUN4)
However, if there are a plurality of values equal to the run length PRUNC, the minimum value of the sum of absolute difference values is output as the busyness (BUSY) PBUSYC. For example, when the run lengths in the direction 1 and the direction 2 are the same and are both the maximum value, the congestion degree for the run lengths in both directions and the minimum value is output as the congestion degree PBUSYC. When run lengths in all directions are 0, 0 is output as the congestion degree PBUSYC.
Example) PBUSY = MIN (BUSY1, BUSY2) (when PRUN = RUN1 = RUN2)
PBUSY = 0 (when PRUN = 0)
As described above, the first inspection circuit 10 receives the image data C, M, and Y from the CCD sensors 1 of the input unit 1 for each color, and inputs the flag signals AVEKC, AVEKM, AVEKY, run lengths PRUNC, PRUNM. , PRUNY, and congestion levels PBUSYC, PBUSYM, and PBUSYY are output to the second inspection circuit 20. Further, during the process, the maximum / minimum pixel calculation circuit (maximum / minimum pixel calculation means, maximum / minimum pixel calculation step) 12 calculates the high density characteristic parameter. In the local maximum / minimum pixel peripheral information detection circuit (local maximum / minimum pixel peripheral information detection means, local maximum / minimum pixel peripheral information detection step) 13, the run length PRUN and the degree of complexity PBUSY are obtained, and the granularity characteristic parameter is calculated.

  The second inspection circuit 20 sets a weight ratio and detects periodicity. The second inspection circuit 20C (FIG. 3A), the second inspection circuit 20M (FIG. 3B), and the second inspection circuit 20Y (FIG. 3C) have the same configuration, so Only the second inspection circuit 20C will be described.

  As shown in FIG. 3A, the second inspection circuit 20C includes a pixel weight switching signal calculation circuit (pixel weight switching signal calculation means) 21C, a weight ratio setting circuit 22C, and a first periodicity detection circuit (period Periodicity calculating means, periodicity calculating step) 23C and a second periodicity detecting circuit (periodicity calculating means, periodicity calculating step) 24C.

  The pixel weight switching signal calculation circuit 21C receives the flag signal AVEKC, the run length PRUNC, and the congestion degree PBUSYC for the cyan image data C output from the first inspection circuit 10C, and calculates the pixel weight switching signal WSC. Output.

  Specifically, in the pixel weight switching signal calculation circuit 21C, the cyan maximum determination value PDATAC, the minimum determination value DDATAC, the run length PRUNC, and the congestion degree PBUSYC are expressed as the maximum determination value PDATA, the minimum determination value DDATA, The cyan pixel weight switching signal WSC is obtained by applying it to the run length PRUN and the congestion degree PBUSY, and is output to the weight ratio setting circuit 22C and the weight switching signal count circuit 31 of the feature amount extraction circuit 30. Although the image data C will be described here, the image data M and Y are similarly processed.

Conditional expression for determining pixel weight switching signal (F)
WS = 0 (when PDATA = 0 or DDATA = 0)
= 1 (when PDATA = 1 and PRUN> THPRO and PB <THPB0)
= 1 (when PDATA = 1 and PRUN> THPR1 and PB <THPB1)
= 1 (when PDATA = 1 and PRUN> THPR2 and PB <THPB2)
= 2 (when PDATA = 1 and PRUN <THPR3 and PB> THPB3)
= 2 (when PDATA = 1 and PRUN <THPR4 and PB> THPB4)
= 2 (when PDATA = 1 and PRUN <THPR5 and PB> THPB5)
= 3 (when PDATA = 1 and other than above)
= 4 (when DDATA = 1 and PRUN> THPR6 and PB <THPB6)
= 4 (when DDATA = 1 and PRUN> THPR7 and PB <THPB7)
= 4 (when DDATA = 1 and PRUN> THPR8 and PB <THPB8)
= 5 (when DDATA = 1 and PRUN <THPR9 and PB> THPB9)
= 5 (when DDATA = 1 and PRUN <THPR10 and PB> THPB10)
= 5 (when DDATA = 1 and PRUN <THPR11 and PB> THPB11)
= 6 (DDATA = 1 and other than above)
However,
PB = "100000000" (when PRUN = 0) [binary notation]
= PBUSY (when PRUN = 1)
= PBUSY / 2 (PRUN = 2)
= PBUSY × (1/4 + 1/16 + 1/64) (when PRUN = 3)
= PBUSY / 4 (when PRUN = 4)
= PBUSY × (1/8 + 1/16 + 1/64) (when PRUN = 5)
= PBUSY × (1/8 + 1/32 + 1/64) (when PRUN = 6)
THPB0 to 11, THPR0 to 11: Threshold value (Register setting value: Any)
THPB3>THPB4>THPB0>THPB1>THPB2> THPB5
THPB9>THPB10>THPB6>THPB7>THPB8> THPB11
THPR0>THPR1>THPR2>THPR3>THPR4> THPR5
THPR6>THPR7>THPR8>THPR9>THPR10> THPR11
Next, the weight ratio setting circuit 22C receives the cyan pixel weight switching signal WSC output from the pixel weight switching signal calculation circuit 21C as an input, and uses the following equation (G) to represent the halftone dot classification flag PDWEC for cyan. Is output to the first periodicity detection circuit 23C, the second periodicity detection circuit 24C, and the OR circuit 33 of the feature amount extraction circuit 30. The image data M and Y are similarly processed in the weight ratio setting circuits 22M and 22Y.

  That is, in each color plane shown in FIG. 10A, the number of pixels that are maximum or minimum where the pixel weight switching signal WS (WSC, WSM, WSY)> 0, and the network where the pixel weight switching signal WS = 2 or 5 is satisfied. The number of pixels closer to the point is obtained, and the halftone dot classification flag PDWE (PDWEC, PDWEM, PDWEY) is obtained by the following conditional expression (G).

Conditional expression for weight ratio determination (G)
PDWEC = 1 ((Number of WSC = 2or5) × 64> (Number of WSC = 0) × THWS0C)
= 0 (other than above)
PDWEM = 1 ((Number of WSM = 2or5) × 64> (Number of WSM = 0) × THWS0M)
= 0 (other than above)
PDWEY = 1 ((Number of WSY = 2or5) × 64> (Number of WSY = 0) × THWS0Y)
= 0 (other than above)
However,
PDWEC = 0 (when the number of WSC> 0 <THWS1)
PDWEM = 0 (when the number of WSM> 0 <THWS1)
PDWEY = 0 (when the number of WSY> 0 <THWS1)
THWS0,1: Threshold (value set by register: arbitrary)
Further, when AVEK = 1 obtained by the maximum / minimum pixel calculation circuits 12C, 12M, and 12Y, the number of (WSC, WSM, WSY = 2or5) is zero. Further, (the number of WSC, WSM, WSY = 2or5) has an upper limit of 127, and values higher than that converge to 127.

  Next, the first periodicity detection circuit 23C calculates the periodicity A of the color dots using the pixel weight switching signal WSC. The first periodicity detection circuit 23C receives the halftone dot classification flag PDWEC for cyan from the weight ratio setting circuit 22C, obtains the periodicity A calculation result P1C for cyan, and sends it to the OR circuit 32C of the feature amount extraction circuit 30. Output.

  The “color dot periodicity A” is a parameter for performing halftone dot discrimination in a narrow area. In the following, the image data C will be described, but the image data M and Y are similarly processed.

  Here, a calculation area as shown in FIG. 11 is set. An area of 31 horizontal pixels × 7 vertical pixels centered on the target pixel P is defined as an area AS1. An area having the same size and adjacent to the left of the target pixel P is referred to as area AS0, and an area adjacent to the right is referred to as area AS2. Therefore, the 15 pixels on the left of the area AS1 are the 15 pixels on the right of the area AS0, and the 15 pixels on the right of the area AS1 are the 15 pixels on the left of the area AS2.

  Furthermore, as shown in FIG. 12, for calculating the feature quantity of periodicity A, areas (calculation areas) C0 to C15 arranged in two vertical rows and eight horizontal rows are set in area AS1. Area C0 is an area of 4 vertical pixels × 5 vertical pixels sharing the right and lower 1 pixel with the adjacent area, and areas C1 to C6 are 4 vertical pixels sharing the right, left, and lower 1 pixel with the adjacent area. An area of 6 pixels in length, area C7 is an area of 4 pixels in length × 4 pixels sharing the left and lower pixels with the adjacent area, and area C8 is 4 pixels in length sharing the right and upper 1 pixels with the adjacent area. Pixel x 5 pixels area, areas C9 to C14 are 4 pixels x 6 pixels area sharing the right, left and top 1 pixels with the adjacent area, area C15 is the adjacent area and left and top 1 pixels This is an area of vertical 4 pixels × vertical 4 pixels sharing the minutes.

  The contents of processing in the first periodicity detection circuit 23C are as follows. First, in each area C0 to C15 (FIG. 12) in the area AS1, the number of pixels CSP0 to CSP15 satisfying the pixel weight switching signal WSC = 2or3 (maximum) and the number of pixels satisfying the pixel weight switching signal WSC = 5or6 (minimum). CSD0 to CSD15 are obtained. Then, the maximum value API (I = 0 to 15) (AP0 to AP15) and the minimum value ADI (I = 0 to 15) (AD0 to AD15) for the areas C0 to C15 are expressed by the conditional expression (H) shown below. calculate.

Conditional expression for determination of extreme value existence (H)
API = 1 (when CSPI> 0 and CSPI <CW0) (I = 0 to 15)
= 0 (other than above)
ADI = 1 (when CSDI> 0 and CSDI <CW0) (I = 0 to 15)
= 0 (other than above)
However,
CW0: Threshold value (Register setting value: Any)
Next, the sum of the maximum value API and the minimum value ADI (I = 0 to 15) is obtained for each area AS0, AS1, AS2.

  Subsequently, the periodicity A is calculated by the following equation, and the periodicity A calculation result P1 (1 bit) is output.

  The periodicity A calculation result P1, which is the above calculation result, is output for each of the image data C, M, and Y. However, the maximum value sum ASPJ obtained for the image data C, M, and Y, respectively, when the parameter RPJ = 1 (J = 0, 1, 2) for the maximum for determining whether or not the halftone dot is periodic. = 0 (J = 0, 1, 2), and when the parameter RDJ = 1 (J = 0, 1, 2) for determining whether the halftone dot is periodic, the sum of the minimum values ASDJ = 0 ( J = 0, 1, 2). Further, when the parameter RJ = 1 (J = 0, 1, 2) for the extreme value for determining whether or not the halftone dot is periodic, ASPJ = ASDJ = 0 (J = 0, 1, 2). , ASPJ = ASDJ = 0 for all output values of the image data C, M, Y. A method for calculating each of the parameters RPJ, RDJ, and RJ will be described later.

  Subsequently, the second periodicity detection circuit 24C calculates the periodicity B of the color dots using the pixel weight switching signal WSC. The second periodicity detection circuit 24C receives the halftone dot classification flag PDWEC for cyan from the weight ratio setting circuit 22C, obtains the periodicity B calculation result P2C for cyan, and sends it to the OR circuit 32C of the feature amount extraction circuit 30. Output.

  The “color dot periodicity B” is a parameter for performing halftone dot discrimination in a wide area. In the following, the image data C will be described, but the image data M and Y are similarly processed.

  Here, as shown in FIG. 13A, areas (calculation areas) D0 to C3 are set in the area AS1. Area AS1 is divided into four by areas D0 to D3. Areas D0 to D2 are areas of horizontal 8 pixels × vertical 7 pixels, and are arranged in order from the left end of area AS1. The area D3 is a region of horizontal 7 pixels × vertical 7 pixels, and is arranged at the right end of the area AS1 and adjacent to the area D2. Therefore, the target pixel P is located at the right end of the area D1. Note that the area in which the areas D0 to D3 are arranged in this manner is referred to as an area BS1, and the left and right areas that share a half of the area with the area BS1 are also referred to as areas BS0 and BS2, respectively.

  Further, as shown in FIG. 13B, areas (calculation areas) E0 to E2 are set in the area AS1. Areas E0 to E2 are areas of horizontal 8 pixels × vertical 7 pixels, and are arranged in order from a position 4 pixels away from the left end of area AS1. Therefore, there is a space of 3 pixels between the right end of the area E2 and the right end of the area AS1. Note that the area in which the areas E0 to E2 are arranged in this manner is referred to as an area CS1, and the left and right areas sharing the half of the area with the area CS1 are also referred to as areas CS0 and CS2, respectively.

  The contents of processing in the second periodicity detection circuit 24C are as follows. First, in each area D0 to D3 (FIG. 13A) in the area BS1, the number of pixels DSP0 to DSP3 that satisfy the pixel weight switching signal WSC = 2or3 (maximum) and the pixel weight switching signal WSC = 5or6 (minimum) are set. The number of pixels DSD0 to DSD3 to be satisfied is obtained.

  Similarly, in each of the areas E0 to E2 in the area CS1 (FIG. 13B), the number of pixels ESP0 to ESP2 satisfying the pixel weight switching signal WSC = 2 or 3 (maximum) and the pixel weight switching signal WSC = 5 or 6 (minimum). The number of pixels ESD0 to ESD2 satisfying the above is obtained.

Subsequently, maximum pixel detection flags BSP and CSP that detect the presence of the maximum pixel in the area and minimum pixel detection flags BSD and CSD that detect the presence of the minimum pixel in the area are obtained by the following equation (I).
Conditional Expression for Existence of Extreme Value (I)
BSPI = 1 (when DSPI> 0 and DSPI <CW2) (I = 0 to 3)
= 0 (other than above)
BSDI = 1 (when DSDI> 0 and DSDI <CW2) (I = 0 to 3)
= 0 (other than above)
CSPI = 1 (when ESPI> 0 and ESPI <CW2) (I = 0 ~ 2)
= 0 (other than above)
CSDI = 1 (when ESDI> 0 and ESDI <CW2) (I = 0 to 2)
= 0 (other than above)

  Next, using the sum BSP and BSD of the areas BS0 to BS2 and CS0 to CS2 calculated by the above formula (I), the maximum pixel flag number SSP and the minimum pixel flag are calculated by the following formula (J). The number SSD, the area number NNP in the areas D0 to D3, and the area number NND in the areas E0 to E2 are obtained.

Conditional expression (J) for determining which area in FIGS. 13A and 13B is to be selected
SPJ = BSPJ, NPJ = 4 (when BSPJ> CSPJ)
SPJ = CSPJ, NPJ = 3 (when BSPJ ≤ CSPJ)
SDJ = BSDJ, NDJ = 4 (when BSDJ> CSDJ)
SDJ = CSDJ, NDJ = 3 (when BSDJ ≤ CSDJ)
However,
J = 0,1,2

  Then, the periodicity B is calculated by the following equation (K), and the periodicity B calculation result P2 (1 bit) is output.

Conditional expression for periodicity determination (K)
P2 = 1 (when (SSP × 16> THBS × NNP) or (SSD × 16> THBS × NND))
P2 = 0 (Other than above)
However,
THBS: Threshold (Set value by register: Arbitrary)
When RPJ = 1 (J = 0,1,2), SPJ = 0, NPJ = 4 (J = 0,1,2) and when RDJ = 1 (J = 0,1,2), SDJ = 0, NDJ = 4 (J = 0,1,2). Furthermore, when RJ = 1 (J = 0,1,2), SPJ = SDJ = 0, NPJ = NDJ = 4 (J = 0,1,2), but all output values of C, M, Y In contrast, SPJ = SDJ = 0, NPJ = NDJ = 4.

  Here, a calculation method of the parameter RPJ for the maximum time, the parameter RDJ for the minimum time, and the parameter RJ for the extreme value used in the above calculation will be described. First, the parameters RPJ and RDJ are calculated by the following conditional expression (L) for each plane of the image data C, M, and Y.

Conditional expression for determining periodicity (L)
RPJ = 1 (when (number of WS = 2,3) × 64 <THRU × (number of WS = 1,2,3))
= 0 (other than above)
RDJ = 1 ((number of WS = 4,5) x 64 <THRU x (number of WS = 4, 5, 6))
= 0 (other than above)
However,
J = 0,1,2
THRU: Threshold (Set value by register: Any)
The parameter RJ is calculated as follows. First, as shown in FIG. 14, if at least one of the planes of the image data C, M, Y satisfies WS = 1, 2, 3 (maximum) and WS = 4, 5, 6 (minimum), CP = 1 (maximum), CD = 1 (minimum).

  Here, an RJ calculation area as shown in FIGS. 15A and 15B is set. In FIG. 15A, the specific area range (FIG. 2) is divided into four areas of areas CU0 to CU3. The areas CU0 to CU2 are areas of 1 vertical pixel × 31 horizontal pixels, and are arranged in this order from the upper end side of the specific area range. The area CU3 is an area of 4 vertical pixels × 31 horizontal pixels, and is arranged adjacent to the area CU2 at the lower end of the specific area range. The division shown in FIG. 15A is a division that can prevent erroneous determination of halftone dot determination at the bottom of the image. In FIG. 15B, the specific area range (FIG. 2) is divided into four areas of areas CT0 to CT3. The areas CT0 to CT2 are areas of 1 vertical pixel × 31 horizontal pixels, and are arranged in this order from the lower end side of the specific area range. The area CT3 is an area of 4 vertical pixels × 31 horizontal pixels, and is arranged adjacent to the area CT2 at the upper end of the specific area range. The division shown in FIG. 15B is a division that can prevent erroneous determination of halftone dot discrimination at the top of the image.

  Then, the following condition processing is performed on the area shown in FIG. 15, and the parameters RPJ and RDJ are calculated.

Conditional expression for extreme value existence determination (M)
Maximum When RPJ = 1 (J = 0,1,2)
(CTP0> THCT0 and CTP1 <THCT1 and CTP2 <THCT2 and CTP3 <THCT3) or
(CUP0> THCU0 and CUP1 <THCU1 and CUP2 <THCU2 and CUP3 <THCU3)
Otherwise, RPJ = 0 (J = 0,1,2)
When the following conditions are met, RDJ = 1 (J = 0,1,2)
(CTD0> THCT0 and CTD1 <THCT1 and CTD2 <THCT2 and CTD3 <THCT3) or
(CUD0> THCU0 and CUD1 <THCU1 and CUD2 <THCU2 and CUD3 <THCU3)
Otherwise, RDJ = 0 (J = 0,1,2)
However,
CTP0-3: Number of CP = 1 in CT0-3 region
CTD0-3: Number of CD = 1 in CT0-3 region
CUP0-3: Number of CP = 1 in CU0-3 region
CUD0-3: Number of CD = 1 in CU0-3 region
THCT0 to 3: Threshold (Set value by register: Any)
THCU0 to 3: Threshold value (Register setting value: Any)
Next, from the obtained result, the parameter RJ is calculated by an OR circuit as a logical sum of the parameter RPJ and the parameter RDJ.

RJ = (RPJ) OR (RDJ)
As described above, for each color, the second inspection circuit 20 uses the flag signal AVEK, run length PRUN, and complicated from the maximum / minimum pixel calculation circuit 12 and the maximum / minimum pixel peripheral information detection circuit 13 of the first inspection circuit 10C. Using the degree PBUSY as an input, the periodicity A calculation result P1, the periodicity B calculation result P2, the pixel weight switching signal WS, and the halftone dot classification flag PDWE are output to the feature amount extraction circuit 30. In the process, the pixel weight switching signal calculation circuit 21C counts the weight ratio WS to obtain the periodicity A and the periodicity B, thereby calculating the periodic characteristic parameter.

  As shown in FIG. 4, the feature quantity extraction circuit 30 includes a weight switching signal count circuit 31, an OR circuit 32C, an OR circuit 32M, an OR circuit 32Y, an OR circuit 33, and an OR circuit 34. It is configured.

  The feature amount extraction circuit 30 receives the periodicity A calculation result P1, the periodicity B calculation result P2, the pixel weight switching signal WS, and the halftone dot classification flag PDWE output from the second inspection circuits 20C, 20M, and 20Y. A halftone discrimination result flag SCRFLG is obtained and output to the correction circuit 40.

  As shown in FIG. 10, the weight switching signal count circuit 31 is obtained for each plane by the pixel weight switching signal calculation circuit (pixel weight switching signal calculation means) 21 (21C, 21M, 21Y) of the second inspection circuit 20. From the obtained pixel weight switching signal WS (WSC, WSM, WSY) (FIG. 10A), the pixel weight W (WC, WM, WY) is obtained for each plane according to the following conditional expression (N) ( In FIG. 10B, the maximum value is represented by one plane as the pixel weight WMAX. The square at the upper left corner of each plane indicates a unit of one pixel having the pixel weight switching signal WS and the pixel weight W value.

Conditional expression for pixel weight determination (N)
WMAX = MAX (WC, WM, WY)
However,
W = 0 (when WS = 0)
W = 0 (when WS = 1or4)
W = 3 (when WS = 2or5)
W = 1 (when WS = 3or6)
Furthermore, the weight switching signal count circuit 31 has the maximum number of continuous lines NRUN at which the pixel weight of 7 pixels continuous in the sub-scanning direction (vertical direction) becomes W = 0 with respect to the areas AS0, AS1, and AS2 shown in FIG. For each. Then, a halftone dot discrimination flag KOUT (1 bit) indicating a halftone dot result is obtained by the following conditional expression (O), and is output to the OR circuit 34.

Conditional expression for weight switching signal count flag determination (O)
When NRUN ≦ THRUN
PDSUM0 = Number of AREA0 W = 1 / Ks (However, PDSUM0 = 128, PDSUM0 = 127)
PDSUM1 = AREA1 W = 1 count / Ks (However, PDSUM1 ≧ 128, PDSUM1 = 127)
PDSUM2 = Number of AREA2 W = 1 / Ks (However, PDSUM2 = 128, PDSUM2 = 127)
When NRUN> THRUN
PDSUM0 = Number of AREA0 W = 1 / Ks (However, PDSUM0 = 128, PDSUM0 = 127)
PDSUM1 = AREA1 W = 1 count / Ks (However, PDSUM1 = 128, PDSUM1 = 127)
PDSUM2 = Number of AREA2 W = 1 / Ks (However, PDSUM2 = 128, PDSUM2 = 127)
However,
Ks = 2or4 (register setting value)
PDSUM = (PDSUM0 + PDSUM1 * X + PDSUM2)
X = 2or3or4 (register setting value)
KOUT = 1 (when PDSUM> THCSUM)
= 0 (other than above)
The OR circuits 32C, 32M, and 32Y include a first periodicity detection circuit (periodicity calculation means, periodicity calculation step) 23C, 23M, and 23Y and a second periodicity detection circuit (periodicity calculation means, Periodicity calculation step) The logical sums LPC, LPM, and LPY of the periodicity A calculation results P1C, P1M, and P1Y and the periodicity B calculation results P2C, P2M, and P2Y respectively input from 24C, 24M, and 24Y are obtained, and the OR circuit 34 Output to.

  The OR circuit 33 obtains a logical sum LPDWE of halftone dot classification flags PDWEC, PDWEM, and PDWEY inputted for each plane from the weight ratio setting circuits 22C, 22M, and 22Y, and outputs the logical sum LPDWE to the OR circuit 34.

  The OR circuit 34 receives the halftone dot discrimination flag KOUT from the weight switching signal count circuit 31, the logical sums LPC, LPM, LPY from the OR circuits 32C, 32M, 32Y, and the logical sum LPDWE from the OR circuit 33. The halftone dot discrimination result flag SCRFLG (1 bit) which is the logical sum of these is obtained and output to the correction circuit 40.

  Note that by inserting a selector on the signal line of the feature quantity extraction circuit 30, the number and type of feature quantities used for region discrimination can be changed according to the specifications of the image data and the image processing apparatus.

  As described above, the feature amount extraction circuit 30 includes the pixel weight switching signal calculation circuit 21, the weight ratio setting circuit 22, the first periodicity detection circuit 23, and the second periodicity detection circuit 24 of the second inspection circuit 20. The halftone dot determination result flag SCRFLG is obtained by inputting the periodicity A calculation result P1, the periodicity B calculation result P2, the pixel weight switching signal WS, and the halftone dot classification flag PDWE from the above and output to the correction circuit 40.

  The correction circuit 40 extracts a high density characteristic parameter, a granularity characteristic parameter, and a periodic characteristic parameter, and performs block correction on the image region discrimination result obtained by each element. Providing the correction circuit 40 can reduce erroneous determination of final image region determination.

  As shown in FIG. 5, the correction circuit 40 includes a FIFO memory 41 and a halftone block correction circuit 42.

  The halftone dot block correction circuit 42 receives the halftone dot discrimination result flag SCRFLG from the FIFO memory 41, obtains a halftone dot discrimination result MESH, and outputs it. Here, to the halftone block correction circuit 42, the signal from the FIFO memory 41 and the signal inputted in real time are inputted as a bus.

  Here, halftone dot block correction areas as shown in FIGS. 16A to 16C and FIGS. 17A to 17C are set and used in the processing of the halftone block correction circuit 42. . As the halftone dot block correction area, there are two types of matrix modes, each including three types of areas (FIGS. 16 and 17).

  FIGS. 16A to 16C show halftone block correction areas in a mode that is block correction in a matrix of 18 × 61 pixels. FIG. 16A shows an area divided so as to perform halftone block correction in the main scanning direction. FIG. 16B shows an area divided so as to perform halftone block correction in the sub-scanning direction. FIG. 16C shows an area divided so as to perform halftone block correction in consideration of the main scanning direction and the sub-scanning direction.

  17A to 17C show halftone block correction areas in a mode that is block correction in a matrix of 14 × 61 pixels. FIG. 17A shows an area divided so as to perform halftone block correction in the main scanning direction. FIG. 17B shows an area divided so as to perform halftone dot block correction in the sub-scanning direction. FIG. 17C shows an area which is divided so as to perform halftone block correction in consideration of the main scanning direction and the sub-scanning direction.

  The halftone dot block correction circuit 42 determines which of the two matrix modes (FIGS. 16 and 17) is prioritized between processing speed and high image quality when halftone block correction is actually performed. Select either one to use. In the three types of areas included in the halftone block correction area of the selected mode, the number of pixels having a halftone discrimination result flag SCRFLG = 1 is obtained. When the following conditional expression (P) is satisfied, the halftone discrimination result MESH is set to “1”, and otherwise, “0” is set as the final output result. That is, when the halftone dot discrimination result MESH = 1, the pixel is determined to be a halftone dot.

Conditional expression for halftone dot discrimination (P)
(MC0> THMC0 and MC1> THMC1 and MC2> THMC2 and MC3> THMC3)
or (MC4> THMC4 and MC5> THMC5 and MC6> THMC6)
or (SC0> THSC0 and SC1> THSC1 and SC2> THSC2 and SC3> THSC3)
or (SC16> THSC16 and SC17> THSC17 and SC18> THSC18 and SC19> THSC19)
or (CC1> THCC1 and CC2> THCC2 and CC3> THCC3 and CC4> THCC4)
However,
THMC0 to 6: Threshold (Set value by register: Any)
THSC 0 to 3, 16 to 19: Threshold value (Register setting value: Any)
THCC1 to 4: Threshold (setting value by register: arbitrary)
Therefore, the correction circuit 40 receives the halftone dot discrimination result flag SCRFLG from the OR circuit 34 of the feature quantity extraction circuit 30 and outputs the halftone dot discrimination result MESH to the image processing unit 2.

  As described above, the image processing apparatus according to the present embodiment determines the region characteristics to which each pixel on the image belongs from the multi-value image data C, M, and Y obtained by scanning the document. Then, a high density characteristic parameter, a granularity characteristic parameter, and a periodic characteristic parameter are extracted, and an image area is determined based on each element.

  As a result, the image processing apparatus can obtain each feature parameter from a specific area composed of neighboring pixels adjacent to the target pixel for C, M, and Y multivalued image data obtained by scanning a document with a CCD sensor or the like. Based on these, it is possible to more reliably determine whether the region to which the target pixel belongs on the image is a character region, a dot region, or a photographic region. Therefore, more optimal image processing is possible, and high image quality can be achieved.

  The image processing apparatus has a maximum / minimum detection circuit for detecting a maximum / minimum pixel for each color component C, M, Y from input multivalued image data obtained by scanning a document image, and has a high density. Sex characteristic parameters are calculated. As a result, the image processing apparatus detects maximum / minimum pixels for each color component for each of the image data C, M, and Y, and discriminates characters, halftone dots, and photographic regions based on the high-density characteristic parameters. Can do.

  Further, the image processing apparatus includes a local maximum / minimum pixel peripheral information calculation circuit that extracts a run length and a complexity for each image data C, M, Y of each color component from the local maximum / minimum pixel obtained from the local maximum / minimum detection circuit. The graininess feature parameter is calculated. As a result, the image processing apparatus extracts the run length and the complexity for each color component image data C, M, and Y from the maximum and minimum pixels, and based on the granularity feature parameter, the halftone dot region and others ( It is possible to discriminate between a continuous point and line area.

  The image processing apparatus calculates a pixel weight switching signal for extracting a pixel weight switching signal for each image data C, M, and Y of each color component from the run length and complexity obtained from the local maximum / minimum pixel peripheral information calculation circuit. It has a circuit.

  Thus, the image processing apparatus extracts a pixel weight switching signal for each image data C, M, and Y of each color component from the run length and the complexity, and counts the weight ratio within a certain region. A feature amount of image data can be extracted, and a character, a halftone dot, and a photograph area can be discriminated.

  The image processing apparatus has a periodicity calculation circuit that extracts the periodicity for each of the image data C, M, and Y of each color component of the pixel weight switching signal obtained from the pixel weight switching signal calculation circuit. A feature parameter is calculated.

  As a result, the image processing apparatus extracts periodicity for each of the image data C, M, and Y of each color component of the pixel weight switching signal, and calculates the periodicity characteristic parameter. It is possible to confirm and distinguish characters, halftone dots, and photo areas.

  The image processing apparatus includes a pixel weighting circuit that synthesizes a pixel weight switching signal obtained from the pixel weight switching signal calculation circuit for each of the color component image data C, M, and Y. Extract the amount.

  As a result, the image processing apparatus can discriminate characters, halftone dots, and photographic regions by synthesizing pixel weight switching signals and extracting feature amounts for region discrimination.

  Further, the image processing apparatus, when determining the region characteristics to which each pixel on the image belongs from multi-valued image data obtained by scanning a document, a high density characteristic parameter, a granularity characteristic parameter, and Periodic feature parameters are extracted, and parameters that use each element for image region discrimination can be selected according to image data.

  As a result, the image processing apparatus can select the parameter to be used for the determination of the image area according to the image data, so that characters, halftone dots, and photographic areas can be efficiently and quickly applied to various image data. Can be determined.

  Further, the image processing apparatus, when determining the region characteristics to which each pixel on the image belongs from multi-valued image data obtained by scanning a document, a high density characteristic parameter, a granularity characteristic parameter, and It has a correction circuit that extracts periodic characteristic parameters and performs block correction on the image region discrimination result obtained by each element.

  Accordingly, the image processing apparatus can reduce erroneous determination of final image region determination by extracting each parameter and performing block correction on the image region determination result obtained by each element.

[Embodiment 2]
The following will describe another embodiment of the present invention with reference to FIG. 6 and FIGS. For convenience of explanation, members having the same functions as those shown in the first embodiment are given the same reference numerals, and explanation thereof is omitted. In addition, the thick signal lines in FIGS. 18, 19, and 22 to 27 indicate signal lines that can transmit multiple bits in parallel.

  The image processing apparatus according to the present embodiment performs color image region determination based on multi-valued image data obtained by scanning a document to determine whether or not the region to which each pixel on the image belongs is a black character region. Process.

  In the present embodiment, the second image region determination circuit 3b provided in the image region determination unit 3 shown in FIG. 6 will be described.

  The outline of the configuration of the second image region discrimination circuit 3b will be described with reference to FIG. The second image area determination circuit 3b includes a line memory 101, a color determination circuit 102, an edge determination circuit (edge determination means, edge determination step) 103, and a determination processing circuit 104.

  The second image area discriminating circuit 3b receives line data 101C, 101M, and 101Y from the image data C, M, and Y obtained for each color of cyan, magenta, and yellow by the CCD sensors 1C, 1M, and 1Y of the input unit 1, respectively. Is converted into a 7-line 8-bit format for each color and input to the color determination circuit 102 and the edge determination circuit 103. The edge determination circuit 103 obtains the edge detection signal S134 and the edge determination signal S136, and based on these, the determination processing circuit 104 calculates the edge determination result EDGE and inputs it to the color determination circuit 102. The color determination circuit 102 receives the image data C, M, Y from the line memories 101C, 101M, 101Y and the edge determination result EDGE from the determination processing circuit 104, and outputs a black character determination result BLACK.

  The line memory 101 (101C, 101M, 101Y) is a line memory arranged in 6 lines in parallel. The line memory 101 is 8-bit image data C, M, 8 bits of each color, which is document image data obtained by scanning a document image by the CCD sensor 1 (1C, 1M, 1Y) (FIG. 6) of the input unit 1. Y is stored in parallel in 6 lines in synchronization with the system clock, and 7 lines of image data C, M, Y are always output in parallel. For example, in the case of 8-bit image data for each color of A3 and 600 dpi, a capacity of about 7 Kbytes per line is required. In the present embodiment, a case where the specific area is 7 pixels × 7 pixels is shown as an example, but the present invention is not limited to this. Therefore, the form of the line memory 101 is not limited to this.

  As described above, the line memories 101C, 101M, and 101Y convert the 8-bit image data C, M, and Y from the CCD sensors 1C, 1M, and 1Y into a format of 7 lines and 8 bits for each color to be described later. The data is output to the determination circuit 102 and the edge determination circuit 103.

  The color determination circuit 102 receives the 7-line 8-bit image data C, M, and Y of each color and the edge determination result EDGE from the determination processing circuit 104, integrates them, and outputs a black character determination result BLACK. .

  As shown in FIG. 19, the color determination circuit 102 includes a color feature amount extraction circuit (color feature amount extraction means) 121, a color detection circuit (color detection means, color detection step) 122, and a count circuit (count means, (Counting step) 123 and a black detection circuit (black detection means, black detection step) 124.

  The color feature quantity extraction circuit 121 includes a maximum value calculation circuit (maximum value calculation means) 121a, a minimum value calculation circuit (minimum value calculation means) 121b, a maximum density difference calculation circuit (maximum density difference calculation means) 121c, A density average value calculating circuit (density average value calculating means) 121d and a density difference total calculating circuit (density difference total calculating means) 121e are provided.

  The maximum value calculating circuit 121a, the minimum value calculating circuit 121b, the maximum density difference calculating circuit 121c, the density average value calculating circuit 121d, and the density difference sum calculating circuit 121e are located at the same coordinates in C, M, and Y in the specific area. For the image data to be processed, a maximum value feature amount MAX, a minimum value feature amount MIN, a maximum density difference feature amount SUB, a density difference average value feature amount AVE, and a density difference total feature amount PLUS are calculated.

  Here, the maximum value calculation circuit 121a, the minimum value calculation circuit 121b, and the maximum density difference calculation circuit 121c have density levels in all the pixels in a specific area centered on the target pixel in each of the image data C, M, and Y. The maximum value, the minimum value, and the maximum density difference are obtained. The density average value calculation circuit 121d calculates an average value of density levels in all pixels in the specific area. The density difference sum calculation circuit 121e is the sum of the absolute value of the density level difference between pixels adjacent in the main scanning direction within the specific area and the sum of the absolute value of the density level difference between pixels adjacent in the sub-scanning direction. Ask for.

  Here, the above-described feature amounts MAX, MIN, SUB, AVE, and PLUS are obtained by comparing and calculating image data C, M, and Y having the same coordinates in a 7 pixel × 7 pixel region for each color. . Therefore, 49 (7 × 7) pieces of each feature amount are obtained.

  The color detection circuit 122 receives the feature amounts calculated by the calculation circuits 121a to 121e of the color feature amount extraction circuit 121 and the image data C, M, and Y, and sets a threshold value set for each of the feature amounts. By comparing, the first color detection result COLOR and the second color detection result signal S126 are output to the count circuit 123 and the black color detection circuit 124.

  As shown in FIG. 19, the color detection circuit 122 includes a first color detection circuit 125 and a second color detection circuit 126.

  As shown in FIG. 20, the first color detection circuit 125 includes seven comparators 201 to 207, two selectors 208 and 209, a combinational circuit 210, an OR circuit 211, and an AND circuit 212. It is prepared for.

  The comparator 201 compares the minimum value feature value MIN calculated by the minimum value calculation circuit 121b with the threshold value THMIN, and when the minimum value feature value MIN is equal to or greater than the threshold value THMIN, "1" is set otherwise. "0" is output to the combinational circuit 210 as a signal S201.

  The comparator 202 compares the density difference total feature amount PLUS calculated by the density difference total calculation circuit 121e with the threshold value THPLUS, and when the density difference total feature amount PLUS is larger than the threshold value THPLUS, “1” is set otherwise. In this case, “0” is output to the combinational circuit 210 as the signal S202.

  The comparator 203 compares the maximum density difference feature quantity SUB calculated by the maximum density difference calculation circuit 121c with the threshold value THSUB. When the maximum density difference feature quantity SUB is smaller than the threshold value THSUB, “1” is set. In this case, “0” is output to the combinational circuit 210 as the signal S203.

  The combinational circuit 210 receives the signals S201, S202, and S203 from the comparators 201 to 203 and inputs “01 (2 bits)” when all are “1”, and “ 10 (2 bits) "is output to the selector 208 as the signal S210.

  The comparator 204 compares the density average value feature value AVE calculated by the density average value calculation circuit 121d with the threshold value THAVE. If the density average value feature value AVE is smaller than the threshold value THAVE, “1” is set. In this case, “0” is output to the selector 208 and the AND circuit 212 as the signal S204.

  The selector 208 uses the signal S204 from the comparator 204 as a control signal, and when the signal S204 is “0”, the signal S210 from the combinational circuit 210 is “01” or “10”, otherwise "00" is output to the selector 209 as a 2-bit signal S208.

  The comparator 205 compares the pixel image data C with the threshold value THC, and if the image data C is greater than or equal to the threshold value THC, “1” is obtained, otherwise “0” is obtained as the signal S205. Output to the circuit 211.

  The comparator 206 compares the image data M of the pixel with the threshold value THM. If the image data M is equal to or greater than the threshold value THM, “1” is set as the signal S206. Output to the circuit 211.

  The comparator 207 compares the image data Y of the pixel with the threshold value THY. If the image data Y is equal to or greater than the threshold value THY, “1” is set as the signal S207. Output to the circuit 211.

  The OR circuit 211 calculates the logical sum of the signals S205, S206, and S207 from the comparators 205 to 207, and outputs the result to the AND circuit 212 as a signal S211.

  The AND circuit 212 calculates a logical product of the signal S204 from the comparator 204 and the signal S211 from the OR circuit 211, and outputs the result to the selector 209 as a signal S212.

  The selector 209 uses the signal S212 from the AND circuit 212 as a control signal, and when the signal S212 is “1”, the signal S208 from the selector 208 is “01 (2 bits)”, otherwise. Outputs “10 (2 bits)” to the count circuit 123 and the black detection circuit 124 as the first color detection result COLOR which is the output of the first color detection circuit 125.

  With the above configuration, the first color detection circuit 125 has the minimum value feature value MIN, the maximum density difference feature value SUB, the density difference average value feature value AVE, the density difference total feature value PLUS, and the image data at the same coordinates in the specific area. C, M, and Y are compared with the threshold values set for each, and the first color detection result COLOR is output in three ways according to the combination. That is, when the first color detection result COLOR is “01”, it means “black character region (emphasis processing is required)”, and when it is “10”, “black character region (no enhancement processing is required)”. In this case, “00” means “an area other than a black character area”. That is, the first color detection result COLOR detects a black portion.

  As shown in FIG. 21, the second color detection circuit 126 includes two comparators 221, 222 and an AND circuit 223.

  The comparator 221 compares the maximum value feature amount MAX calculated by the maximum value calculation circuit 121a (FIG. 19) with the threshold value THMAX, and when the maximum value feature amount MAX is equal to or greater than the threshold value THMAX, In other cases, “0” is output to the AND circuit 223 as the signal S221.

  The comparator 222 compares the maximum density difference feature quantity SUB calculated by the maximum density difference calculation circuit 121c (FIG. 19) with the threshold value THSUB, and when the maximum density difference feature quantity SUB is equal to or greater than the threshold value THSUB, “1”. ", Otherwise," 0 "is output to the AND circuit 223 as the signal S222.

  The AND circuit 223 calculates the logical product of the signals S221 and S222 from the comparators 221 and 222, and outputs the result to the count circuit 123 as a second color detection result signal S126 that is an output of the second color detection circuit 126. To do.

  With the above configuration, the second color detection circuit 126 compares the maximum value feature amount MAX and the maximum density difference feature amount SUB at the same coordinates in the specific region with the threshold values set for each, and the second color The detection result signal S126 is output in two ways. That is, when the second color detection result signal S126 is “1”, it means “dark region”, and when it is “0”, it means “region other than the dark region”.

  Here, the optimum value of each threshold value set by the first color detection circuit 125 and the second color detection circuit 126 is shown below. All values are in decimal notation. In addition, the numerical range and the like can be set within 8 bits (0 to 255) except for the threshold value THPLUS which is a 10-bit signal. The setting reference varies depending on the state of the image processing apparatus (CCD or the like).

THMIN = 26, THPLUS = 73, THSUB = 25, THAVE = 88, TH (THC, THM, THY) = 25, THMAX = 60
The count circuit 123 (FIG. 19) counts the total number of the first color detection result COLOR and the second color detection result signal S126 in the specific area, and outputs the count result COUNT to the black color detection circuit 124. That is, the count result COUNT represents the number of colors other than black and black.

  The black detection circuit 124 (FIG. 19) includes a first color detection result COLOR from the first color detection circuit 125, a count result COUNT from the count circuit 123, and an edge determination result EDGE from the determination processing circuit 104 described later. Is input, and the black character discrimination result BLACK for the target pixel is output as the result of the second image area discrimination circuit 3b.

  As shown in FIG. 22, the black color detection circuit 124 includes three combinational circuits 231 to 233, four selectors 234 to 237, and a comparator 238.

  The combinational circuit 231 receives the edge discrimination result EDGE and the first color detection result COLOR transmitted in parallel in 2 bits, and inputs “1” when EDGE = 01 and COLOR = 01, and otherwise. Outputs “0” to the selector 234 as a 1-bit signal S231.

  The combinational circuit 232 receives the edge discrimination result EDGE and the first color detection result COLOR transmitted in 2-bit parallel, and inputs “1” when EDGE = 10 and COLOR = 01, and otherwise. Outputs “0” to the selector 235 as a 1-bit signal S232.

  The combination circuit 233 receives “EDGE = 01 and COLOR = 10” or “EDGE = 10 and COLOR = 10” with the edge determination result EDGE and the first color detection result COLOR transmitted in parallel in 2 bits as inputs. “1” is output to the selector 236 as a 1-bit signal S233.

  The selector 234 uses the signal S231 from the combinational circuit 231 as a control signal, and “01 (2 bits)” when the signal S231 is “1”, and “00 (2 bits)” otherwise. Is output to the selector 235 as a signal S234.

  The selector 235 uses the signal S232 from the combinational circuit 232 as a control signal, and when the signal S232 is “1”, it is “10 (2 bits)”, otherwise it is the signal S234 from the selector 234. A certain “01” or “00” is output to the selector 236 as a signal S235.

  The selector 236 uses the signal S233 from the combinational circuit 233 as a control signal. When the signal S233 is “1”, the selector 236 uses “10 (2 bits)”; otherwise, the selector 236 uses the signal S235 from the selector 235. Any one of “10”, “01”, and “00” is output to the selector 237 as a signal S236.

  The comparator 238 compares the count result COUNT with the threshold value THCOUNT. When the count result COUNT is equal to or greater than the threshold value THCOUNT, “1” is set, and “0” is set otherwise, and the selector 237 is set as the signal S238. Output to.

  The selector 237 uses the signal S238 from the comparator 238 as a control signal, and when the signal S238 is “1”, “00 (2 bits)”; otherwise, the selector 237 uses the signal S236 from the selector 236. Any one of “10”, “01”, and “00” is output as the black character discrimination result BLACK which is the output of the black color detection circuit 124.

  With the above configuration, the black detection circuit 124 detects the edge determination result EDGE that detects the location where the color change suddenly occurs, the first color detection result COLOR that detects the black portion, and the colors other than black and black. Detection of black characters, particularly detection of edge portions of black characters, is performed in combination with the count result COUNT obtained by counting the number.

  Here, the threshold value THCOUNT set by the comparator 238 is preferably a numerical value in the range of 10 to 20, and the optimum value is 14 (decimal notation). In addition, about the range of a numerical value etc., it can set within 8 bits (0-255). The setting reference varies depending on the state of the image processing apparatus (CCD or the like).

  The black character discrimination result BLACK is input from the second image region discrimination circuit 3b to a post-processing (filtering) program in the image processing unit 2, and edge enhancement processing is performed on the image.

  As shown in FIG. 18, the edge determination circuit 103 receives image data C, M, and Y of 7 colors and 8 bits from each line memory 101 and detects whether or not the pixel of interest is an edge, which will be described later. The edge detection signal S134 and the edge determination signal S136 to be output are output to the determination processing circuit 104. Then, the determination processing circuit 104 determines the edge determination result EDGE based on the input edge detection signal S134 and the edge determination signal S136, and outputs it to the color determination circuit 102.

  As shown in FIG. 23, the edge discrimination circuit 103 includes edge discrimination pre-processing circuits 130C, 130M, and 130Y provided for each color, a first LUT (look up table) 133, a second edge detection circuit 134, and the like. The second edge threshold selection circuit (edge threshold selection means) 135 and the edge feature quantity extraction circuit (edge feature quantity extraction means) 136 are provided.

  Since the edge determination preprocessing circuits 130C, 130M, and 130Y have the same configuration, the edge determination preprocessing circuit 130C will be described below as an example.

  The edge discrimination pre-processing circuit 130C includes a first edge detection circuit 131C, and receives 7-line 8-bit image data C from the line memory 101C as an input and a mask filter MF (described later) for edge extraction. By performing a convolution operation, an edge detection result signal S131C of the image data C is obtained and output to the first LUT 133. Similarly, the edge discrimination preprocessing circuits 130M and 130Y include first edge detection circuits 131M and 131Y.

  Further, the edge discrimination pre-processing circuit 130C includes a first edge threshold selection circuit (edge threshold selection means) 132C, and receives the 7-line 8-bit image data C from the line memory 101C as the input, and focuses on the target pixel. The binarized image data C of the area to be binarized and outputs the binarized image data S132C to the second edge threshold selection circuit 135. Similarly, the edge discrimination preprocessing circuits 130M and 130Y include first edge threshold selection circuits (edge threshold selection means) 132M and 132Y.

  As shown in FIG. 24, the first edge detection circuit 131C receives 7 lines and 8 bits of image data C from the line memory 101C, and inputs 9 pixels centered on the pixel of interest P0 (shaded portion) in the specific area. A mask filter MF for edge extraction is convolved with the (3 pixels × 3 pixels) image data C, and the comparison result between the obtained value and the threshold value THCONVC is used as an edge detection result signal S131C of the image data C. Output to the LUT 133.

  Here, the first edge detection circuit 131C includes two arithmetic systems 317h and 317v, an adder 315, and a comparator 316.

  The arithmetic system 317h includes three subtracters 301 to 303, three integrators 304 to 306, and an adder 307. The arithmetic system 317v includes three subtracters 308 to 310, three integrators 311 to 313, and an adder 314. Note that the calculation system 317h and the calculation system 317v have substantially the same configuration, and only the direction of pixel combination in the calculation process is different.

  Specifically, the arithmetic system 317h first multiplies the image data C by the pixel value of the pixel P and the filter coefficient W of the mask filter MF (FIG. 31A) for each corresponding position in the matrix. Subsequently, the subtracter 301 subtracts the value of (m−1, n−1) and the value of (m + 1, n−1), and the subtractor 302 calculates the value of (m−1, n + 1). The result of subtracting the value of (m + 1, n + 1), the value of (m−1, n) and the value of (m + 1, n) each multiplied by 2 in the integrators 304 and 305 are sent to the subtractor 303. The adder 307 adds the results of the subtraction. The accumulator 306 squares the result of the adder 307 and outputs the result to the adder 315.

  Similarly, the arithmetic system 317v first multiplies the image data C by the pixel value of the pixel P and the filter coefficient W of the mask filter MF (FIG. 31A) for each corresponding position in the matrix. Subsequently, the subtracter 308 subtracts the value of (m−1, n−1) and the value of (m−1, n + 1), and the subtractor 309 calculates the value of (m + 1, n−1). The result of subtracting the value of (m + 1, n + 1), the value of (m, n−1) and the value of (m, n + 1) multiplied by 2 respectively in the integrators 311, 312 are sent to the subtractor 310. The result of subtraction is added by the adder 314. The accumulator 313 squares the result of the adder 314 and outputs it to the adder 315.

  The adder 315 calculates the final result of the convolution operation of the mask filter MF by adding the results of the operation systems 317h and 317v, and outputs the result to the comparator 316 as the convolution operation result signal S315.

  As described above, the mask filter MF used in the present embodiment is a mask filter in which filter coefficients are arranged in a 3 × 3 matrix. For example, a mask filter having filter coefficients as shown in FIGS. 32A to 32D can be used as the mask filter MF. 32A to 32D are mask filters for edge detection in the 0 ° direction, 45 ° direction, 90 ° direction, and 135 ° direction, respectively.

  The comparator 316 compares the convolution calculation result signal S315 calculated by the adder 315 with the threshold value THCONVC, and sets “1” when the convolution calculation result signal S315 is greater than the threshold value THCONVC, otherwise. Outputs “0” to the first LUT 133 as the edge detection result signal S131C of the image data C.

  Here, the threshold value THCONVC has a setting range of 21 bits. The threshold value THCONVC is preferably a numerical value in the range of 20000 to 25000, and the optimum value is 24025 (decimal notation). However, the threshold THCONVC setting criterion varies depending on the state of the CCD sensor 1 of the image processing apparatus.

  The first LUT 133 receives the edge detection result signals S131C, S131M, and S131Y obtained from the image data C, M, and Y for each color in the first edge detection circuits 131C, 131M, and 131Y. A value (for example, “0” or “1”) is extracted from a numerical table preliminarily set and stored in a RAM or the like, and is output to the second edge detection circuit 134 as an edge information signal S133.

  As a result, the first LUT 133 receives “1” when two or more of the edge detection result signals S131C, S131M, and S131Y are “1”, and “0” otherwise. Are output to the second edge detection circuit 134.

  The second edge detection circuit 134 (FIG. 23) receives the edge information signal S133 of the pixel of interest P0 and its surrounding pixels obtained by the first LUT 133 as an input, and determines the edge determination result EDGE of the pixel of interest P0 by combining them. Obtained and output to the determination processing circuit 104 as the edge detection signal S134.

  As shown in FIG. 25, the second edge detection circuit 134 includes two adders 321, 322, a maximum value calculator 323, a comparator 324, an OR circuit 325, and a second LUT 326. Has been.

  In the calculation in the second edge detection circuit 134, first, as shown in FIG. 30A, a cross-shaped region ACONV including nine pixels centered on the target pixel P0 is set for the target pixel P0 in the specific region. . That is, as shown in FIG. 31B, the cross-shaped region ACONV is centered on the pixel of interest P0 = P (i, j), and the pixels P (i−2, j), P (i−) on the jth row. 1, j), P (i + 1, j), P (i + 2, j), and pixels P (i, j-2), P (i, j-1), P (i, j + 1), i columns, P (i, j + 2).

  The edge detection signal S134 of the target pixel P0 is based on the edge information signal S133 obtained in the first LUT 133 for each of all nine pixels P included in the cross-shaped area ACONV centered on the target pixel P0. Desired.

  Specifically, the following calculation is performed on the edge information signal S133. The values of the pixels P (i−2, j), P (i−1, j), P (i, j), P (i + 1, j), and P (i + 2, j) on the j row are added to the adder 321. Of the pixels P (i, j-2), P (i, j-1), P (i, j), P (i, j + 1), and P (i, j + 2) on the i column. Of the results of adding the values by the adder 322, the larger value is obtained by the maximum value calculator 323. Then, the comparator 324 compares the result of the maximum value calculator 323 with the threshold value THCSMAX. If the result of the maximum value calculator 323 is smaller than the threshold value THCSMAX, “1” is set. Otherwise, “0” is set. Is output to the second LUT 326 as the signal X.

  Further, the OR circuit 325 calculates the logical sum of the values of all the pixels P included in the cross-shaped area ACONV, and outputs the result as a signal Y to the second LUT 326. Further, the value of the target pixel P0 is output to the second LUT 326 as a signal Z.

  The second LUT 326 extracts a value (for example, “0” or “1”) corresponding to the combination of the input signals X, Y, and Z from a numerical table preliminarily set and stored in a RAM or the like, and an edge determination circuit 103 is output to the determination processing circuit 104 as an edge detection signal S134 as a result of 103.

  As described above, whether or not the pixel of interest P0 is a pixel in which the color change is abruptly generated by the first edge detection circuits 131C, 131M, and 131Y, the first LUT 133, and the second edge detection circuit 134 is determined. Can be detected and output as an edge detection signal S134.

  That is, the cross-shaped area ACONV is set for the pixel of interest P0 in the specific area (FIG. 30A), and the 3 × 3 mask filter MF (FIG. 31) is set for all the pixels P included in the cross-shaped area ACONV. The edge information signal S133 of each pixel is obtained by performing a convolution operation with (a)). Specifically, in order to obtain the edge detection signal S134 of the pixel P (i, j), as shown in FIG. 30B, a filter that is the center of the pixel P (i, j-2) and the mask filter MF. A convolution operation is performed by setting the mask filter MF so that the coefficient W (m, n) corresponds to the edge information signal S133 of the pixel P (i, j-2). Then, based on the edge information signal S133 of all nine pixels in the cross-shaped region ACONV, the second LUT 326 is referred to extract the edge detection signal S134.

  The edge feature amount extraction circuit 136 (FIG. 23) receives the image data C, M, and Y for each color from the line memory 101 and inputs a 7 pixel × 7 pixel region (first area A1) centered on the target pixel. And the sum of the maximum value, the minimum value, the maximum density difference, and the standard deviation (only the first area A1) is obtained for the 7 pixel × 31 pixel region (second area A2), and this is used as edge information to be described later as a second edge By comparing with the edge threshold value selected by the main edge threshold value selection signal S135 calculated by the threshold value selection circuit 135 (FIG. 28), the edge determination signal S136 is obtained, and the determination processing circuit 104 is obtained as a result of the edge determination circuit 103. Output to.

  As shown in FIG. 26, the edge feature quantity extraction circuit 136 includes first calculation circuits 331C, 331M, 331Y and second calculation circuits 332C, 332M, 332Y provided for each color, a sum calculation circuit 336, and edge determination. And a circuit 337.

  The first calculation circuits 331C, 331M, 331Y and the second calculation circuits 332C, 332M, 332Y are provided in the same configuration for each of the image data C, M, Y. Therefore, only the first calculation circuit 331C and the second calculation circuit 332C that perform the calculation of the image data C will be described below.

  The first calculation circuit 331C receives the image data C from the line memory 101C as an input, and outputs a maximum of 7 pixels × 7 pixels first area A1 and 7 pixels × 31 pixels second area A2 centered on the target pixel. The value, minimum value, and maximum density difference are calculated and output to the sum calculation circuit 336 as a 32-bit first edge information signal S331C.

  The second calculation circuit 332C receives the image data C from the line memory 101C as an input, calculates a standard deviation for the first area A1 of 7 pixels × 7 pixels centered on the pixel of interest, and sends it to the total calculation circuit 336. The bit is output as the second edge information signal S332C.

  The sum calculation circuit 336 includes first edge information signals S331C, S331M, S331Y (maximum value, minimum value) calculated for each color by the first calculation circuits 331C, 331M, 331Y and the second calculation circuits 332C, 332M, 332Y, respectively. , Maximum density difference) and second edge information signals S332C, S332M, and S332Y (standard deviation) are obtained and output to the edge determination circuit 337.

  The total calculation circuit 336 includes a 7 × 31 maximum value total circuit 333a, a 7 × 31 minimum value total circuit 333b, a 7 × 31 maximum density difference total circuit 333c, a 7 × 7 maximum value total circuit 334a, A 7 × 7 minimum value summation circuit 334b, a 7 × 7 maximum density difference summation circuit 334c, and a 7 × 7 standard deviation summation circuit 335 are provided.

  The 7 × 31 maximum value summation circuit 333a, the 7 × 31 minimum value summation circuit 333b, and the 7 × 31 maximum density difference summation circuit 333c are respectively calculated for the image data C, M, and Y by the first calculation circuits 331C, 331M, and 331Y. The sum of the respective values is calculated by inputting the obtained maximum value, minimum value, and maximum density difference of the second area A2 of 7 pixels × 31 pixels centered on the target pixel, and the 7 × 31 maximum value total signal The p1, 7 × 31 minimum value sum signal p2, and the 7 × 31 maximum density difference sum signal p3 are output to the edge determination circuit 337.

  Similarly, the 7 × 7 maximum value summation circuit 334a, the 7 × 7 minimum value summation circuit 334b, and the 7 × 7 maximum density difference summation circuit 334c are imaged by the first calculation circuits 331C, 331M, and 331Y, respectively. The maximum value, minimum value, and maximum density difference of the first area A1 of 7 pixels × 7 pixels centered on the pixel of interest obtained for each Y are input, and the sum of each value is calculated to obtain the maximum of 7 × 7 The value sum signal p4, 7 × 7 minimum value sum signal p5, and 7 × 7 maximum density difference sum signal p6 are output to the edge determination circuit 337.

  The 7 × 7 standard deviation summation circuit 335 is a first area of 7 pixels × 7 pixels centered on the pixel of interest obtained by the second calculation circuits 332C, 332M, and 332Y for each of the image data C, M, and Y. The sum total is calculated by using the standard deviation of A1 as an input, and is output to the edge determination circuit 337 as a 7 × 7 standard deviation sum signal p7.

  The edge determination circuit 337 includes summation signals p1 to p7 from the summation calculation circuit 336, which is edge information, and a main edge threshold selection signal S135 calculated by a second edge threshold selection circuit 135 (FIG. 28) described later. Is input to each of the sum signals p1 to p7 and the edge threshold selected by the predetermined edge threshold and the main edge threshold selection signal S135 to obtain an edge determination signal S136, and the result of the edge determination circuit 103 is obtained. To the determination processing circuit 104.

  The edge determination circuit 337 includes a selection circuit 341, six comparators 342a, 342b, 343a, 343b, 344, 345, two AND circuits 346, 347, and an OR circuit 348. Yes.

  The selection circuit 341 calculates each of the maximum value, the minimum value, and the maximum density difference of the second area A2 of 7 pixels × 31 pixels centered on the pixel of interest, which is calculated by the total circuits 333a to 333c of the total calculation circuit 336. Sum signals p1 to p3, which are sums, and a main edge threshold selection signal S135 from a second edge threshold selection circuit 135 (FIG. 28), which will be described later, are input, and each sum signal p1 to p3 and main edge threshold selection signal S135 are input. Are compared with the edge threshold value selected in step S341, signals S341a, S341b, and S341c are obtained and output to the OR circuit 348.

  As shown in FIG. 27, the selection circuit 341 includes three selectors 351 to 353 and three comparators 354 to 356. Here, the main edge threshold selection signal S135 is a signal including selection signals S135e, S135f, and S135g, and is transmitted separately in the selection circuit 341.

  The selector 351 selects one of the three edge thresholds THMAX3A, THMAX3B, and THMAX3C, which are input signals, according to the combination of the selection signals S135e and S135g, and outputs the selected signal to the comparator 354 as a signal S351. The comparator 354 compares the 7 × 31 maximum value sum signal p1 from the 7 × 31 maximum value summation circuit 333a with the signal S351 (edge threshold values THMAX3A to 3C) from the selector 351, and the 7 × 31 maximum value summation. If the signal p1 is smaller than the signal S351, “0” is output to the OR circuit 348 as “0”, otherwise “1” is output as the signal S341a.

  The selector 352 selects one of the two edge thresholds THMIN3A and THMIN3B, which are input signals, according to the combination of the selection signals S135e and S135g, and outputs the selected signal to the comparator 355 as a signal S352. Then, the comparator 355 compares the 7 × 31 minimum value sum signal p2 from the 7 × 31 minimum value summation circuit 333b with the signal S352 (edge threshold value THMIN3A, 3B) from the selector 352, thereby comparing the 7 × 31 minimum value. When the sum signal p2 is larger than the signal S352, “0” is output to the OR circuit 348 as “0”, and “1” is output as the signal S341b.

  The selector 353 selects one of the three edge thresholds THMDIFA, THMDIFB, and THMDIFC that are input signals according to the combination of the selection signals S135e, S135f, and S135g, and outputs the selected signal to the comparator 356 as a signal S353. Then, the comparator 356 compares the 7 × 31 maximum density difference sum signal p3 from the 7 × 31 maximum density difference summation circuit 333c with the signal S353 (edge threshold value THMDIFA to C) from the selector 353, and the 7 × 31 maximum density difference summation signal p3. If the density difference sum signal p3 is smaller than the signal S353, “0” is output to the OR circuit 348 as “0”, and “1” is output as the signal S341c otherwise.

  Thus, the selection circuit 341 outputs signals S342a to S342c that are “0” when the pixel of interest is included in the character region in the 7 × 31 region, and “1” in other cases. Is done. Therefore, the edge thresholds THMAX3A to 3C are set to values for performing edge determination from the maximum density value of the 7 × 31 region. The edge thresholds THMIN3A and 3B are set to values for performing edge determination from the minimum density value of the 7 × 31 region. The edge threshold values THMDIFA to C are set to values for performing edge determination from the maximum density difference of the 7 × 31 region.

  The comparator 342a compares the 7 × 7 maximum value sum signal p4 from the 7 × 7 maximum value sum circuit 334a with a preset edge threshold value THMAX1, and the 7 × 7 maximum value sum signal p4 is compared with the edge threshold value THMAX1. If it is smaller, “0” is output to the AND circuit 347 as a signal S342a.

  The comparator 343a compares the 7 × 7 minimum value sum signal p5 from the 7 × 7 minimum value sum circuit 334b with a preset edge threshold value THMIN1, and the 7 × 7 minimum value sum signal p5 is compared with the edge threshold value THMIN1. If it is larger, “0” is output to the AND circuit 347 as a signal S343a.

  The AND circuit 347 receives the signals S342a and S343a from the comparators 342a and 343a, calculates a logical product of these signals, and outputs the logical product as a signal S347 to the OR circuit 348.

  Similarly, the comparator 342b compares the 7 × 7 maximum value sum signal p4 from the 7 × 7 maximum value sum circuit 334a with a preset edge threshold value THMAX2, and the 7 × 7 maximum value sum signal p4 is obtained. If it is smaller than the edge threshold value THMAX2, “0” is output to the AND circuit 346 as a signal S342b otherwise “1”.

  The comparator 343b compares the 7 × 7 minimum value sum signal p5 from the 7 × 7 minimum value sum circuit 334b with a preset edge threshold value THMIN2, and the 7 × 7 minimum value sum signal p5 is compared with the edge threshold value THMIN2. If it is larger, “0” is output to the AND circuit 346 as a signal S343b.

  The comparator 344 compares the 7 × 7 maximum density difference sum signal p6 from the 7 × 7 maximum density difference sum circuit 334c with a preset edge threshold value THMD to obtain a 7 × 7 maximum density difference sum signal p6. If it is smaller than the edge threshold THMD, “0” is output to the AND circuit 346 as a signal S344, and “1” is output otherwise.

  The comparator 345 compares the 7 × 7 standard deviation sum signal p7 from the 7 × 7 standard deviation sum circuit 335 with a preset edge threshold value THSD, and the 7 × 7 standard deviation sum signal p7 is compared with the edge threshold value THSD. If it is smaller, “0” is output to the AND circuit 346 as a signal S345.

  The AND circuit 346 receives the signals S342b, S343b, S344, and S345 from the comparators 342b, 343b, 344, and 345, calculates a logical product of these signals, and outputs the logical product as a signal S346 to the OR circuit 348.

  As a result, the AND circuits 346 and 347 indicate that the signals S346 and 347 are “0” when the pixel of interest is included in the character region in the 7 × 7 region and “1” otherwise. Is output. Therefore, the edge thresholds THMAX1 and THMAX2 are set to values for performing edge determination from the maximum density value of the 7 × 7 region. The edge threshold values THMIN 1 and 2 are set to values for performing edge determination from the minimum density value of the 7 × 7 area. The edge threshold value THMD is set to a value for performing edge determination from the maximum density difference of the 7 × 7 region. The edge threshold value THSD is set to a value for performing edge determination from the standard deviation value of the 7 × 7 area.

  The OR circuit 348 receives the signals S341a to S341c from the selection circuit 341 and the signals S346 and S347 from the AND circuits 346 and 347, calculates a logical sum of them, and outputs the result of the edge feature quantity extraction circuit 136. Is output to the determination processing circuit 104 as an edge determination signal S136.

  Thus, the OR circuit 348 can output to the determination processing circuit 104 a 1-bit signal that is “0” when the target pixel is a character region and “1” when the target pixel is a non-character region.

  Here, the optimum value of each threshold set by the edge determination circuit 337 is shown below. All values are in decimal notation. As for the range of numerical values, only THSD is 16 bits, and all others can be set within 10 bits (0 to 1023). The set value varies depending on the state of the image processing apparatus (CCD or the like).

THMIN1 = 330, THMIN2 = 45, THMAX1 = 432, THMAX2 = 339, THMD = 330, THSD = 1800,
THMAX3A = 147, THMAX3B = 210, THMAX3C = 420, THMIN3A = 330, THMIN3B = 135,
THMDIFA = 135, THMDIFB = 210, THMDIFC = 225
Next, as shown in FIG. 23, the main edge threshold value selection signal S135 to the edge feature quantity extraction circuit 136 includes first edge threshold value selection circuits 132C, 132M, and 132Y of the edge discrimination preprocessing circuits 130C, 130M, and 130Y, Calculated from the second edge threshold selection circuit 135.

  The first edge threshold selection circuits 132C, 132M, and 132Y have the same configuration and are provided in the edge determination pre-processing circuits 130C, 130M, and 130Y, respectively. Therefore, only the first edge threshold selection circuit 132C that performs the calculation of the image data C will be described below.

  The first edge threshold selection circuit 132C receives the 7-line 8-bit image data C from the line memory 101C and receives the image data C in the 7 pixel × 7 pixel region (first area A1) centered on the target pixel. Using the average value as the binarization threshold, the image data C in the 7 pixel × 31 pixel region (second area A2) centered on the target pixel is binarized.

  As shown in FIG. 28, the first edge threshold selection circuit 132C includes a binarization threshold calculation circuit 361C, a binarization circuit 362C, and a 7 × 31 inversion count calculation circuit 363C.

  The binarization threshold value calculation circuit 361C receives the image data C from the line memory 101C, obtains an average value of the image data C in the first area A1, and supplies the binarization threshold value signal S361C to the binarization circuit 362C. Output.

  The binarization circuit 362C receives the image data C from the line memory 101C, compares the image data C in the second area A2 with the binarization threshold signal S361C from the binarization threshold calculation circuit 361C, If the image data C in the second area A2 is smaller than the binarization threshold signal S361C, it is binarized to “1”, otherwise it is binarized to “0”, and 7 × 31 inversion as the binarization signal S362C Output to the number calculation circuit 363C.

  The 7 × 31 inversion number calculating circuit 363C receives the binarized signal S362C from the binarizing circuit 362C and inputs the main scanning and sub scanning lines of the binarized image data C in the second area A2. Each time, the number of inversions of “0” and “1” is calculated and output to the second edge threshold selection circuit 135 as an inversion number signal S132C which is a result output of the first edge threshold selection circuit 132C.

  The second edge threshold selection circuit 135 receives the inversion frequency signals S132C, S132M, and S132Y from the first edge threshold selection circuits 132C, 132M, and 132Y, calculates a main edge threshold selection signal S135, and generates an edge feature quantity extraction circuit. To 136.

  As shown in FIG. 28, the second edge threshold selection circuit 135 is configured to include a monochrome pixel maximum number calculation circuit 371 and a threshold selection circuit 372.

  The black-and-white pixel maximum number calculation circuit 371 determines the number of inversions of the image data binarized for each color in the second area A2 from the first edge threshold selection circuits 132C, 132M, and 132Y. The maximum number of white pixels and the maximum number of black pixels for each color after binarization in the second area A2 are obtained by using the inversion number signals S132C, S132M, and S132Y calculated for each line, respectively. The pixel number maximum value signal S371w and the black pixel number maximum value signal S371b are output to the threshold selection circuit 372.

  The threshold selection circuit 372 receives the white pixel maximum value signal S371w and the black pixel maximum value signal S371b from the monochrome pixel maximum number calculation circuit 371 as input, and based on a preset threshold THSEL, the main edge threshold selection signal S135. Is output to the edge feature quantity extraction circuit 136 as a result output of the second edge threshold selection circuit 135. The main edge threshold selection signal S135 is a signal including three selection signals S135e, S135f, and S135g. The threshold value THSEL includes four threshold values (TP1, TP30, TP31, TME).

  The specific processing contents of the threshold selection circuit 372 are as follows. When the number of inversions of the 5 × 5 area is larger than the threshold value TP1, the comparison result is “1” (otherwise “0”), and the average value of the 7 × 7 area is compared with the threshold value TME. If it is larger, a logical product with the comparison result “1” (otherwise “0”) is calculated and output as the selection signal S135e. The product of the number of inversions in the 31 × 7 area and the threshold value TP31 is compared with the maximum value of the number of white pixels in the 31 × 7 area of CMY (white pixel number maximum value signal S371w). Otherwise, “0”) is output as the selection signal S135f. The product of the number of inversions of the 31 × 7 area and the threshold value TP30 is compared with the maximum value of black pixels in the 31 × 7 area of CMY (black pixel number maximum value signal S371b). Otherwise, “0”) is output as the selection signal S135g.

  As a result, the threshold selection circuit 372 outputs a main edge threshold selection signal S135 for selecting thresholds (THMAX3A to 3C, THMIN3A, 3B, and THMDIFA to C) in the selection circuit 341 according to the combination of inputs. The Therefore, the threshold value THSEL (TP1, TP30, TP31, TME) is set to a value that can determine an optimum threshold value for edge determination from the C, M, and Y image regions.

  Specifically, the threshold value TP1 can be set within 4 bits, a numerical value in the range of 1 to 10 is preferable, and the optimum value is 4. The threshold value TP30 can be set within 8 bits, a numerical value in the range of 30 to 70 is preferable, and an optimal value is 40. The threshold value TP31 can be set within 8 bits, a numerical value in the range of 30 to 70 is preferable, and the optimum value is 50. The threshold value TME can be set within 8 bits, is preferably a numerical value in the range of 20 to 50, and the optimum value is 38. All the above numerical values are in decimal notation. Each set value varies depending on the state of the image processing apparatus (CCD or the like).

  As described above, whether or not the target pixel (center pixel) belongs to the character region is determined by the first edge threshold selection circuits 132C, 132M, and 132Y, the second edge threshold selection circuit 135, and the edge feature amount extraction circuit 136. It is possible to output to the determination processing circuit 104 a 1-bit edge determination signal S136 that is “0” when the target pixel is a character region and “1” when the target pixel is a non-character region.

  The first area A1 may be a narrow range centered on the target pixel. On the other hand, the second area A2 only needs to be wider than the first area A1. However, the 7 pixel × 7 pixel region centered on the target pixel is optimal for the first area A1, and the 7 pixel × 31 pixel region centered on the target pixel is optimal for the second area A2, respectively. This is an important area.

  As shown in FIG. 18, the determination processing circuit 104 receives the edge detection signal S134 and the edge determination signal S136 from the edge determination circuit 103 as input, and whether or not the color changes abruptly at the target pixel due to the combination thereof. The edge determination result EDGE indicating this is output to the color determination circuit 102. The color determination circuit 102 performs integration processing including the edge determination result EDGE for black detection.

  As shown in FIG. 29, the determination processing circuit 104 includes an enhancement switching circuit 400 and a third LUT 420.

  The emphasis switching circuit 400 is a circuit that determines whether or not the emphasis processing of the target pixel P0 is necessary in a 5 × 5 area centered on the target pixel P0 (i, j). Specifically, the emphasis switching circuit 400 includes emphasis switching pre-processing circuits 411C, 411M, and 411Y provided in C, M, and Y, two AND circuits 407 and 409, and an OR circuit 410, respectively. It is configured.

  Since the emphasis switching pre-processing circuits 411C, 411M, and 411Y have the same configuration, the emphasis switching pre-processing circuit 411C will be described below as an example.

  The emphasis switching pre-processing circuit 411C includes four subtracters 401 to 404, an adder 405, and two comparators 406 and 408.

  The subtractor 401 calculates an absolute difference value between the pixel of interest P0 and the pixel P (i-2, j-2) in a 5 × 5 area centered on the pixel of interest P0 (i, j) for each cyan table. Calculate and output to the adder 405 as a signal S401. Similarly, the subtracters 402, 403, and 404 are the target pixel P0 and the pixel P (i + 2, j-2), the target pixel P0 and the pixel P (i + 2, j + 2), and the target pixel P0 and the pixel P (i-2, j + 2). ) And outputs the signals to the adder 405 as signals S402, S403, and S404, respectively.

  The adder 405 receives the signals S401 to S404 from the subtractors 401 to 404, calculates the sum of these signals, and outputs the sum as a signal S405 to the comparator 406.

  The comparator 406 compares the signal S405, which is the sum of the absolute difference values centered on the pixel of interest P0, with the threshold value THET. If the signal S405 is smaller than the threshold value THET, "1" is set otherwise. Is output to the AND circuit 407 as a signal S406C (1 bit) as a result of the emphasis switching pre-processing circuit 411C. Note that the threshold value THET is desirably a value in the range of 10 to 50, and the optimum value is 15 (all decimal values are described above).

  Further, the comparator 408 compares the target pixel P0 with 0 which is a threshold value. When the target pixel P0 is smaller, “1” is displayed, and “0” is displayed otherwise. The signal is output to the AND circuit 409 as a signal S408C (1 bit) as a result of the processing circuit 411C.

  In this way, two signals are output from the emphasis switching pre-processing circuits 411C, 411M, and 411Y provided for C, M, and Y, respectively. That is, signals S406C, S406M, and S406Y are output to AND circuit 407. Also, signals S408C, S408M, and S408Y are output to AND circuit 409.

  Subsequently, the AND circuit 407 calculates a logical product of the input signals S406C, S406M, and S406Y, and outputs the logical product to the OR circuit 410 as a signal S407 (1 bit).

  On the other hand, the AND circuit 409 calculates the logical product of the input signals S408C, S408M, and S408Y, and outputs the logical product to the OR circuit 410 as a signal S409 (1 bit).

  The OR circuit 410 calculates the logical sum of the signals S407 and S409 input from the AND circuits 407 and 409, and outputs the logical sum to the third LUT 420 as the enhancement switching signal S400 that is the result of the enhancement switching circuit 400.

  The third LUT 420 receives the emphasis switching signal S400 from the emphasis switching circuit 400, the edge detection signal S134 and the edge determination signal S136 from the edge discrimination circuit 103 (FIG. 18), and inputs a value (2 Bits “00”, “01”, and “10”) are extracted from a numerical table preliminarily set and stored in a RAM or the like and output to the color determination circuit 102 as an edge determination result EDGE. When the edge discrimination result EDGE is “01”, it means “edge region (emphasis processing is required)”, and when it is “10”, it means “edge region (emphasis processing is not required)” and “00”. Sometimes it means “an area other than an edge area”.

  As described above, in the image processing apparatus according to the present embodiment, the area to which each pixel on the image belongs is a black character area from the C, M, and Y multivalued image data obtained by scanning the document. When determining whether or not there is a target pixel, the target pixel belongs based on all image data in a specific area including the target pixel to be determined and a plurality of neighboring pixels located around the target pixel. A feature amount indicating the characteristics of the region is extracted, and the region to which the target pixel belongs is determined based on each feature amount.

  As a result, the image processing apparatus obtains a feature amount from a specific area composed of neighboring pixels adjacent to the target pixel with respect to C, M, and Y multi-valued image data obtained by scanning a document with a CCD sensor or the like. Based on the result of the output as an identification signal, it is possible to perform optimal image processing for the region to which the pixel of interest belongs on the image, and high image quality can be achieved.

  The image processing apparatus includes a color determination circuit that determines whether each pixel on the image is black or not from multi-value image data of C, M, and Y obtained by scanning a document.

  As a result, the image processing apparatus determines whether or not the pixel of interest on the image is black in the color determination circuit with respect to the input multivalued image data of C, M, and Y, so that the optimum image as a black area Processing becomes possible, and higher image quality can be achieved.

  In the image processing apparatus, from C, M, and Y multivalued image data obtained by scanning a document, an area to which each pixel on the image belongs is a black character edge region (an edge portion of a black character, a black line image, or the like). An edge discriminating circuit for discriminating whether it belongs to or not.

  Thus, the image processing apparatus determines whether or not the pixel of interest on the image belongs to the black character edge region in the edge determination circuit with respect to the input multivalued image data of C, M, and Y, and the black character, Optimum image processing is possible as an edge portion of a black line image or the like, and further image quality can be achieved.

  Further, in the image processing apparatus, the color determination circuit determines the maximum value, the minimum value, and the maximum density difference of density levels in all pixels within a specific area centered on the target pixel in each of the C, M, and Y image data. Each calculation circuit to be calculated; a density average value calculation circuit for calculating an average value of density levels in all pixels in the specific area; and a sum of absolute values of density level differences between adjacent pixels in the main scanning direction in the specific area; A color feature amount extraction circuit for extracting a color feature amount of an area to which the pixel of interest belongs from a result of a density difference sum calculation circuit for obtaining a sum of absolute values of density level differences between adjacent pixels in the sub-scanning direction; The color determination result is obtained from the color feature amount. Thereby, the image processing apparatus can further improve the black discrimination accuracy.

  In addition, the image processing apparatus includes a color detection circuit that outputs a color discrimination result of the pixel of interest with a predetermined threshold with respect to the result obtained from the color feature amount extraction circuit. Thereby, the image processing apparatus can perform black determination with higher accuracy.

  Further, the image processing apparatus includes an edge detection circuit in which the edge determination circuit performs edge detection by convolving a filter with respect to each pixel in a specific area in each of the C, M, and Y image data, and the target pixel. Calculation circuits for obtaining the maximum value, minimum value and comparison difference value of density levels in all pixels in one or a plurality of specific areas, and the average value and each of density levels in all pixels in the specific area A standard deviation calculating circuit for obtaining a standard deviation from image data and an edge discriminating circuit for obtaining an edge discriminating result from a threshold obtained by a binarizing circuit are provided.

  Accordingly, the image processing apparatus can obtain a black character edge discrimination result from each calculation result and the color determination result. Therefore, it is possible to extract a black character edge portion with high accuracy, and it is possible to perform an optimal image processing as a black character edge area, and it is possible to achieve further higher image quality.

  Further, the image processing apparatus binarizes the values of pixels in one or more specific regions centered on the pixel of interest with a predetermined threshold in the binarization circuit, and performs each main scanning and sub-scanning each line. Is provided with an inversion number calculating circuit for obtaining the inversion number of the binary signal. Thereby, the image processing apparatus can perform edge part detection with higher accuracy.

  In the binarization circuit, the image processing apparatus includes a maximum white pixel number calculation circuit that calculates a maximum value of the number of white pixels after binarization, and a maximum that calculates a maximum value of the number of black pixels after binarization. And a black pixel number calculation circuit. As a result, the image processing apparatus can improve detection accuracy such as preventing erroneous recognition of the edge portion.

[Embodiment 3]
The first and second embodiments do not limit the scope of the present invention, and various modifications are possible within the scope of the present invention. For example, the first and second embodiments can be configured as follows.

  The present invention can be applied to a system composed of a plurality of devices (for example, a host computer, an interface device, a reader, a printer, etc.) or an apparatus composed of a single device (for example, a copier, a facsimile machine, etc.). May be.

  Another object of the present invention is to supply a recording medium in which a program code of software that realizes the functions of the first and second embodiments described above is readable by a computer, to the system or apparatus, and the computer of the system or apparatus This can also be achieved by (or CPU or MPU) reading and executing the program code stored in the storage medium. In this case, the program code itself read from the storage medium realizes the functions of the above-described embodiment, and the storage medium storing the program code constitutes the present invention. As a storage medium for supplying the program code, for example, a floppy disk, hard disk, magnetic tape, optical disk, magneto-optical disk, CD-ROM, CD-R, MD, nonvolatile memory card, ROM, RAM, or the like is used. be able to.

  Further, by executing the program code read out by the computer, not only the functions of the above-described embodiments are realized, but also the OS running on the computer based on the instruction of the program code performs the actual processing. In some cases, the functions 1 and 2 of the above-described embodiment are realized by performing part or all of the above.

  Further, after the program code read from the storage medium is written into a memory provided in a function expansion board inserted into the computer or a function expansion unit connected to the computer, the function expansion is performed based on the instruction of the program code. This includes a case where the CPU or the like provided in the board or the function expansion unit performs part or all of the actual processing, and the functions of the first and second embodiments described above are realized by the processing.

  Further, in order to solve the above-described problem, the image processing apparatus of the present invention relates to the density value of a pixel in a specific region including a target pixel in image data for each color component and a pixel in the vicinity of the target pixel. The density average value of the pixels in the specific area, the density difference sum that is the sum of absolute differences between the target pixel and the peripheral pixels extracted in the specific area, and the same density value as the target pixel It is provided with a maximum / minimum pixel calculation means for determining whether the density value of the target pixel is a maximum value or a minimum value by obtaining the number of density matching pixels which is the number of the peripheral pixels. It is said.

  With the above configuration, the average / minimum pixel calculation means extracts the average density value, the total density difference, and the number of density matching pixels from the specified area consisting of the target pixel and neighboring pixels for the input image data. Thus, it can be determined whether or not the density value of the target pixel is a maximum value or a minimum value.

  Therefore, it is possible to determine with high accuracy whether the region to which the target pixel belongs in the image is a character region, a halftone dot region, or a photo region. Therefore, it is possible to perform image processing optimal for the region to which the target pixel belongs on the image, and achieve high image quality.

  Further, in order to solve the above problems, the image processing apparatus of the present invention provides a maximum value or a minimum value of the density value within the specific region obtained by the maximum / minimum pixel calculation means in addition to the above configuration. A run length in which a pixel having a density value equal to or higher than the reference value based on the average density value or lower than the reference value from the target pixel having a value is a maximum value of the number of pixels that are continuous in the main scanning direction or the sub-scanning direction; There may be provided local maximum / minimum pixel peripheral information detection means for calculating a degree of complexity that is a sum of absolute differences between each pixel having a run length and the target pixel.

  With the above configuration, the run length and the degree of complexity can be extracted by the local maximum / minimum pixel peripheral information detection unit based on the local maximum / minimum pixel of the image data for each color component.

  Therefore, it can be determined with high accuracy whether the region to which the target pixel belongs in the image is a halftone dot region or a region other than a halftone dot such as a continuous dot or line. Therefore, it is possible to perform optimal image processing for the region to which the target pixel belongs on the image, and to achieve higher image quality.

  Further, in order to solve the above problems, the image processing apparatus of the present invention, in addition to the above-described configuration, includes the above-mentioned specific information obtained by the above-mentioned maximum / minimum pixel calculation means and maximum / minimum pixel peripheral information detection means. A pixel weight switching signal calculating means for extracting a pixel weight switching signal for each color component based on the run length and the complexity of the pixel of interest having a maximum or minimum density value in the region; Periodicity calculation means for detecting the periodicity of the halftone dots of the image data by counting pixels in the predetermined calculation area that are equal to or greater than the reference value based on the reference value may be provided.

  With the above configuration, the pixel weight switching signal calculation means extracts the pixel weight switching signal for each image data of each color component from the run length and the complexity, and counts the weight ratio within a certain area, thereby obtaining an image. The feature amount of the data can be extracted, and characters, halftone dots, and photo areas can be discriminated. Furthermore, the periodicity calculation means can extract the periodicity for each image data of each color component of the pixel weight switching signal.

  Therefore, it can be determined with high accuracy whether the region to which the pixel of interest belongs in the image is a halftone dot region. Therefore, it is possible to perform optimal image processing for the region to which the target pixel belongs on the image, and to achieve higher image quality.

  Further, in order to solve the above-described problem, the image processing method of the present invention relates to the density value of a pixel in a specific area including a target pixel in image data for each color component and a pixel in the vicinity of the target pixel. The density average value of the pixels in the specific area, the density difference sum that is the sum of absolute differences between the target pixel and the peripheral pixels extracted in the specific area, and the same density value as the target pixel A maximum / minimum pixel calculating step for determining whether or not the density value of the target pixel is a maximum value or a minimum value by obtaining a density matching pixel number that is the number of the peripheral pixels, and the maximum / minimum pixel From the target pixel having the maximum value or the minimum value of the density value in the specific region obtained by the calculation step, a pixel having a density value greater than or less than the reference value based on the average density value is changed in the main scanning direction. Or sub-scanning direction Maximum / minimum pixel peripheral information detection step for calculating the run length that is the maximum value of the number of consecutive pixels, and the degree of complexity that is the sum of absolute differences between each pixel that forms the run length and the target pixel, and Each color component based on the run length and complexity of the pixel of interest having a maximum or minimum density value in the specific area obtained by the local maximum / minimum pixel calculation step and the local maximum / minimum pixel peripheral information detection step The pixel weight switching signal is extracted for each pixel, and the periodicity of the halftone dots of the image data is detected by counting pixels that are greater than or less than the reference value based on the pixel weight switching signal within a predetermined calculation area. And a periodicity calculating step.

  With the above configuration, in the maximum / minimum pixel calculation process, for the input image data, the density average value, the density difference sum, and the number of density matching pixels are extracted from the specific area consisting of the target pixel and neighboring pixels adjacent to it. Thus, it can be determined whether or not the density value of the target pixel is a maximum value or a minimum value. In the local maximum / minimum pixel peripheral information detection step, the run length and the degree of complexity can be extracted based on the local maximum / minimum pixels of the image data for each color component. In the pixel weight switching signal calculation step, the pixel weight switching signal is extracted for each image data of each color component from the run length and the degree of complexity, and the weight ratio is counted within a certain area, thereby calculating the feature amount of the image data. Extraction can be performed to distinguish characters, halftone dots, and photo areas. Furthermore, in the periodicity calculation step, the periodicity for each image data of each color component of the pixel weight switching signal can be extracted.

  Therefore, it is possible to determine with high accuracy whether the region to which the target pixel belongs in the image is a character region, a halftone dot region, or a photo region. Therefore, it is possible to perform image processing optimal for the region to which the target pixel belongs on the image, and achieve high image quality. In addition, since the parameters used for discriminating the image area can be selected according to the image data, it is possible to discriminate characters, halftone dots, and photographic areas efficiently and quickly from various image data.

FIG. 7 is a block diagram illustrating an outline of a configuration of a first image region determination circuit provided in the image processing apparatus illustrated in FIG. 6. FIG. 2 is a block diagram showing an outline of a configuration of a first inspection circuit of the first image region discrimination circuit shown in FIG. 1, and FIGS. 2A, 2B, and 2C correspond to cyan, magenta, and yellow, respectively. FIG. 3 is a block diagram showing an outline of a configuration of a second inspection circuit of the first image area discrimination circuit shown in FIG. 1, and FIGS. 3A, 3B, and 3C correspond to cyan, magenta, and yellow, respectively. FIG. 2 is a block diagram illustrating an outline of a configuration of a feature amount extraction circuit of the first image region determination circuit illustrated in FIG. 1. FIG. 2 is a block diagram showing an outline of a configuration of a correction circuit of the first image region discrimination circuit shown in FIG. 1. It is a block diagram which shows the outline of a structure of the image processing apparatus concerning embodiment of this invention. It is explanatory drawing which shows the specific area range used with the 1st image area discrimination circuit shown in FIG. FIG. 4 is a matrix used for maximum / minimum pixel detection performed by the first image region discrimination circuit shown in FIG. 1, wherein (a) is a 3 × 3 matrix, and (b) is an explanatory diagram showing a 5 × 5 matrix. FIGS. 9A to 9G are explanatory views showing the lengths and directions of run lengths in the main and sub-scanning directions in the run length detection performed by the first image region discrimination circuit shown in FIG. It is explanatory drawing which shows pixel weight distribution for every plane in the weight ratio setting performed by the 1st image area discrimination circuit shown in FIG. It is explanatory drawing which shows the calculation area of the feature-value used with the 1st image area discrimination circuit shown in FIG. It is explanatory drawing which shows the calculation area used for the periodicity detection performed with the 1st image area discrimination circuit shown in FIG. FIGS. 13A and 13B are explanatory diagrams showing calculation areas used for other periodicity detection performed by the first image region discrimination circuit shown in FIG. It is explanatory drawing which shows the calculation method of parameter RJ used for the periodicity detection performed with the 1st image area discrimination circuit shown in FIG. FIGS. 15A and 15B are explanatory diagrams showing calculation areas of the parameter RJ used for periodicity detection performed by the first image region discrimination circuit shown in FIG. FIGS. 16A to 16C are explanatory diagrams showing calculation areas of feature amounts used for halftone dot block correction performed by the first image area discrimination circuit shown in FIG. FIGS. 17A to 17C are explanatory views showing other calculation areas of feature amounts used for the halftone block correction performed by the first image region discrimination circuit shown in FIG. FIG. 7 is a block diagram illustrating an outline of a configuration of a second image region determination circuit provided in the image processing apparatus illustrated in FIG. 6. It is a block diagram which shows the outline of a structure of the color determination circuit of the 2nd image area determination circuit shown in FIG. It is a block diagram which shows the outline of a structure of the 1st color detection circuit of the color determination circuit shown in FIG. It is a block diagram which shows the outline of a structure of the 2nd color detection circuit of the color determination circuit shown in FIG. It is a block diagram which shows the outline of a structure of the black detection circuit of the color determination circuit shown in FIG. It is a block diagram which shows the outline of a structure of the edge discrimination circuit of the 2nd image area discrimination circuit shown in FIG. FIG. 24 is a block diagram showing an outline of a configuration of a first edge detection circuit and a first LUT of the edge determination circuit shown in FIG. 23. FIG. 24 is a block diagram illustrating an outline of a configuration of a second edge detection circuit of the edge determination circuit illustrated in FIG. 23. FIG. 24 is a block diagram showing an outline of a configuration of an edge feature quantity extraction circuit of the edge determination circuit shown in FIG. 23. FIG. 27 is a block diagram showing an outline of a configuration of a selection circuit provided in an edge determination circuit of the edge feature quantity extraction circuit shown in FIG. 26. FIG. 24 is a block diagram showing an outline of a configuration of a first edge threshold selection circuit and a second edge threshold selection circuit of the edge determination circuit shown in FIG. 23. It is a block diagram which shows the outline of a structure of the determination process circuit of the 2nd image area discrimination circuit shown in FIG. (A) is explanatory drawing of the cross-shaped area | region which is a processing range of the 2nd edge detection circuit shown in FIG. 25, (b) is explanatory drawing which shows convolution of the mask filter in the 1st edge detection circuit shown in FIG. It is. (A) is explanatory drawing of the mask filter used by the 1st edge detection circuit shown in FIG. 24, (b) is explanatory drawing of the cross-shaped area | region which is a process range of the 2nd edge detection circuit shown in FIG. . 25 is an example of filter coefficients of a mask filter used in the first edge detection circuit shown in FIG. 24, and (a) to (d) are edge detections in 0 ° direction, 45 ° direction, 90 ° direction, and 135 ° direction, respectively. It is a mask filter for.

Explanation of symbols

12 (12C, 12M, 12Y) Maximum / minimum pixel calculation circuit
(Maximum / minimum pixel calculation means, maximum / minimum pixel calculation process)
13 (13C, 13M, 13Y) Maximum / minimum pixel peripheral information detection circuit (maximum / minimum pixel peripheral information detection means, maximum / minimum pixel peripheral information detection step)
21 (21C, 21M, 21Y)
Pixel weight switching signal calculation circuit (pixel weight switching signal calculation means)
23 (23C, 23M, 23Y)
First periodicity detection circuit (periodicity calculating means, periodicity calculating step)
24 (24C, 24M, 24Y)
Second periodicity detection circuit (periodicity calculating means, periodicity calculating step)
AVE (AVEC, AVEM, AVEY) density average values C0 to C15, D0 to D3, E0 to E2 area (calculation area)
EQ (EQC, EQM, EQY) Number of density matching pixels P, P0 Pixel of interest Pij Peripheral pixel PBUSY (PBUSYC, PBUSYM, PBUSYY) Congestion level PRUN (PRUNC, PRUNM, PRUNY) Run length PS (PSC, PSM, PSY) Density difference Total WS (WSC, WSM, WSY) Pixel weight switching signal 103 Edge discrimination circuit (edge discrimination means, edge discrimination step)
121 color feature quantity extraction circuit (color feature quantity extraction means)
121a Maximum value calculation circuit (maximum value calculation means)
121b Minimum value calculation circuit (minimum value calculation means)
121c Maximum density difference calculation circuit (maximum density difference calculation means)
121d Density average value calculation circuit (density average value calculation means)
121e Density difference sum calculation circuit (density difference sum calculation means)
122 color detection circuit (color detection means, color detection process)
123 Count circuit (counting means, counting process)
124 Black detection circuit (black detection means, black detection step)
132C, 132M, 132Y
First edge threshold selection circuit (edge threshold selection means)
135 Second edge threshold selection circuit (edge threshold selection means)
136 Edge feature extraction circuit (edge feature extraction means)
S135 Main edge threshold selection signal S136 Edge discrimination signal A1 First area A2 Second area C, M, Y Image data MF Mask filter

Claims (8)

Color detection means for detecting a black portion of a pixel in a first area composed of a target pixel in image data for each color component and a pixel in the vicinity of the target pixel;
Counting means for counting the number of pixels of black and a color other than black for the pixels in the first area,
Edge discriminating means for detecting whether or not the color changes rapidly at the target pixel, and
Based on the output results of these color detection means, counting means, and edge determination means, it is provided with black detection means for detecting whether the target pixel belongs to a black character area or a line drawing area ,
The edge discrimination means is
The pixels in the second area including the first area are binarized by a predetermined threshold of density level, and the main pixels are binned based on the number of inversions for each line in the main scanning direction and the sub scanning direction. Edge threshold selection means for generating an edge threshold selection signal;
For the first area and the second area, the maximum value, the minimum value, and the maximum density difference of the density levels of the included pixels are obtained, and the standard deviation of the density level of the pixels in the first area is obtained. An image comprising: an edge feature amount extraction unit that generates an edge determination signal indicating whether the pixel of interest is a character region or a non-character region based on the main edge threshold selection signal. Processing equipment. Based on the image data in the first area, provided with a color feature amount extraction means for extracting a feature amount indicating the characteristics of the first area,
The image processing apparatus according to claim 1, wherein the color detection unit and the counting unit perform processing based on an output result of the color feature amount extraction unit.   The color feature amount extraction means includes a maximum value calculation means for obtaining a maximum value, a minimum value calculation means for obtaining a minimum value, and a maximum density difference calculation means for obtaining a maximum density difference for the density levels of the pixels in the first area. A density average value calculating means for obtaining an average value, and a sum of absolute values of density level differences between pixels adjacent in the main scanning direction and a sum of absolute values of density level differences between pixels adjacent in the sub-scanning direction. 3. The image processing apparatus according to claim 2, further comprising density difference sum total calculating means for obtaining a sum.   The edge determination means includes edge detection means for performing edge detection of the pixel of interest by convolving a mask filter for edge extraction with respect to the pixels in the first area. The image processing apparatus according to claim 3. An image processing apparatus including an edge determination unit that detects whether or not a color changes rapidly at a target pixel in image data for each color component,
The edge discrimination means is
The pixels in the second area including the first area are binarized by a predetermined threshold of density level, and the main pixels are binned based on the number of inversions for each line in the main scanning direction and the sub scanning direction. Edge threshold selection means for generating an edge threshold selection signal;
For the first area and the second area, the maximum value, the minimum value, and the maximum density difference of the density levels of the included pixels are obtained, and the standard deviation of the density level of the pixels in the first area is obtained. An image comprising: an edge feature amount extraction unit that generates an edge determination signal indicating whether the pixel of interest is a character region or a non-character region based on the main edge threshold selection signal. Processing equipment. The edge determination means includes edge detection means for performing edge detection of the pixel of interest by convolving a mask filter for edge extraction with respect to the pixels in the first area. The image processing apparatus according to claim 5. A color detection step of detecting a black portion of a pixel in a first area including a target pixel in image data for each color component and a pixel in the vicinity of the target pixel;
For the pixels in the first area, a counting step of counting the number of pixels of black and a color other than black,
An edge determination step for detecting whether or not the color changes rapidly at the target pixel,
Based on the output results of the color detection step, the counting step, and the edge determination step, the black detection step of detecting whether the pixel of interest belongs to a black character region or a line drawing region,
The edge discrimination step is
The pixels in the second area including the first area are binarized by a predetermined threshold of density level, and the main pixels are binned based on the number of inversions for each line in the main scanning direction and the sub scanning direction. An edge threshold selection step for generating an edge threshold selection signal;
For the first area and the second area, the maximum value, the minimum value, and the maximum density difference of the density levels of the included pixels are obtained, and the standard deviation of the density level of the pixels in the first area is obtained. An image processing method comprising: an edge feature amount extraction step of generating an edge determination signal indicating whether the pixel of interest is a character region or a non-character region based on the main edge threshold selection signal . An image processing method including an edge determination step for detecting whether or not a color changes rapidly at a target pixel in image data for each color component,
The edge discrimination method is as follows:
The pixels in the second area including the first area are binarized by a predetermined threshold of density level, and the main pixels are binned based on the number of inversions for each line in the main scanning direction and the sub scanning direction. An edge threshold selection step for generating an edge threshold selection signal;
For the first area and the second area, the maximum value, the minimum value, and the maximum density difference of the density levels of the included pixels are obtained, and the standard deviation of the density level of the pixels in the first area is obtained. An image processing method comprising: an edge feature amount extraction step of generating an edge determination signal indicating whether the pixel of interest is a character region or a non-character region based on the main edge threshold selection signal .

Images