JPH0638054A - Image processor - Google Patents

Abstract

PURPOSE:To sharply reproduce black characters in color images by surely extracting picture elements at the edge parts of black characters and emphasizing the black color of these picture elements. CONSTITUTION:M data corresponding to Magenta are impressed to an edge detection part 4M after the density gradient is smoothed at a smoothing part 2M. When the density gradient near the notice picture element is sharp, the edge detection part 4M judges this notice picture element as the picture element of the edge part. Based on the three primary color data Y, M and C corresponding to Yellow, Magenta and Cyan, a black picture element detection part 7 defines an uncolored high-density picture element as a black picture element candidate. A detection area enlargement part 14 forcedly defines picture elements adjacent to the periphery of the picture element, which is defined as the black picture element candidate, as the black picture element candidate. When the picture element of the black picture element candidate is the picture element of the edge part at the same time, a selector 10 selects correction data from a color adjustment part 11. The color adjustment part 11 performs processing to emphasize the black color to the three primary color data of Y, M and C and BK data corresponding to black.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image processing apparatus such as a color copying machine, and more particularly to an image processing apparatus capable of favorably detecting and reproducing black characters.

[0002]

2. Description of the Related Art Conventionally, a color original is optically read by a scanner composed of a CCD (charge coupled device) or the like and three primary colors of additive colors of red (R), green (G) and blue (B) are added. A color copying machine is used which is converted into an electric signal and a color image is formed based on the signal. The R, G, and B primary color signals output from the scanner are converted into complementary color data of cyan (C), magenta (M), and yellow (Y) subtractive primary color data. This C,
Appropriate correction is applied to the three primary color data of M and Y, and black (BK) data is generated based on the corrected three primary color data.

For example, the surface of the photoconductor is scanned by a laser beam modulated based on the C signal, and an electrostatic latent image corresponding to cyan is formed on the surface of the photoconductor. This electrostatic latent image is visualized as a toner image using cyan toner and then transferred to a copy sheet. Similarly, in response to the M signal, the Y signal, and the BK signal, magenta, yellow, and black toner images are transferred onto the copy sheet in an overlapping manner, and finally the toner is heat-fixed to achieve a color copy. C, M,
A color image can be reproduced with only the toners of the three primary colors of Y, but if black toner is used, a high density portion can be satisfactorily expressed using black toner, so that color toner can be saved. .

[0004]

In the color copying machine as described above, when a manuscript containing black characters is copied,
There is a problem that the edges of black characters are hard to see.
That is, since the image is expressed by the toners of four colors of cyan, magenta, yellow, and black, bleeding due to the chromatic color component inevitably occurs at the edge portion of the character, and the boundary between the character portion and the background portion is clearly defined. I can't. Therefore, there is a problem that a clear character image cannot be obtained.
Moreover, in a color copying machine, generally, so-called dither processing is applied to each of C, M, Y, and BK data in order to express an intermediate gradation, so that the above tendency is promoted.

SUMMARY OF THE INVENTION An object of the present invention is to solve the above technical problems and to provide an image processing apparatus in which black characters can be detected or reproduced remarkably well.

[0006]

In order to achieve the above object, the image processing apparatus according to claim 1 is an image including an image including data corresponding to respective densities of three primary colors by optically reading a color image. Conversion means for converting to data,
Based on the image data, when the pixel of interest is an achromatic pixel and the density is equal to or higher than a predetermined value, a black pixel detection unit that makes this pixel of interest a black pixel candidate, and a detection result by the black pixel detection unit Based on the above, when a predetermined pixel around the target pixel is a black pixel candidate, the detection area enlarging unit forcibly making the target pixel a black pixel candidate, image data of the target pixel and the target pixel Edge detecting means for detecting a density gradient in the vicinity of the pixel of interest based on image data of a pixel adjacent to the edge pixel and determining the pixel of interest as a pixel of an edge portion when the density gradient is equal to or more than a predetermined value. When the target pixel is determined to be a pixel of an edge portion and is a black pixel candidate, the target pixel is determined to be a pixel forming an edge portion of a black character.

According to the above arrangement, the black pixel detecting means selects black pixel candidates of achromatic color and high density on the basis of the image data corresponding to the respective densities of the three primary colors created by the converting means. Detected as. Furthermore, based on the detection result of the black pixel detection means, the detection area expansion means
When a predetermined pixel around the target pixel is a black pixel candidate, the target pixel is set as a black pixel candidate. That is, when a certain pixel is detected as a black pixel candidate by the black pixel detecting means, the pixels surrounding the pixel in a predetermined positional relationship also become the black pixel candidate.

For example, due to color bleeding occurring at the edge portion of a character in a color image printed in color, the reading characteristics of the optical reading means, and the like, black characters are detected only by the processing by the black pixel detecting means. There is a possibility that the pixels in the edge portion cannot be detected as black pixel candidates. Even in such a case, the pixel of the edge portion of the black character can be surely set as the black pixel candidate by the function of the detection area expanding unit.

On the other hand, when the density gradient in the vicinity of the pixel of interest is equal to or larger than a predetermined value, it is determined that such a pixel constitutes an edge portion of some image. Then, if the pixels that are determined to form the edge portion are pixels that are also black pixel candidates at the same time, it is determined that the pixel is a pixel that constitutes the edge portion of the black character. In this way, the edge part of the black character is detected by using the detection of the edge part based on the density gradient and the detection of the black pixel candidate together.
In addition, as described above, the pixels at the edge portion of the black character can be surely set as the black pixel candidates without depending on the color blurring at the edge portion of the black character, so that the edge portion of the black character can be detected reliably. You can do it.

According to another aspect of the present invention, there is provided an image processing apparatus, wherein a color image is optically read and converted into image data including data corresponding to respective densities of three primary colors, and based on the image data, Black pixel detection means for selecting the target pixel as a black pixel candidate when the pixel is an achromatic pixel and the density is equal to or more than a predetermined value, and image data of the target pixel and image data of pixels adjacent to the target pixel. Based on the above, the density gradient in the direction from the adjacent pixel to the target pixel is detected, and when the density gradient is equal to or more than a predetermined value, the target pixel is determined as an edge part pixel, And a means for determining the pixel of interest as a pixel forming an edge portion of a black character when the pixel is determined to be a pixel of an edge portion and is a black pixel candidate.

According to this structure, the edge detecting means
The density gradient in the direction from the pixel adjacent to the target pixel toward the target pixel is detected. Then, when the density gradient is equal to or more than a predetermined value, the pixel of interest is determined to be a pixel of the edge portion. Therefore, even if the density gradient in the direction from the target pixel to the adjacent pixel is large, the target pixel is not determined to be the pixel of the edge portion. Therefore, only pixels inside the outline of the black character image are determined to be pixels of the edge portion, and pixels outside the outline are not determined to be pixels of the edge portion. In this way, the contour line of the black character image can be detected with high accuracy.

According to a third aspect of the present invention, there is provided an image processing apparatus, wherein a color image is optically read and converted into image data including data corresponding to respective densities of three primary colors, and based on the image data, When the pixel is an achromatic pixel and the density is equal to or more than a predetermined value, a black pixel detection unit that makes this pixel of interest a black pixel candidate, and based on the detection result of this black pixel detection unit, When a predetermined pixel is a black pixel candidate, a detection area expanding unit that forcibly sets the target pixel as a black pixel candidate, image data of the target pixel, and image data of pixels adjacent to the target pixel Based on the above, the density gradient in the direction from the adjacent pixel to the target pixel is detected, and when the density gradient is equal to or more than a predetermined value, the target pixel is determined to be an edge pixel. Di detecting means, when the pixel of interest is determined that the pixel of the edge portion, and is a black pixel candidate,
And a means for determining the pixel of interest as a pixel forming an edge portion of a black character.

In this configuration, the processing for enlarging the area of the black pixel candidate according to claim 1 and the edge detection processing with directionality according to claim 2 are used together. Therefore, the edge portion of the black character can be detected reliably and accurately.
An image processing apparatus according to claim 4, wherein the conversion means optically reads the color image and converts it into three primary color data corresponding to respective densities of the three primary colors, and further, based on the three primary color data, a black density is obtained. Corresponding black data is created, and the image processing apparatus further reduces the densities of the three primary colors and increases the densities of black with respect to the pixels determined to be the pixels of the edge portion of the black character. , And data correction means for correcting the three primary color data and the black data.

According to this structure, for the pixels forming the edge portion of the black character, the black data and the three primary color data in which the density of black is increased are generated from the data correcting means. As a result, black is emphasized in the edge portion of the black character, so that image data clearly expressing the boundary portion of the black character can be obtained. In this case, if the detection area expansion means is provided and the processing for expanding the area of the black pixel candidate is performed, the black color of the pixel at the edge portion of the black character can be surely emphasized. In addition, if the density gradient is determined in the edge detection process with directionality and only the pixels within the contour line of the black character image are determined to be edge pixels, the contour portion of the character image can be accurately measured. It can be reproduced. For this reason,
You can prevent the characters from getting fat, and even if the characters are close to each other with a narrow space, such as "quantity",
The margin between the lines can be reproduced well. In this way
High-quality image reproduction is possible.

According to a fifth aspect of the present invention, in the image processing apparatus, the edge detecting means is based on data of one specific color among the three primary color data of the target pixel and the pixel adjacent to the target pixel. It is characterized by determining whether or not the pixel of interest is a pixel in the edge portion. As a result, the structure of the edge detecting means can be simplified.

Magenta is preferably applied to the specific color when, for example, the three primary color data correspond to yellow, magenta, and cyan. This is because the magenta component forming the color image is the closest to the human visual characteristics. The image processing apparatus according to claim 6, further comprising a smoothing unit that creates data in which a density gradient in the vicinity of the target pixel is smoothed based on image data of the target pixel and a pixel adjacent to the target pixel.
The edge detection means is provided with the data smoothed by the smoothing means, and based on the smoothed image data of the target pixel and the pixel adjacent to the target pixel, the target pixel is a pixel of the edge portion. It is characterized by determining whether or not

According to this structure, since the edge detecting process based on the density gradient is performed by the edge detecting means after the density gradient is smoothed by the smoothing means, pixels in a region where the density gradient is relatively gentle are detected. It is prevented that the pixel is detected as an edge pixel. Therefore, it is possible to reliably detect the edge portion of the image. The image processing device according to claim 7, wherein the smoothing unit calculates a density gradient in the vicinity of the pixel of interest based on data of any one specific color of the three primary color data in pixels adjacent to the pixel of interest. In order to create smoothed data, the edge detection means is given the data smoothed by the smoothing means,
It is characterized in that it is determined whether or not the target pixel is a pixel of an edge portion based on the smoothed specific color data of the target pixel and the pixel adjacent to the target pixel.

According to this structure, the smoothing process and the edge detecting process are performed based on the data of one specific color, so that the structures of the smoothing unit and the edge detecting unit can be simplified. Moreover, since the edge detection is performed based on the smoothed data, the edge portion of the image can be surely detected.

[0019]

Embodiments of the present invention will be described in detail below with reference to the accompanying drawings. FIG. 1 is a block diagram showing an electrical configuration of a main part of a color copying machine which is an image processing apparatus according to an embodiment of the present invention. This image processing apparatus is read by a scanner or the like on the basis of three primary color data of yellow (Y), magenta (M), and cyan (C) and black (BK) data created from these three primary color data. Pixels constituting the edge portion of a black character existing in a read image are extracted, and the boundary between the black character portion and the background portion is made clear to reproduce the black character clearly.

A color original to be copied is optically read by a scanner (not shown) composed of a CCD (charge coupled device) or the like, and is subjected to a red (R), green (G), and blue (B) color addition method. It is photoelectrically converted into three primary color signals. 3 by this addition method
The primary color signals are converted into three primary color signals by the subtractive color method of yellow (Y), magenta (M), and cyan (C), which are complementary colors, and the signals are subjected to correction for color reproduction faithful to the original image. Then, Y, M, and C primary color image data is created. Further, BK data is created based on the three primary color data of Y, M and C. This BK data is data for compensating for the insufficient density in the high density portion with black toner, and for example, a predetermined correction coefficient α (for example, 0.5 to 1) is added to the minimum value of the three primary color data of Y, M, and C. It is created by removing the multiplied value from each of the three primary color data and using the removed value as the BK data. In the present embodiment, the conversion means is configured to include the above-mentioned scanner and the components for creating the Y, M, C and BK data.

Of the three primary color data of Y, M and C, M data corresponding to magenta is supplied to the smoothing unit 2M for smoothing the density gradient through the timing adjusting unit 1M. The data smoothed by the smoothing unit 2M is given to the edge detecting unit 4M for detecting the edge portion of the image such as a character through the timing adjusting unit 3M. The edge detection unit 4M detects whether or not each pixel is a pixel of an edge portion by detecting a density gradient between pixels in the vicinity of the target pixel, and outputs an output signal to the pixel of the edge portion. EDGE becomes logic "1". The signal of this logic "1" becomes an edge detection signal, and the pixel to which this edge detection signal is output is the pixel of the edge part of some image.

The reason for detecting the edge of the image based on the M data corresponding to magenta is that the magenta component forming the color image is close to the human visual characteristic. On the other hand, the three primary color data of Y, M, and C are given to the black pixel detection unit 7 via the timing adjustment units 6Y, 6M, and 6C, respectively. This black pixel detection unit 7
An achromatic pixel is detected based on the difference between the three primary color data of Y, M, and C, and when the density of this achromatic pixel is equal to or higher than a predetermined value, the pixel is regarded as a black pixel candidate. The output signal BLACK is set to logic “1” for the target pixel selected as the black pixel candidate.

The output signal of the black pixel detection unit 7 is input to the detection area expansion unit 14 via the timing adjustment unit 120.
As will be described later, the detection area enlarging unit 14 forcibly sets the target pixel as a black pixel candidate when a black pixel candidate pixel exists around the target pixel, and the target pixel is a black pixel candidate. It outputs a signal of logic "1" indicating that it is a candidate. As a result of the processing in the detection area expanding unit 14, all the pixels adjacent to the pixels of the black pixel candidate detected by the black pixel detecting unit 7 become the black pixel candidates. As a result, the black pixel candidate detected by the black pixel detection unit 7 is thickened by one dot.

This signal is input to one input terminal of the AND circuit 8. The output signal EDGE of the edge detector 4M is applied to the other input terminal of the AND circuit 8.
Therefore, the output of the AND circuit 8 is when the edge portion is detected based on the M data, and the detection area expansion unit 14 outputs the signal of logic “1” indicating that the target pixel is the black pixel candidate. Only when it is output, it becomes logical "1". In this case, it can be said that the pixel is a pixel forming an edge portion of a black image such as an edge portion of a black character.

The output signal of the AND circuit 8 is given to the selector 10 which will be described later from the line 9 as a selection control signal. A data correction unit is configured by including the logical product circuit 8, the color adjustment unit 11, the selector 10, and the like.
The three primary color data of Y, M, and C that have passed through the timing adjustment units 6Y, 6M, and 6C are also given to the color adjustment unit 11. The BK data whose timing is adjusted by the timing adjusting unit 6BK is also input to the color adjusting unit 11. The color adjusting unit 11 emphasizes the BK data and generates a signal corresponding to an image that is, so to speak, black and is close to a binary image. The output of the color adjusting unit 11 is input to the selector 10 via the timing adjusting unit 140.

The selector 10 also includes a color adjusting section 11
Timing adjustment section 6Y, 6M, 6C, 6BK
The three primary color data of Y, M, and C and BK data are given through the line 12 and further through the timing adjusting unit 13. The selector 10 responds to the selection control signal from the line 9 with Y,
M, C, BK data and Y from the timing adjustment unit 13,
One of Y, M, and M, C, and BK data
C and BK data will be selected. That is, the data from the color adjusting unit 11 is selected for the pixels forming the edge portion of the black character, and the data from the timing adjusting unit 13 is selected for the remaining pixels.

Yout, M output from the selector 10
BKout corresponding to 3 primary color data of out and Cout and black
Based on the data, the laser beam that forms an electrostatic latent image corresponding to each color image on the photoconductor is modulated. FIG. 2 is a block diagram showing a configuration example of the timing adjusting units 1M and 3M. The M data input according to the scanning order of the laser beam is first passed through the D (Delayed) type flip-flop (DFF) 15 as the data D2 and the smoothing unit 2
Input to M. Further, the data D2 is delayed by one line in a first-in first-out memory (hereinafter referred to as "FIFO") 16 and input to the smoothing unit 2M as data D1 via the flip-flop 17. This data D1 is FIFO
Flip-flop 19 after being delayed by one line at 18
Is input to the smoothing unit 2M as data D0.
As a result, the data D0, D1, and D2 provided in parallel to the smoothing unit 2Y are data corresponding to adjacent pixels in the sub-scanning direction when the laser beam scans the photoconductor.

The timing adjusting section 3M has substantially the same structure, and FIFOs 20, 21 and 22 each for delaying data by one line and D-type flip-flops 23, 24 and 25 for latching data from each FIFO.
And the output data DL of each flip-flop
2, DL1 and DL0 are input to the edge detection unit 4M.

FIG. 3 is a diagram for explaining the smoothing process in the smoothing unit 2M. This smoothing process is a process for smoothing the density gradient. The smoothing process for the target pixel S0 is performed by the density Q0 of this pixel, the densities Q1 and Q4 of the pixels S1 and S4 adjacent in the sub-scanning direction 27, and the main scanning direction 28.
Is performed on the basis of the densities Q2 and Q3 of the pixels S2 and S3 adjacent to each other. That is, the weighting factors of the pixels S0 to S4 are C0, C1, C2, C3, C4 (where C0 + C1
+ C2 + C3 + C4 = 1) and C1 = C2 = C3 = C4 = C (constant) (1), the density Qs of the pixel S0 after the smoothing process is obtained by the following equation (2).

[0030]

[Equation 1]

Here, C = 1 / K (3) (where K is an integer and K ≧ 5), and Q1 + Q2 + Q3 + Q4 = Qn (4) Then, Qs = (1-4 / K) Q0 + (Q1 + Q2 + Q3 + Q4) / K = Q0-4Q0 / K + Qn / K ... (5) is obtained.

The concrete structure of the smoothing unit 2M for realizing such smoothing processing is shown in FIG. The data D1 (see FIG. 2) from the timing adjustment unit 1M is
The D-type flip-flops 31, 32, and 33 delay the signal by 3 clocks (clock signal is represented by CLK) and input to the adder 34. In addition, 1 in the flip-flop 31
The data delayed by the clock is also given to the adder 34. The output data of the flip-flop 32 is the target pixel S0.
Supposing that the density Q0 corresponding to the above is represented, the outputs of the flip-flops 31 and 33 represent the densities Q3 and Q2 corresponding to the pixels S3 and S2, respectively, considering that the data is input in the order of scanning by the laser beam. It will be.

On the other hand, the data D0 from the timing adjusting unit 1M is delayed by 2 clocks by the flip-flops 35 and 36 and given to the adder 34. The output data of the flip-flop 36 represents the density Q1 of the pixel S1. Further, the data D2 is the flip-flop 3
The data is delayed by 2 clocks by 7, 38 and input to the adder 34 as data representing the density Q4 of the pixel Q4. Input data is added in the adder 34, and data corresponding to the above value Qn is created.

The output data of the flip-flop 32 representing the density Q0 is multiplied by 4 by the integrator 39 to generate 4Q0. Further, in the divider 40 that divides the input data by the constant K, 4Q0 / K is created, and this data is given to the subtractor 41. Data representing the density Q0 is also input to the subtractor 41 from the line 42, and the result (Q0-4Q0 / K) of the subtraction is given to the adder 44 from the line 43.

The adder 44 also outputs the output Q of the adder 34.
A value Qn / K obtained by dividing n by a constant K in the divider 45 is input, and the addition result of the adder 44 is
The density Qs after the smoothing process shown by the equation (5) is obtained. The data corresponding to this density Qs is the timing adjustment unit 3M.
(See FIG. 2). FIG. 5 is a diagram for explaining the processing in the edge detection unit 4M. The processing by the edge detection unit 4M is processing for detecting a portion having a steep density gradient as an edge of an image. In the edge detection process, the density q0 of the target pixel s0 and the surrounding pixels s1 to s1
It is performed based on the respective concentrations q1 to q8 of s8. That is, the density q0 of the target pixel s0 and the surrounding pixels s1 to s
Absolute value of the density difference with the density q1 to q8 of 8 | q0-q1
|, | Q0-q2 |, | q0-q3 |, | q0-q4
|, | Q0-q5 |, | q0-q6 |, | q0-q7
|, | Q0-q8 | is calculated. Then, each of these values is compared with a predetermined threshold value t, and when any of the values is greater than or equal to the predetermined value t, it is determined that the pixel of interest s0 is an edge pixel. It is determined that the target pixel s0 is not an edge pixel.

In summary, | q0-q1 | ≧ t ... (6) | q0-q2 | ≧ t ... (7) | q0-q3 | ≧ t ... (8) | q0 -Q4 | ≧ t ... (9) | q0-q5 | ≧ t ... (10) | q0-q6 | ≧ t ... (11) | q0-q7 | ≧ t ... · (12) | q0-q8 | ≧ t ······················· (13), the pixel of interest s0 is an edge pixel, and when none of the conditions is satisfied, the pixel of interest s0 is not an edge pixel. To be judged. By such a process, when a sharp change in density occurs in any direction around the pixel of interest s0, the pixel of interest s
0 will be detected as an edge pixel.

Since the edge pixel detection processing is performed based on the density value smoothed by the smoothing unit 2M, it is possible to prevent the portion where the density change is gentle from being determined as an edge pixel.
The edge pixels can be accurately detected. Figure 6
FIG. 4 is a block diagram showing a specific configuration example of an edge detection unit 4M. Data DL0, DL1, DL2 for three lines that have been subjected to the smoothing process given from the timing adjustment unit 3M
Are D-type flip-flops 163, 165, 1
68 is input. The data DL1 of the line including the pixel of interest s0 is delayed by one clock in each of the flip-flops 165, 160, 164 and input to the subtractor 170 as the density data q4 of the pixel s4. At this time,
The output of the flip-flop 160 corresponds to the density data q0 of the pixel of interest s0, and the output of the flip-flop 165 corresponds to the density data q5 of the pixel s5. These density data q0 and q5 are also input to the subtractor 170.

Similarly, the data DL0 is sequentially delayed by one clock by the flip-flops 163, 162 and 161. Then, each flip-flop 163, 16
The outputs of 2 and 161 are given to the subtractor 170 as density data q3, q2 and q1 of the pixels s3, s2 and s1, respectively. Similarly, the data DL2 is delayed by one clock by the flip-flops 168, 167, 166 connected in series. Then, each flip-flop 16
The outputs of 8, 167 and 166 are given to the subtractor 170 as density data q8, q7 and q6 of the pixels s8, s7 and s6.

The subtractor 170 has values (q0-q1) and (q0) obtained by subtracting the density data q1 to q8 of the target pixels s1 to s8 from the density data q0 of the target pixel s0.
-Q2), (q0-q3), (q0-q4), (q0-
q5), (q0-q6), (q0-q7), (q0-q
8) is calculated. This calculation result is the absolute value calculation unit 17
1 is input and positive values | q0-q1 |, | q0-q2
|, | Q0-q3 |, | q0-q4 |, | q0-q5
|, | Q0-q6 |, | q0-q7 |, | q0-q8 |
Is converted to.

The calculation result is input to the comparator 172 and compared with the threshold value t. Then, when any of the calculation results exceeds the threshold value t, a signal of logic "1" indicating that the target pixel s0 is a pixel at the edge portion is output. FIG. 7 is a diagram showing a situation of edge detection processing in the edge detection unit 4M. FIG. 7A shows input image data, and shows changes in M data in the main scanning direction. Moreover, FIG. 7 (b)
Shows the output of the comparator 172 according to the configuration of this embodiment. It is understood that the edge detection signals of logic "1" are obtained at the edge portions at both ends of the image as indicated by reference signs A1, A2, A3, and the like. In addition, in FIG. 7 (a),
"H" indicates that the symbol described before it is a hexadecimal number.

By such processing, for example, with respect to the original image shown in FIG. 8, the pixels in the area SE1 shown by hatching in FIG. 9 are extracted as edge pixels.
In this way, the pixels adjacent to the contour line 151 of the character image are detected as edge pixels. Note that, in FIG. 8, the symbol “●” represents a black pixel.

For the edge detection, a second derivative method using a derivative filter may be applied. That is, for example, the discriminant value a of the following formula (14) is used.

[0043]

[Equation 2]

Here, ci is a weighting coefficient, and the following (1)
Formula (5) is satisfied.

[0045]

[Equation 3]

In general, a positive integer is given to the weighting coefficient c0 for the target pixel s0, and negative values satisfying the above expression (15) are given to the pixels for the surrounding pixels s1 to s8.
That is, the weighting factors c0 to c8 are defined as, for example, c0 = 1 ... (16) c1 = c2 = .... = C8 = -1 / 8 ... (17).

Whether or not the pixel is an edge pixel is determined by comparing the discrimination value a with a predetermined threshold value t1. That is, when a> t1 ... (18), the target pixel s0 is determined to be an edge pixel, and when a ≦ t1 ... (19), the target pixel s0 is determined to be a non-edge pixel.

In this way, when the discriminant value a corresponding to the density gradient is larger than the threshold value t1, the target pixel s0 is determined to be an edge pixel. The result of applying the edge detection processing by this differential filter is shown in Fig. 7 (c).
Is shown in. From the comparison between FIG. 7 (b) and FIG. 7 (c), the edge of the image is detected as two pixels in the case of the configuration of FIG. 6, whereas the edge of the image is 1 when the differential filter is used. It is understood that the pixel can be detected.

FIG. 10 is a block diagram showing the structure of the timing adjusting section 6Y. This timing adjustment unit 6Y has a 2
FIFOs 61 and 62 and two flip-flops 6
4 and 65 are alternately connected, and the Y data is blacked in synchronization with the edge detection signal derived from the edge detection unit 4M in consideration of the data delay time in the timing adjustment units 1M and 3M. Pixel detection unit 7 and color adjustment unit 11
It is for giving to etc. The same applies to the configurations of the other timing adjusting units 6M, 6C, and 6BK.

FIG. 11 is a block diagram showing the basic structure of the black pixel detecting section 7. In the black pixel detection unit 7, an achromatic pixel having a high density is set as a black pixel candidate, and the black pixel detection signal BLACK is set to logic "1" for the black pixel candidate. It is a thing. The three primary color data of Y, M and C from the timing adjusting units 6Y, 6M and 6C are given to the subtractor 70, and the difference (Y−
M), (MC), and (CY) are created. The differences (Y−M), (M−C), and (C−Y) between the respective data are given to the absolute value calculation unit 71, and their absolute values | Y−M
|, | M-C |, | C-Y | are created, and the absolute value is given to the comparator 72. Each absolute value | Y
-M |, | MC |, | CY | and the threshold value T1 are compared, and all absolute values | Y-M |, | MC-, | CY
When | is equal to or less than the threshold value T1, it is determined that the pixel of interest is a gray pixel which is an achromatic pixel, and a signal of logic “1” is derived on the line 73. Absolute value | Y-M |, | M
When any one of −C | and | C−Y | exceeds the threshold T1, it is determined that the pixel is a color pixel.

That is, in an achromatic pixel, the difference between the three primary color data of Y, M, and C is basically zero, and a large difference does not occur even after various corrections. Therefore, B = Max {| Y−M |, | M−C |, | C−Y |} (20) The maximum value B of the differences between the three primary color data is given by a predetermined value. When there is a relation of B ≦ T1 ... (21) with respect to T1, it can be determined that the pixel is a gray pixel which is an achromatic pixel, and there is a relation of B> T1 ... (22). At times, the pixel can be determined to be a color pixel.

Such processing is performed by the difference between the three primary color data |
It is possible to replace with the above-described processing for determining whether or not any of Y-M |, | MC |, | C-Y | exceeds the predetermined value T1. Can be simplified. On the other hand, the three primary color data of Y, M, C from the timing adjustment units 6Y, 6M, 6C are
It is also provided to the comparator 74. The comparator 74 determines the yellow density level Y of the pixel of interest represented by each data.
d, the density level Md of magenta and the density level Cd of cyan are all equal to or greater than a predetermined threshold value T2,
A signal of logic "1" is derived from the line 75 as a high density pixel detection signal. When any of the density levels is lower than the threshold value T2, a signal of logic "0" is derived on the line 75 as the low density pixel detection signal.

The signals from the lines 73 and 75 are input to the AND circuit 76. In the logical product circuit 76, after all, the comparator 72 determines that the pixel of interest is a gray pixel, and the comparator 7
The signal of logic "1" is derived only when the pixel 4 determines that the pixel of interest is a high density pixel. This signal of logic “1” indicates that the pixel of interest is a black pixel candidate. Figure 1
2 is a diagram for explaining the processing in the detection area expansion unit 14. The detection area expansion unit 14 determines that at least one of the eight pixels P1 to P8 adjacent to the target pixel P0 is a pixel determined by the black pixel detection unit 7 to be a black pixel candidate, and The target pixel P0 is forcibly made a black pixel candidate. Such processing is achieved by the configuration shown in FIG. That is, the output signal BLACK of the black pixel detection unit 7 is input to the timing adjustment unit 120. The timing adjustment unit 120 includes a D-type flip-flop 121, 122, 123 and a FIFO memory 12 capable of storing pixel data for one line.
4, 125 are alternately connected.

The output signal of the black pixel detecting section 7 is first delayed by one clock in the flip-flop 121, and this signal is input to the detection area expanding section 14 as data DA2. The output data of the flip-flop 121 is further FI
The data is delayed by one line in the FO memory 124, further delayed by one clock in the flip-flop 122, and provided to the detection area expansion unit 14 as data DA1. The data DA1 is further delayed by one line in the FIFO memory 125, further delayed by one clock in the flip-flop 123, and then input to the detection area expansion unit 14 as data DA0.

In this way, the data DA0, DA1, DA2 for three adjacent lines are arranged in parallel in the detection area expansion unit 1.
4 will be given. The detection area expansion unit 14 includes
Three flip-flops 133, 132, 131 for delaying the data DA0 by one clock, and one data DA1
Three flip-flops 13 that delay each clock
5, 130, 134 and three flip-flops 138, 13 for delaying the data DA2 by one clock.
7,136 are provided. The output data of these nine flip-flops is the pixel of interest P shown in FIG.
It is the output data of the black pixel detection unit 7 corresponding to each of the nine pixels P0 to P8 centered on 0. The correspondence relationship between the output data of each of the flip-flops 131 to 139 and the pixels P0 to P8 is as follows.

P0 ··· flip-flop 130 P1 ··· flip-flop 131 P2 ··· flip-flop 132 P3 ··· flip-flop 133 P4 ··· Flip-flop 134 P5 ··· Flip-flop 135 P6 ··· Flip-flop 136 P7 ··· Flip-flop 137 P8 ··· Flip-flop 138 Data of each pixel P0 to P8 Are commonly used as the OR circuit 150.
Entered in. Therefore, the output signal of the logical sum circuit 150 becomes a logic "1" if at least one of the target pixels P0 to P8 is a black pixel candidate. This logic "1" signal is input to the AND circuit 8 of FIG. 1 as a black pixel detection signal.

In this way, when at least one of the pixels P1 to P8 around the target pixel P0 is a black pixel candidate, the target pixel P0 is forcibly a black pixel candidate. For example, a case of copying the document image shown in FIG. 8 will be taken as an example. When the black pixel detection unit 7 processes this original image, a black pixel area SB1 shown by hatching in FIG. 14 is obtained. As you can see from this Figure 14,
In many cases, the pixels adjacent to the contour 151 cannot be detected as black pixel candidates only by the processing in the black pixel detection unit 7. This is due to color bleeding occurring at the edge portion of the color image and the reading characteristics of the CCD scanner.

When the processing by the detection area expanding section 14 is applied to the processing result of the black pixel detecting section 7 shown in FIG.
5, a black pixel area SB2 indicated by hatching is obtained. From this FIG. 15, it is understood that the pixels adjacent to the contour 151 are surely detected as black pixel candidates. In order to surely reach the pixel adjacent to the contour line 151 in the black pixel region, it is conceivable to set the threshold value T2 regarding the density in the black pixel detection unit 7 to be low. But,
By doing so, it is possible that even high-density pixels with a darkness that do not reach black pixels at the same time will be extracted as black pixel candidates, and pixels other than the edge portions of black characters are mistakenly recognized as edge pixels of black characters. There is a possibility that a problem that it is processed as a result occurs. Such a defect can be prevented by applying the above-described configuration in which the output of the black pixel detection unit 7 is processed by the detection area expansion unit 14.

In the logical product circuit 8 of FIG.
When both the output signal of M and the output signal of the detection area expansion unit 14 are logic "1", a logic "1" indicating that the pixel is a pixel of an edge portion of a black image area such as a character.
Derive the signal of. That is, when the edge pixel is also a black pixel candidate at the same time, it is determined that the pixel is a pixel of an edge portion of a black image area such as a character.

Specifically, for the original image of FIG.
It is assumed that the pixels in the region SR1 shown by double hatching in FIG. 16, which is the overlapping portion of the edge region SE1 of FIG. 9 and the black pixel region SB2 of FIG. 15, are the pixels of the edge portion such as characters. Will be. The detection area expansion unit 1
4 is omitted, the edge area SE1 of FIG.
Since the overlapping portion with the black pixel area SB1 is the pixel of the edge portion of the character or the like, the pixel indicated by double hatching in FIG. 17 is extracted as the pixel of the edge portion of the character or the like. . That is, the pixels adjacent to the contour 151 cannot be reliably extracted as the pixels of the edge portion.

FIG. 18 is a block diagram showing the basic arrangement of the color adjusting section 11. In this color adjustment unit 11,
Input Y, M, C three primary color data and BK data are data Y ', M', shown by the following formulas (23) to (26).
It is corrected to each data of C'and BK '. Y '= Y-BK / L ... (23) M' = M-BK / L ... (24) C '= C-BK / L ... (25) BK' = BK + BK / L (26) However, in the above formulas (23) to (26), L is a correction coefficient, and when the correction coefficient when generating the BK data is α, 1 / L It is preferable that ≦ 1 / α−1 (27) The correction coefficient α at the time of generating the BK data is a ratio when the minimum data of the three primary color data of Y, M, and C is replaced with the BK data.

By the above correction, a set of correction data Y ', M', C ', BK' in which the BK data is emphasized can be obtained. The BK data from the timing adjustment unit 6BK is the divider 8 that divides the input data by the above correction coefficient L.
Input to 1. The output of the divider 81 becomes BK / L, and this data is added to BK by the adder 82 to produce BK '.
= BK + BK / L is generated. This correction data BK '
Are provided to the selector 10.

The output of the divider 81 is also given to the subtractor 83. In this subtractor 83, the three primary color data of Y, M, C from the timing adjustment units 6Y, 6M, 6C and the data BK / K from the divider 83 are subtracted, and the correction data Y '= Y-BK. / L, M '= M-BK / L,
C '= C-BK / L is generated. The correction data Y ′, M ′, C ′ are input to the selector 10.

In the selector 10, as described above, the three primary color data of Y, M and C from the timing adjusting section 13 and BK
Data has been input, and based on the selection control signal from the line 9, either the data set of Y, M, C, BK or the data set of Y ', M', C ', BK' is set to the pixel. Output data Yout, Mout, C selected for each
out, BKout.

The selection control signal derived from the AND circuit 8 to the line 9 becomes the logic "1" indicating that the output signal EDGE of the edge detection unit 4M has detected an edge pixel, and the output of the detection area expansion unit 14 further. The signal becomes logic "1" only when the signal becomes logic "1" indicating that the pixel of interest is a black pixel candidate. At this time, the selector 10 outputs Y ', M',
The data sets of C'and BK 'are selected and output. In this way, the selector 10 makes the Y ', where the BK data is emphasized only for the constituent pixels of the edge portion of the black character.
The data set of M ', C', and BK 'will be selected.

In this way, in the data Yout, Mout, Cout, BKout output from the selector 10, the BK data is emphasized only for the constituent pixels of the edge portion of the black character. Therefore, in an image of each color of yellow, magenta, cyan, and black formed based on the signal from the selector 10, the density of yellow, magenta, and cyan is increased with respect to the constituent pixels of the edge portion of the black character. Is lowered and black is increased in its density. As a result,
A binary image of only black is formed on the edge portion of the black character, so that an image with a clear boundary between the character portion and the background portion can be formed on the copy sheet. In this way, clear black characters can be reproduced.

In the present embodiment, the detection result of the black pixel detection unit 12 is further processed by the detection area expansion unit 14 to reliably detect the pixels forming the outline of the character as black pixel candidates. I am trying to do it. That is, even if the pixels in the outline portion of the character are not detected as black pixel candidates due to the influence of the color blur occurring in the outline portion of the character or the reading characteristics of the CCD scanner for reading a color image, Such pixels are forcibly made into black pixel candidates by the processing in the detection area expansion unit 14. For this reason,
The output of the logical product output circuit 8 can be surely made to be logic "1" with respect to the pixel of the edge portion of the black character.

As described above, according to the present embodiment, the black color can be surely emphasized with respect to the pixel at the edge portion of the black character. FIG. 19 is a block diagram for explaining the configuration of another embodiment of the present invention. In the present embodiment, in the configuration similar to that of FIG. 1, instead of the edge detection unit 4M shown in FIG. 6, the edge detection unit 4M of the configuration shown in FIG.
a is used. Note that, in FIG. 19, portions having the same functions as the respective portions shown in FIG. 6 described above are denoted by the same reference numerals. Further, FIG. 5 described above will also be referred to in order to explain the configuration and operation of the edge detection unit 4Ma.

Edge detection unit 4Ma used in this embodiment
Then, the output of the subtractor 170 is directly input to the comparator 172. That is, the absolute value calculation is omitted.
Therefore, the processing in the edge detection unit 4Ma is summarized as follows: q0-q1 ≧ t ... (28) q0-q2 ≧ t ... (29) q0-q3 ≧ t. ) q0-q4 ≧ t (...) (31) q0-q5 ≧ t (...) (32) q0-q6 ≧ t (...) (33) q0-q7 ≧ t (...) (34) q0 -Q8≥t ... Process for determining the target pixel s0 as an edge pixel when any of the following (35) is satisfied, and determining that the target pixel s0 is not an edge pixel when none of the following is satisfied. Is.

As a result of such processing, in the edge detecting section 4M having the configuration shown in FIG. 6, if the density change in any direction is sharp, it is determined as an edge pixel.
In the present embodiment, the target pixel s0 is determined to be an edge pixel only when the density sharply increases from the surrounding pixels s1 to s8 toward the target pixel s0. As a result, for example, in the configuration of FIG. 6, the pixels on both sides of the outline 151 of the character image are determined to be edge pixels as shown in FIG. 9, whereas in the configuration of the present embodiment,
As shown by hatching in FIG. 20, the outline 151
Only the pixels inside the are determined to be edge pixels.

Therefore, as shown by the double hatching in FIG. 21, the BK data is emphasized only for the pixels inside the contour 151. As a result,
As can be seen from the comparison between FIG. 6 and FIG. 21, the contour of the image forming the character can be accurately reproduced. That is, it is possible to prevent the line forming the character from becoming thick. Moreover, since the image width in which the BK data is emphasized is constant for one pixel, black characters can be reproduced with high quality.

Such a process is particularly effective in reproducing a character in which thin lines are adjacent to each other with a narrow interval, such as a Chinese character “quantity”. That is, it is assumed that the document image on the left shoulder of the kanji "quantity" is an image as shown in FIG. In this case, in the above-described first embodiment, the BK data is emphasized with respect to the pixels in the region SR10 indicated by the double diagonal line in FIG. On the other hand, in the present embodiment, FIG.
Area SR1 located inside the contour line 152 as shown in FIG.
BK only for pixels of 1 (shown with double diagonal lines)
The data will be highlighted.

As a result, in the state of FIG. 23, black pixels may be filled in between thin lines, whereas
In this embodiment, as can be seen from FIG. 24, the blank area between the thin lines can be reproduced well. The fact that the blank area is filled with the black pixels as shown in FIG. 23 is due to the fact that the black pixel area is enlarged in the detection area expanding section 14. As described above, the pixel adjacent to the contour line 15 cannot be extracted as a black pixel candidate if the above is omitted. That is, the detection area expansion unit 14 expands the black pixel area, and further the edge detection unit 4M
By detecting only the pixels inside the contour line 15 as edge pixels in a, the BK data can be surely emphasized only to the pixels adjacent to the inside of the contour line 15. In this way, the contour of a black image such as a character can be clearly reproduced.

Although the embodiments of the present invention have been described above, the present invention is not limited to the above embodiments. For example, in the above-described embodiment, only the density data of magenta is used for the edge detection by utilizing the fact that the density of the magenta component forming the color image is close to the human visual characteristics, but Y , M and C data may be used. In this case, as shown in FIG. 25, Y data and C data are processed in the same manner as M data, and the output signals of the edge detectors 4Y, 4M, and 4C are input to the AND circuit 5. The output signal EDGE of the AND circuit 5 may be input to the AND circuit 8 (see FIG. 1). As a result, when the density of all three primary colors is steep, it is determined that the pixel of interest is an edge pixel.

Note that, in FIG. 25, in each component that applies the same processing to each component that processes M data to Y data and C data, the numeral portion of the reference numeral attached to each component corresponding to M data is added. The same numbers are shown with symbols Y and C, respectively. In addition, it goes without saying that in the configuration of FIG. 25, instead of the edge detection units 4Y, 4M, and 4C, an edge detection unit similar to the configuration illustrated in FIG. 19 may be applied.

Further, in the above-mentioned embodiment, the detection area enlarging unit 14 has eight pixels P1 adjacent to the target pixel P0.
-P8 (see FIG. 12) includes at least one black pixel candidate, the target pixel P0 is forcibly set as a black pixel candidate. Instead of such a process, for example, at least one of the pixels P2, P4, P5, and P7 that are vertically and horizontally adjacent to the target pixel P0.
That one is a black pixel candidate is used as a condition for forced conversion into a black pixel candidate, or a pixel P adjacent to the right and left of the target pixel P0
It is also possible to set a condition that at least one of P4 and P5 is a black pixel candidate.

For example, when at least one of the pixels P2, P4, P5, and P7 adjacent to the target pixel in the upper, lower, left, and right directions is a black pixel candidate, when the target pixel P0 is forced to be a black pixel candidate, In the configuration of 13,
Flip-flops 132, 134, 135 and 137
It is only necessary to input only the output of the above to the OR circuit 150 and not to output the output of the flip-flop to the OR circuit 150.

In the above embodiment, the target pixel P0 is forcibly set as the black pixel candidate when the black pixel candidate exists within the range of one pixel around the target pixel P0. The presence of black pixel candidates in the range of, for example, two pixels around P0 may be a condition for forced conversion into black pixel candidates. Such processing is performed, for example, in FIG.
This can be achieved by providing two detection area expansion units 14 having the configuration shown in (1) and connecting the two detection area expansion units in series. That is, in the detection area expansion unit 14 shown in FIG. 13, when a certain pixel is a black pixel candidate, the pixels adjacent to it are regarded as black pixel candidates. Therefore, if two detection area expansion units having the same configuration as the detection area expansion unit 14 are connected in series, when a certain pixel is a black pixel candidate, pixels in the surrounding two pixels are black. It will be a pixel candidate. As a result, the target pixel P0
If there is a black pixel candidate in the range of two pixels around, the target pixel P0 is forcibly set as a black pixel candidate.

In the above embodiment, the target pixel s0 and the eight surrounding pixels s1 to s8 are used to detect edge pixels.
Is determined and whether or not the target pixel s0 is an edge pixel is determined based on this difference. However, this process is simplified and the target pixel s0 and the four pixels s above, below, left and right thereof are
The edge determination process may be performed based on the density difference between 2, s4, s5, and s7.

Further, although the above embodiment has been described by taking a color copying machine as an example, the present invention is a color facsimile or other color image processing in which an optically read color image is stored or reproduced as data. The present invention can be easily applied to various image processing devices such as an image reading device that optically reads an image and converts it into a color signal.

In addition, various design changes can be made without changing the gist of the present invention.

[0082]

As described above, according to the image processing apparatus of the present invention, the pixels determined to constitute the edge portion of the image are
At the same time, when the pixel is a black pixel candidate, the pixel is determined to be a pixel forming an edge portion of a black character.
As described above, since the detection of the edge portion based on the density gradient and the extraction of the black pixel candidate are used together, the edge portion of the black character is surely detected.

As a result, it becomes possible to reliably detect the black character, and if the correction of emphasizing the black color is performed on the data of the pixel which is determined to be the pixel forming the edge portion of the black character, the black character and the background portion are removed. It is possible to obtain image data in which the boundary between and is clearly expressed. Further, in addition to the processing by the black pixel detecting means that uses a high-density achromatic pixel as a black pixel candidate, when there is a pixel which is a black pixel candidate by the black pixel detecting means around the target pixel, the target pixel concerned By forcibly processing as a black pixel candidate, the pixel adjacent to the contour line of the black character image can be reliably extracted as a black pixel candidate. As a result, the influence of the color bleeding that occurs at the edge of the black character and the reading characteristics when reading a color image can be eliminated, and the pixel at the edge of the black character can be reliably extracted as a black pixel candidate.

When detecting the edge portion of a black character, if the density gradient in the direction from the pixel adjacent to the pixel of interest toward the pixel of interest is steep,
Only pixels inside the contour of the black character image are detected as edge pixels. Therefore, the lines forming the characters do not become thick, and the image is not crushed even with respect to characters such as "quantity" in which the lines are densely packed at narrow intervals. In this way, each line forming a character can be reproduced well, and a high-quality image can be reproduced.

[Brief description of drawings]

FIG. 1 is a block diagram showing an electrical configuration of a main part of a color copying machine which is an embodiment of an image processing apparatus of the present invention.

FIG. 2 is a block diagram showing a configuration of timing adjusting units 1M and 3M.

FIG. 3 is a diagram for explaining a smoothing process.

FIG. 4 is a block diagram showing a configuration of a smoothing unit 2M.

FIG. 5 is a diagram for explaining edge detection processing.

FIG. 6 is a block diagram showing a configuration of an edge detection unit 4M.

FIG. 7 is a diagram for explaining the operation of the edge detection unit 4M.

FIG. 8 is a diagram showing an example of a document image.

9 is a diagram showing a result of performing edge detection processing on the document image of FIG.

FIG. 10 is a block diagram showing a configuration of a timing adjusting unit 6Y.

FIG. 11 is a block diagram showing a configuration of a black pixel detection unit 7.

FIG. 12 is a diagram for explaining a process in a detection area expansion unit 14.

FIG. 13 is a block diagram showing a configuration of a detection area expansion unit 14.

FIG. 14 is a diagram showing a result of performing a black pixel candidate extraction process by a black pixel detection unit 7 on the original image in FIG.

FIG. 15 is a diagram showing a result of performing processing by a detection area expanding unit 14 on a processing result by the black pixel detecting unit 7 for the original image in FIG.

16 is a diagram showing pixels extracted as edge portions of black characters in the original image of FIG.

FIG. 17 is a diagram showing pixels extracted as an edge portion of a black character when the processing in the detection area expanding unit 14 is omitted.

FIG. 18 is a block diagram showing a configuration of a color adjusting unit 11.

FIG. 19 is a block diagram showing the configuration of an edge detection unit 4Ma used in another embodiment of the present invention.

20 is a diagram illustrating the edge detection unit 4 for the original image of FIG.
It is a figure which shows the edge pixel extracted by performing the process by Ma.

FIG. 21 is a diagram showing pixels in which black is emphasized when the edge detection unit 4Ma is applied.

FIG. 22 is a diagram showing a part of a document image forming a character with dense lines.

FIG. 23 is a diagram showing pixels in which black is emphasized when the document image of FIG. 22 is processed according to the first embodiment.

FIG. 24 is a diagram showing pixels in which black is emphasized when processed using the edge detection unit 4Ma.

[Fig. 25] Fig. 25 is a block diagram illustrating another configuration example for extracting pixels in an edge portion.

[Description of Reference Signs] 2M smoothing unit 4M edge detection unit 7 black pixel detection unit 8 AND circuit 10 selector 11 color adjustment unit 14 detection area expansion unit 4Ma edge detection unit 2Y smoothing unit 2C smoothing unit 4Y edge detection unit 4C Edge detection unit 5 AND circuit

Claims (7)

[Claims] 1. A conversion means for optically reading a color image and converting it into image data containing data corresponding to respective densities of three primary colors; and, based on the image data, the pixel of interest is an achromatic pixel. Moreover, when the density is equal to or higher than a predetermined value, a black pixel detecting unit that makes the target pixel a black pixel candidate, and a predetermined pixel around the target pixel is a black pixel candidate based on the detection result of the black pixel detecting unit. Detection area enlarging means for forcibly making the target pixel a black pixel candidate when the target pixel is a selected pixel, and the vicinity of the target pixel based on the image data of the target pixel and the image data of the pixel adjacent to the target pixel. Edge detecting means for detecting the density gradient of the target pixel and determining that the target pixel is a pixel of the edge portion when the density gradient is equal to or more than a predetermined value; An image processing apparatus comprising: a unit that determines a pixel of interest as a pixel forming an edge portion of a black character when it is determined as a pixel candidate. 2. A conversion means for optically reading a color image and converting it into image data containing data corresponding to respective densities of the three primary colors, and the pixel of interest is an achromatic pixel based on the image data. In addition, based on the image data of the pixel of interest and the image data of the pixel adjacent to the pixel of interest, the black pixel detection means that makes this pixel of interest a black pixel candidate when the density is greater than or equal to a predetermined value An edge detection unit that detects a density gradient in the direction toward the target pixel and determines that the target pixel is a pixel at the edge portion when the density gradient is equal to or greater than a predetermined value, and determines that the target pixel is a pixel at the edge portion. And an image processing device that determines that the pixel of interest is a pixel forming an edge portion of a black character when the pixel of interest is determined to be a black pixel candidate. 3. A conversion means for optically reading a color image and converting it into image data containing data corresponding to respective densities of the three primary colors; and, based on the image data, the pixel of interest is an achromatic pixel. Moreover, when the density is equal to or higher than a predetermined value, a black pixel detecting unit that makes the target pixel a black pixel candidate, and a predetermined pixel around the target pixel is a black pixel candidate based on the detection result of the black pixel detecting unit. A detected pixel enlarging unit that forcibly sets the target pixel as a black pixel candidate when the target pixel is a selected pixel, and based on the image data of the target pixel and the image data of the pixel adjacent to the target pixel, An edge detection unit that detects a density gradient in the direction toward the target pixel and determines that the target pixel is a pixel of an edge portion when the density gradient is equal to or more than a predetermined value, and the target pixel is An image processing apparatus comprising: a unit that determines that the pixel of interest is a pixel forming an edge portion of a black character when the pixel of interest is determined to be a pixel of a black portion and is a black pixel candidate. 4. The conversion means optically reads a color image and converts it into three primary color data corresponding to respective densities of the three primary colors, and further, based on the three primary color data, black data corresponding to a black density. The image processing apparatus further reduces the densities of the three primary colors and increases the densities of black with respect to the pixels determined to be the pixels of the edge portion of the black character.
4. The image processing apparatus according to claim 1, further comprising data correction means for correcting the primary color data and the black data. 5. The edge detecting means determines that the pixel of interest is an edge pixel based on the data of one specific color among the three primary color data of the pixel of interest and a pixel adjacent to the pixel of interest. The image processing apparatus according to claim 1, wherein the image processing apparatus determines whether or not there is. 6. The edge detecting means further includes a smoothing means for creating data in which a density gradient in the vicinity of the target pixel is smoothed based on image data of the target pixel and a pixel adjacent to the target pixel. The data smoothed by the smoothing means is given, and it is determined whether the pixel of interest is an edge pixel based on the smoothed image data of the pixel of interest and the pixel adjacent to the pixel of interest. The image processing apparatus according to any one of claims 1 to 4, wherein the image processing apparatus makes a determination. 7. The smoothing means smoothes a density gradient in the vicinity of the target pixel based on data of any one specific color among the three primary color data in pixels adjacent to the target pixel. The edge detecting means is provided with data smoothed by the smoothing means, based on the smoothed specific color data of the pixel of interest and the pixel adjacent to the pixel of interest, The image processing apparatus according to claim 6, wherein it is determined whether or not the pixel of interest is a pixel of an edge portion.