JPH09149240A - Image processor - Google Patents

Abstract

PROBLEM TO BE SOLVED: To allow the image processor to magnify/reduce binary data at an optional magnification, to apply smoothing a character part at magnification, to reserve black thin lines at reduction and to reproduce a half tone part in medium tone gradation. SOLUTION: In this image processor, an image area separate circuit 3 separates image data in an image memory 2 into a character image and a medium tone image, a control circuit 15 generates a projection coefficient, a projection address, a character address, and a character correction processing circuit 7 generates smoothing image data at magnification, a projection processing circuit 12 outputs projection image data based on the projection coefficient, an image signal synthesis circuit 18 binarizes data from the character correction processing circuit 7 in the case of a character image based on the result of discrimination from the image area separate circuit 3 and receives data from the projection processing circuit 12 in the case of a medium tone image to synthesize the data into a multi-value processing image and an error spread processing circuit 19 binarizes the synthesis data.

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 for digital data such as a facsimile, a scanner and a digital copy, and more particularly to an image processing apparatus for enlarging and reducing a binarized image at an arbitrary magnification.

[0002]

2. Description of the Related Art Conventionally, an image processing apparatus having a function of enlarging / reducing a binarized image of this type performs thinning processing at intervals corresponding to a reduction rate at the time of reduction, and at the time of enlargement, performs an integer multiple enlargement and then reduction. By the same process as that at the time, the thinning process is performed at a cycle corresponding to the enlargement ratio to perform the enlargement / reduction process of an arbitrary magnification. This technique is disclosed in JP-A-6-164896.

[0003]

However, in the above-mentioned conventional technique, the thinning process is periodically performed regardless of the input image data, so that a smoothed image is obtained for the character portion of the input image. However, noise is increased in the halftone halftone portion, and the reproducibility of gradation is lost. In addition, there is a problem that the fine black line is erased at the time of reduction and the reproducibility is lost.

The present invention has been made in view of the above-described problems, and it is possible to perform enlargement / reduction processing on a binarized image at an arbitrary magnification, a character portion is smoothed at the time of enlargement, and a black thin line is made at the time of reduction. It is an object of the present invention to provide an image processing apparatus in which the halftone part is stored and halftoned, and the halftone part retains halftone gradation reproducibility without causing an obstacle such as noise due to the value of the scaling ratio. .

[0005]

In the image processing apparatus of the present invention, smoothing processing is performed on binary image data at the time of enlarging, and processing of not erasing black fine lines is performed at the time of reducing. A scaling factor is set, a projection coefficient is generated from this, and multivalued projection pixel data corresponding to the ratio of the input black pixel projected on the output image is generated by this projection coefficient and the binary image data. Also, the image area of the binary image data is determined, and in the case of a character image, the image data is subjected to smoothing processing or black thin line erasure prevention processing, and in the case of a halftone image, it is multi-valued projection image data. Then, it is converted into binary data including halftone by error diffusion processing. According to the present invention, a binary image can be enlarged / reduced at an arbitrary magnification, a character portion is smoothed at the time of enlargement, a black thin line is preserved at the time of reduction, and a halftone portion has noise or the like depending on the enlargement / reduction value. Gradation reproducibility of halftone is maintained without occurrence.

[0006]

According to a first aspect of the invention, image area discrimination means for inputting binary image data from an image memory and discriminating a character image and a halftone image by referring to a predetermined area of the image data. And inputting binary image data from the image memory, and generating smoothing image data around the output pixel on which the center of the input pixel is projected and its periphery during the enlargement processing,
Character correction processing means for performing a saving process of a black thin line projected on an output image at the time of reduction, and control means for generating a projection coefficient indicating a ratio of projection of an input pixel to an output pixel according to a set magnification of the output image with respect to the input image. A projection processing means for generating multi-valued projection image data according to the ratio of the input black pixel projected to the output pixel by the projection coefficient and the binary image data input from the image memory; and the image area discrimination means. In the case of a character image based on the output data of, the output data of the character correction processing means is input, in the case of a halftone image, the output data of the projection processing means is input, and combining means for combining these into a multi-valued image, An error diffusion processing unit that performs an error diffusion process on the data output from the synthesizing unit is provided.

According to the present invention, the character correction processing means performs smoothing processing at the time of enlargement and processing that does not erase the black thin line at the time of reduction. The control means generates a projection coefficient according to the set enlargement / reduction ratio, and the projection processing means generates multi-valued projection image data according to the ratio of the input black pixel projected onto the output pixel by this projection coefficient and the binary image data. . The image area discrimination means discriminates between the character image and the halftone image by referring to the predetermined area of the input binary image, and the compositing means inputs the binary data of the character correction processing means in the case of the character image according to this discrimination. However, in the case of a halftone image, the multi-valued image of the projection processing means is input, and the binary data is set to the highest level and the lowest level of the multi-valued data to combine both data, and this multi-valued combined data is obtained. It is converted into binary data including halftone by the error diffusion process and is output.

As a result, the halftone portion of the input image data is subjected to projection processing to be multi-valued, and the local reflection of the multi-valued projection image matches the local reflectance of the input binary image. . The halftone image obtained by performing the error diffusion processing on the multivalued data in which the local reflectance is preserved in this way has very little noise generation, and the reproducibility of gradation is preserved. In the character portion, smoothing interpolation data is enlarged and inserted in the enlargement processing, and black thin lines are saved in the reduction processing.

According to a second aspect of the present invention, in the first aspect of the present invention, a discrimination result memory for storing the discrimination result of the image area discriminating means, a character image memory for storing output data of the character correction processing means, and And a projection image memory for storing output data of the projection processing means, wherein the control means generates a projection address of data to be stored in the projection image memory and corresponds to the discrimination result memory and the character image memory with the projection address. Generate a character address.

According to the present invention, since the projection address of the projection image memory is generated by the control means and the character address corresponding to this projection address is generated in the discrimination result memory and the character image memory, the projection image data can be obtained at any magnification. And the position of the character image data coincide with each other, and the image quality deterioration can be suppressed to a minimum even when the image area discrimination is erroneously made.

According to a third aspect of the present invention, in the second aspect of the invention, the control means projects a grid point representing a position in the main scanning direction of a pixel input from the main scanning magnification of the set magnification onto the output pixel. The main scanning grid point is obtained, and from this value, the main scanning projection coefficient of the projection coefficient in the main scanning direction, the main scanning projection address indicating the position of the main scanning grid point on the output pixel, and the main scanning direction of the input pixel The main scanning character address is generated based on the address of the output pixel in the main scanning direction at the position where the center is projected, and the grid point representing the position in the sub scanning direction of the pixel input from the sub scanning magnification of the set magnification is the output pixel. The sub-scanning grid point to be projected on is calculated from this value, the sub-scanning projection coefficient of the projection coefficient in the sub-scanning direction, the sub-scanning projection address indicating the position of the sub-scanning grid point on the output pixel, and the input pixel Center in sub-scanning direction Generating a sub-scan character address based on the sub-scanning direction of the output address of a location to be projected.

In the present invention, the main-scanning grid point is obtained from the main-scanning magnification, the main-scanning projection coefficient, the main-scanning projection address, and the main-scanning character address are obtained based on this, and the sub-scanning lattice point is found from the sub-scanning magnification. Since the sub-scanning projection coefficient, the sub-scanning projection address, and the sub-scanning character address are generated based on, the positions of the projection image data and the character image data represented by the projection address and the character address match, and the image deterioration is reduced. In addition, since the addresses can be generated independently in the main scanning direction and sub-scanning direction, smoothing interpolation data insertion and black thin line saving processing operate independently in the main scanning direction and sub-scanning direction, expanding in one direction and reducing in the other direction. Also in the case, the smoothing interpolation and the saving process of the black thin line are performed respectively, and the reproducibility of the character is improved.

According to a fourth aspect of the present invention, in the second aspect of the present invention, the projection processing means sets a white level as an initial level of an output image of the projection image memory, and when an input pixel is black,
The output pixel indicated by the projection address of that pixel is set to a multi-valued level according to the projection area calculated by the projection coefficient.

In the present invention, the projection processing means sets the initial level of the output pixel to the white level, and when the input pixel is the black level, the output pixel has a level corresponding to the projection area calculated by the projection coefficient. Black may be projected onto a single output pixel from a plurality of input pixels, and in this case, the multi-value level corresponds to the value obtained by integrating the projected areas of these black pixels. In this way, the input image of the binary image can be converted into the multi-valued image.

According to a fifth aspect of the present invention, in the first aspect of the present invention, the synthesizing means determines that the image area determination result is a character image,
When the output data of the character correction processing means is black, the composite multilevel output level is set to the minimum level, the image area discrimination result is a character image, and when the output data of the character correction processing means is white, the combination is performed. When the multi-value output level is the maximum level and the image area discrimination result is a halftone image, the projection image data of the projection processing means is output.

According to the present invention, when the binary image output from the character correction processing means and the multivalued image output from the projection processing means are combined, the white level of the binary image is defined as the maximum level of the multivalued image. By setting the black level of the value image to the minimum level of the multivalued image, the binary image and the multivalued image can be combined into the multivalued image.

According to a sixth aspect of the present invention, binary image data is input from the image memory, and the output pixel on which the center of the input pixel is projected during the enlargement process is used as the core pixel of the character, and the smoothing interpolation pixel for correction is provided around the pixel. If the main scanning character address of the smoothing interpolation pixel in the main scanning direction is the same as the main scanning character address of the previously written character core pixel, the main scanning is performed. When writing the smoothing interpolation pixel in the direction to the image memory is prohibited, and the sub-scanning character address of the smoothing interpolation pixel in the sub-scanning direction becomes the same as the sub-scanning character address of the previously written character core pixel, A character correction processing unit for prohibiting writing of the smoothing interpolation pixel in the sub-scanning direction into the image memory is provided.

In the present invention, when the main scanning character address of the smoothing interpolation pixel in the main scanning direction is the same as the main scanning character address of the previously written character core pixel, the smoothing interpolation pixel stores the output pixel. It is prohibited to write in the image memory. When the enlargement ratio is small, the smoothing interpolation pixel may overlap the previously written character core pixel. Since the character core pixel is a projection of the input pixel and the smoothing interpolation pixel is a pixel estimated based on the character core pixel, the image is improved by deleting the pixel and leaving the character core pixel. This also applies to smoothing interpolation pixels in the sub-scanning direction.

According to the seventh aspect of the invention, binary image data is input from the image memory, and the output pixel on which the center of the input pixel is projected is used as the character core pixel, and the previous pixel in the main scanning direction within the predetermined area of the input image is selected. When the reference pixel pattern including the pixel of interest and the thin line pattern match and the main scanning magnification is less than 1, the character core pixel corresponding to the pixel of interest this time is prohibited from being stored in the image memory, and the predetermined area of the input image is prohibited. If the reference pixel pattern including the previous target pixel in the sub-scanning direction and the thin line pattern match and the sub-scanning magnification is less than 1,
A character correction processing unit for prohibiting the storage of the pixel core pixel corresponding to the pixel of interest this time in the image memory is provided.

In the present invention, the previous pixel of interest of the input pixel corresponds to the thin line pattern to form a black thin line, and the character core pixel of the previous pixel of interest is black. Since the character core pixel of the pixel of interest this time is a reduction process, it may overlap with the previous core pixel. At this time, if the core pixel of the character this time is black, there is no problem, but if it is white, the thin black line is erased. For this reason, it is prohibited to write the character core pixel of this time to the image memory storing the output pixel. This prevents the black thin line from being erased during the reduction and prevents the deterioration of the image quality. This is true in both the main scanning direction and the sub scanning direction.

According to an eighth aspect of the present invention, in the fifth aspect of the present invention, a black level is output when the input data to the error diffusion processing means is the minimum level, and a white level is output when the input data is the maximum level. In other cases, a black-and-white forcing circuit that outputs the output data of the error diffusion processing means is provided.

According to the present invention, the minimum level of the input data is set to the black level and the maximum level is set to the white level before being processed by the error diffusion processing means, the binary image of the character image is reproduced, and the data of the other levels have an error. The diffusion processing means produces halftone binary data. The minimum level and the maximum level of the binarized data are obtained by converting the binarized data of the character image into multi-valued data by the synthesizing means. Data can be used.
If all the data are passed through the error diffusion processing means, white or black data may become halftone data, which is prevented to obtain a clear character image.

Hereinafter, embodiments of the present invention will be described with reference to the drawings. FIG. 1 is a block diagram showing an outline of an image processing apparatus according to an embodiment of the present invention. 1 is input of binary image data, 2 is an image memory for storing a plurality of lines of image input data, 3 is a predetermined area of 4 × 4 pixels read from the image memory 2, and a pixel of interest is a halftone (halftone) Is an image area separation circuit that determines whether each pixel is a character or a character and outputs the image area determination result in accordance with the timing of image processing. Image area determination that determines whether the image is a halftone or a character depending on the characteristics of a predetermined area. It is provided with a circuit 4, a discrimination data write control circuit 5 for controlling writing of the discrimination result by an address from a control circuit 15 described later, and a discrimination result memory 6 for storing the discrimination result.

Reference numeral 7 reads out 4 × 4 pixels of the image memory 2, refers to a predetermined area, performs smoothing processing at the time of enlargement setting, saves black thin lines at the time of reduction setting, and makes a projection processing circuit described later. A character correction processing circuit that outputs character correction processing result data in synchronization with each other, and a black thin line determination circuit 8 that determines whether a pixel in a predetermined area is a black thin line, and smoothing interpolation data from pixels in the predetermined area. The generated smoothing processing circuit 10, the character data writing control circuit 9 for controlling the writing of smoothing interpolation data according to the black thin line determination result and the character address from the control circuit 15, and the character data subjected to the smoothing process or the black thin line process. And a character image memory 11 for storing

A projection processing circuit 12 outputs the projection data of the target pixel of the input image to the output pixel of the input image according to the set magnification. The projection processing circuit 12 obtains the projection data of the target pixel to the output pixel from the target pixel and a projection coefficient to be described later. If there are a plurality of target pixels, the projection data calculation circuit 13 that calculates the integrated value of the projection values and calculates the multi-valued projection data from the area ratio of the projected black pixels, and the projection image that stores this projection data It comprises a memory 14. The control circuit 15 outputs the projection coefficient for calculating the projection area of the input pixel to the output pixel to the projection data calculation circuit 13 according to the magnification specified by the main scanning magnification input 16 and the sub-scanning magnification input 17, and the projection image memory 14
And a character address to the discrimination result memory 6 and the character image memory 11 are generated.

Reference numeral 18 is an image signal synthesizing circuit for synthesizing the multi-valued image data by the image area discrimination result data, the character image processing result data and the projection image data, and 19 is the image signal synthesizing circuit 1.
Error diffusion processing is performed on the multivalued image data synthesized by
An error diffusion processing circuit that outputs binary image data of bits, an image signal correction circuit 20 that corrects an input image signal with past error data, a binarization circuit 21 that binarizes this correction result, and this binary An error calculation circuit 22 that obtains an error from the conversion result and the correction result, an error distribution integrated circuit 23 that distributes this error data to subsequent processing pixels and obtains the error data of the current processing pixel, and an error memory 24 that stores the error data.
And Reference numeral 26 forcibly makes the output of the binarization circuit 21 black when the level of the error diffusion input data is the minimum level, forcibly whites it when it is the maximum level, and otherwise the output of the binarization circuit 21. Is a black-and-white compulsory circuit that outputs as is.

The operation of the apparatus configured as above will be described. 2 to 6 are operation flow charts of the present embodiment. 7 to 9 show the coordinates of the input pixel and the projection output pixel, and show the relationship when the magnifications in the main scanning direction and the sub scanning direction are the same, FIG. 7 is 1.8 times, and FIG. 8 is 1.2 times. FIG. 9 shows the case of 0.8 times. An example will be described with reference to FIG. 7. The grid shown by the broken line shows the input pixels, and the grid shown by the solid line shows the projection output pixels. Coordinates X and Y indicate the coordinates of the input pixel,
Coordinates AX and AY indicate the coordinates of the projection output pixel. AX is called a main scanning projection address, and AY is called a sub scanning projection address. In order to show how the input pixel is magnified 1.8 times and projected onto the projection output pixel, the input pixel is magnified 1.8 times and displayed. Considering the input pixels indicated by halftone dots,
It shows that one input pixel (Y = 2, X = 2) is projected on nine projection output pixels. Although FIG. 9 shows the case of reduction, an input pixel (Y = 3, X = 3) is projected on four projection output pixels. In the subsequent coordinate display, the Y coordinate and the X coordinate are displayed in the order of (Y, X).

The projection output pixel is expressed based on the grid point a of the input pixel. FIG. 10 is a diagram showing the relationship between input pixels and projection output pixels. In the present embodiment, the magnification of the projection output pixel with respect to the input pixel is explained as being up to 2 times, but it is possible to be 2 times or more. If the magnification is set to a maximum of 2, the maximum number of projection output pixels on which the input pixel is projected is 9, as shown in FIG. It should be noted that (AY, A
The (X) coordinate indicates the absolute coordinate, and is shown in FIG. 10 (AY, A
X) coordinates are relative coordinates based on the grid point a. This relative coordinate is also defined for each grid point, for example, grid point b, and the coordinates based on grid point a (AY (0), AX (2)) are based on grid point b (AY (2) ', AX (2 ) ') Represents the same coordinates. The input pixels projected on AY = 3 and A = 5 indicate that there are two input pixels that are +1 in the X direction in addition to the input pixels a and b. On the other hand, the center of the input pixel at the grid point a is projected (AY = 4, AX = 4)
The projection output pixel of the coordinate is projected from only one input pixel of the grid point a.

Coordinates AY and AX representing projection output pixels are called a projection address, AX is called a main scanning projection address, and AY is called a sub scanning projection address. KY and KX are called projection coefficients. The projection coefficient is represented by a ratio of the side lengths of the input pixel projected on each output pixel in the main scanning direction and the sub scanning direction divided by the side length of the output pixel. That is, it is the projected length when one side of the output pixel is 1. KX is called a main-scanning projection coefficient, and KY is called a sub-scanning projection coefficient. The projection coefficient is also relatively expressed based on the grid point a, and KY (2) to KY
(1), KX (2) to KX (0) are displayed. The projection factor is used to describe the area that an input pixel projects onto an output pixel. For example, the area where the input pixel of the grid point a is projected at the coordinates (AY = 5, AX = 5) is KY.
It is represented by (2) × KX (2). With this projected area,
The density (multivalue) of each projection output pixel is calculated as described later. Note that the projection coefficient is “0” for a projection address where the input pixel is not projected depending on the input pixel position and the magnification.

FIG. 11 is a diagram showing the relationship between input pixels and character addresses. The character address is composed of a character core address and a character interpolation address. The output image on which the center point c of the input pixel (target pixel) is projected is the character core data, and this address is the character core address CY (1), CX (1). Is expressed as relative coordinates based on the center point c of the input pixel.
The absolute coordinates of the character core data are the same as the (AY, AX) coordinates shown in FIGS. 7 to 9, and are displayed in (CY, CX) coordinates. In FIG. 11, as an example, (CY (1), CX).
(1)) Coordinates are represented by (CY = 4, CX = 5) coordinates.

The character core data is expressed as S (1,1), and the data above, to the left, and diagonally to the upper left of this S (1,1) are smoothed interpolation data S (0,1), S (1,).
0) and S (0,0). The character core address of S (1,1) is (CY (1), CX (1)) and S (0,
The character interpolation address of 1) is (CY (0), CX
(1)), S (1,0) character interpolation address is (CY
The character interpolation addresses of (1), CX (0)) and S (0,0) are (CY (0), CX (0)). The smoothing interpolation data is an interpolation pixel to be inserted in order to smooth the output image when the input pixel is enlarged. Character addresses are CY (1), CY (0), CX (1), CX
It consists of 4 types of (0), and by combining these, S
Four data items (1,1) to S (0,0) are displayed.
The Y direction is called the sub-scanning direction and the X direction is called the main scanning direction.

Next, the operation of the apparatus shown in FIG. 1 will be described with reference to FIGS. First, the operation of the control circuit 15 shown in FIG. 1 will be described. The control circuit 15 inputs the main scanning magnification MX and the sub-scanning magnification MY, and outputs a control signal LX indicating whether the enlargement or reduction is performed.
LY, projection addresses AX, AY, projection coefficients KX, KY,
Character addresses CX and CY are generated.

In FIG. 2, when the main scanning magnification MX and the sub-scanning magnification MY are input (S1), the control signals LX and LY are set according to the magnifications (S1 to S7). LX = 1 (S3) and LY = 1 (S6) when the magnification is 1 time or more, and LX = 0 (S4) and LY = when the magnification is less than 1 time.
It is 0 (S7).

First, generation of the projection address AY, the projection coefficient KY, and the character address CY in the sub-scanning direction will be described. The address of the input pixel of the image memory 2 is represented by (Y, X). The position of the grid point of the projection output pixel is represented by the projection counter value. As shown in FIG. 10, assuming that the length of one side of the output pixel in the X and Y directions is 1, the length of one side of the input pixel is MX, M.
Represented by Y. Letting XC be the value in the main scanning (X) direction and YC be the value in the sub-scanning (Y) direction of the counter that counts grid points, XC (n) = XC (n-1) + MX, YC
It is represented by (n) = YC (n-1) + MY. Here, n is the position of the nth grid point. Note that these are XC = X
It is expressed as C + MX.

Image data required for image processing is written in the image memory 2, and the stored data is rewritten every time image data of a new line is required. In this embodiment, at least 4 lines of data are written. First, Y = 0, YC = 0, and AY (2) = 0 are set as initial settings (S8). Referring to FIG. 10, let AY (2) at the previous grid point b be AYP. P in AYP, AXP, and CYP is a character representing the previous time. Sub-scanning projection counter value YC
A value obtained by rounding down the number after the decimal point is represented by INT (YC), and this value is represented by AY (2). INT is a symbol indicating that the number after the decimal point is rounded down to an integer value. AY
(1) represents an address smaller than AY (2) in the one-pixel sub-scanning direction, and AY (0) represents an address smaller than AY (2) in the two-pixel sub-scanning direction (S9). As a result, the sub-scanning projection address AY is generated.

Next, the sub-scanning projection coefficient KY will be described with reference to FIG. The value of the projection coefficient changes depending on the magnification.
First, when the current AY (2) value is the same as the previous AY (2) value (S10), this means that the previous grid point b and the current grid point a are the same as shown in FIG. 12 (A). This is the case when the image is projected on the projection output pixel. In this case, as shown in the figure, the projection coefficient is KY (2) = MY, and the other KY (1),
Since KY (0) does not exist, it becomes "0" (S11).
Next, when the previous AY (2) becomes the current AY (1) (S12), this means that the number of input pixels is 2 as shown in (B).
This is the case when it extends over two projection output pixels. In this case, KY (2) and KY (1) are KY as shown in (B).
(2) = YC-AY (2), KY (1) = MY-KY
Since it becomes (2) and KY (0) does not exist, it becomes "0" (S13). When the input pixel spans three projection output pixels, KY (2) = YC-A as shown in (C).
Y (2), KY (1) = 1, KY (0) = MY− (1+
KY (2)) (S14). Thereby, the sub-scanning projection coefficient is generated.

Next, the character address will be described. First, the previous character core address CY (1) is set to CYP (S
15). As shown in FIG. 11, since the character core address represents the center point c of the input pixel, it is represented by an integer value obtained by subtracting 1/2 of MY from the value of the sub-scanning projection counter YC representing the grid point a and discarding the decimal point. That is, CY (1) = INT (YC-M
Y / 2). The character interpolation address CY (0) is C
The address is obtained by subtracting 1 from Y (1) (S16). As a result, the sub-scanning character address is generated.

Main scan projection address AX (2), AX
(1), AX (0), main scanning projection coefficient KX (2), KX
(1), KX (0), main scanning core address CX (1),
The main scanning character interpolation address CX (0) is calculated in the same manner as the address in the sub scanning direction as shown in FIG. 3 (S17 to S).
25). Thereby, the main scanning projection address, the main scanning projection coefficient, and the main scanning character address are generated.

Next, using the main scanning image memory address X and the sub scanning image memory address Y, the image memory 2 to the main scanning 4 are scanned.
16 pixel data PD in the area of pixel x sub-scan 4 pixel
Read (00) to PD (3, 3). S26 of FIG.
S33 represents this image memory 2 reading operation. S29
PIN indicates the input data of the image data to the image memory 2. M represents the number of pixels in the main scanning direction, and N represents the number of pixels in the sub scanning direction.

Next, the operations of the projection data operation circuit 13 and the projection image memory 14 of the projection processing circuit 12 will be described. The circuit 12 obtains a projected area by using the projection coefficients KX and KY of an input image consisting of a binary image input from the image memory 2, and multivalued pixels are obtained according to the area ratio of the black area of the projected area to the projection output pixel. To generate. Since one output pixel receives a projection from a maximum of four input pixels, the projection area is calculated each time the projection is received, the projection area is input to the projection image memory 14, and it is read again to add the area of the next projection. Is repeated up to 4 times.

In the case of converting a binary image into a multi-valued image, in the present embodiment, as an example, the gradation is normalized to 64 gradations of 0 to 63, and white is set to 63 as a white level. Also, black is set to 0. A value corresponding to the projected area ratio is entered during this period. This will be described with reference to FIG. 10. When the input pixel indicated by the halftone dot is white, it becomes the center (AY (1), AX (1)).
The output pixel of the coordinates is 63 of white. If the halftone dot pixel is black, the output pixel of the (AY (1), AX (1)) coordinate is black 0.
Becomes Other projection output pixels, such as (AY (2),
The output pixel at the AX (2) coordinate is projected from the four input pixels. The projected area F1 projected from the halftone dot input pixel is K
Represented by Y (2) * KX (2), the other three projected areas F
2 to F4 are also represented by the product of the main scanning projection coefficient and the sub-scanning projection coefficient of the other input pixels, and the sum of the areas projected by the black input pixels among these four projected areas is the output pixel (AY (2 ), And AX (2)) in black. When this is normalized as described above, it becomes 63 (1-sum of black areas).

The operation of the projection processing circuit 12 will be described with reference to FIG. The pixel of interest of the 16 image data PD (0,0) to P (3,3) previously read from the image memory 2 is PD.
(2, 2). This pixel of interest is the data indicating the approximately center of the 16 pixels, and PD (1,1) may be used instead. If the pixel of interest is 1, it is determined to be a white pixel, and it is checked whether it is a white pixel (S34). If it is a white pixel, the coefficient PT is set to 0.
(S35), and for black pixels, PT = 63 (S3)
6). As shown in FIG. 10, since the input pixels are projected onto a maximum of 9 output pixels, the projection data calculation circuit 13 outputs the main scanning projection coefficient KX (0) output from the control circuit 15.
˜KX (2), sub-scanning projection coefficients KY (0) to KY (2)
For nine different combinations, the main scanning projection coefficient and the sub-scanning projection coefficient are multiplied to obtain the projection area. Since the projection coefficient is expressed as a ratio of one side of the output pixel,
This projected area is the projected area ratio of the input pixel to the output pixel. Steps S35 and S3 are added to this projected area ratio.
The PT obtained in 6 is multiplied.

Projection image data (output pixel) MS (AY) represented by projection addresses AY (N) and AX (M)
(N), AX (M)) are calculated as follows.
First, the projection image memory data MP (AY (N), AX) indicated by the projection addresses AY (N), AX (M)
The initial level of (M)) is set to the white level (63 in this example), and KY (showing the effect of black pixel projection on the output pixel represented by the projection addresses AY (N) and AX (M). N)
When * KX (M) * PT is calculated, this value is calculated as MP (AY
(N), AX (M)) to subtract the normalized density level (multi-valued data) MS (AY) of the output pixel.
(N), AX (M)) are obtained, and this value is stored again as the MP data in the same address of the projection image memory 14. As shown in FIG. 10, one output pixel receives projections from a maximum of four input pixels, and thus the above calculation is repeated each time to update the density level. Also, since one input pixel is projected onto a maximum of 9 output pixels, M and N are each changed in the range of 0 to 2 to generate the addresses of the 9 output pixels, and the number of pixels shown in step S39 is increased. A binarization operation is performed. (S37-S43). If the pixel of interest is at the white level and the projection coefficient is 0, KY
(N) * KX (M) * PT becomes 0, and the white level is maintained. In this way, the binary input pixel is multivalued into 64 gradations. In the present embodiment, the case has been described in which the magnification is up to 2 times and the normalization is 64 gradations, but these magnifications and gradations are examples and can be realized with arbitrary values.

Next, the operation of the character correction processing circuit 7 will be described. In the smoothing processing circuit 10, the image memory 2
16 pieces of image data PD (0,0) to PD (3,3) in an area of 4 pixels in the main scanning × 4 pixels in the sub-scan read from
11, the character core data S (1,1) on which the center of the target pixel (input pixel) is projected and the smoothing interpolation data S (0,1), S (1,
0) and S (0,0) are generated (S44 in FIG. 4). Since this interpolation data is interpolated between two character core data, it is used only at the time of enlargement. A known method is used to generate the interpolation data.

Similarly, the black fine line discrimination circuit 8 also inputs 16 pieces of image data PD (0,0) to PD (3,3), and performs pattern matching to determine the previous pixel of interest in the main scanning direction (this pixel of interest). BX = 1 when it is determined that the input pixel immediately before in the main scanning direction) is a thin line pattern, and BX when it does not match.
= 0 (S45), and when the previous pixel of interest in the sub-scanning direction (input pixel one pixel before in the sub-scanning direction of the current pixel of interest) is determined to be a thin line pattern, BY = 1, when they do not match B
Y = 0 is set (S46).

The character data write control circuit 9 interpolates the smoothing interpolation data at the time of enlargement between the character core data to obtain a smoothing output image. When the enlargement ratio is small, this interpolation data is already projected. It may overlap with the core data and overwrite it to erase the original character core data. The character core data is the data in which the center of the input pixel is projected, and is more important for maintaining the image quality than the interpolation data generated by estimation. Therefore, when overlapping in this way, overwriting the interpolation data is prohibited. .

Therefore, when LX = 1, that is, when the main scanning magnification is 1 or more and the previous core address CXP coincides with the smoothing interpolation address CX (0) of this time in the main scanning direction, the smoothing interpolation data S (1 , 0) is prohibited (S47), and if they do not match, the address CY
(1), CX (0) data KD (CY (1), CX
S (1,0) is written as (0) (S48). Similarly, when LY = 1 in the sub-scanning direction, that is, when the sub-scanning magnification is 1 or more and the previous core address CYP matches the current smoothing interpolation address CY (0), the smoothing interpolation data S (0, 1 ) Is prohibited (S4
9), if they do not match, addresses CY (0), CX (1)
S as the data KD (CY (0), CX (1)) of
Write (0, 1) (S50). Similarly, LX = 1
And LY = 1 and CXP ≠ CX (0) and CYP ≠ CY
Only in the case of (0), the data of KD (CY (0), CX (0)) is set to S (0,0) (S52), and in other cases, writing is prohibited (S51).

The character data write control circuit 9 prevents the thin black line from disappearing during reduction. FIG. 13 is a diagram for explaining a case where the black thin line is erased. The dashed grid represents input pixels and the solid lines represent output pixels. G1 and G2 represent input pixels, and the centers of both pixels are projected into the output pixel (character core data) G3. G1 is a pixel forming a black thin line. The thin black line is shown by a halftone dot.
The character core data G3 first becomes a black pixel due to the projection of G1, but when it is next subjected to the projection of G2, it is rewritten as a white pixel and a black thin line is missing. Therefore, at the time of reduction, when the addresses of the previous character core data and the current character core data match, writing of the current character core data is prohibited, and a black thin line is left.

S53 to S55 in FIG. 5 indicate the above-mentioned operation. When the main scanning magnification is 1 or less (LX = 0) and the previous main scanning input pixel matches the black fine line pattern (BX =
1), (S53), or the sub-scanning magnification is 1 or less (LY =
0) and when the previous sub-scanning input pixel matches the black fine line pattern (BY = 1) (S54), writing of the character core data of this time is prohibited, and steps S53 and S54 are performed.
Write the character core data S (1,1) to the character core addresses CY (1), CX (1) only when the above condition is not satisfied (S
55).

Next, the operation of the image area separation circuit 3 will be described. The image area separation circuit 4 reads 16 pieces of image data PD (0,0) to PD (3,3) from the image memory 2, counts the change points, and if there are more change points than a predetermined value, halftone, If the number is small, it is determined as a character. These are known techniques and are disclosed, for example, in JP-A-6-253140. In S56 of FIG. 5, determination is performed using this method,
In case of character discrimination, ZD = 1, and in case of halftone discrimination, ZD = 0.

The discrimination data write control circuit 5 discriminates the output pixel by using the discrimination result ZD discriminated by the image area discrimination circuit 4 for the input pixel. In this case, in principle, the character core data and the smoothing interpolation data around the character core data use the discriminant value ZD of the corresponding input pixels at the time of enlargement, but when the smoothing interpolation data overlaps with the previous character core data, it is determined. The writing of the value is prohibited, and the discriminant value of the previous character core data is used. For the character core data, when both the main scanning magnification and the sub-scanning magnification are 1 or more,
At the time of reduction, when the previous character core data and the current character core data have the same address and the previous time is determined to be the halftone, the determination result ZD of the current character core data is written. In this way, the determination result of the character core data is given priority, and the character determination result is given priority.

In FIG. 5, the main scanning magnification is 1 or more (LX
= 1), if the previous main scanning character core address (CXP) matches the current main scanning character interpolation address CX (0), it indicates whether the smoothing interpolation data S (1,0) is a character or a halftone. Discrimination data VD (CY (1), CX
It is prohibited to enter the discrimination result ZD in (0) (S5
7) Fill in only when they do not match (S58). When the sub-scanning magnification is 1 or more (LY = 1), the previous sub-scanning character core address (CYP) is the current sub-scanning character interpolation address CY.
If it matches with (0), smoothing interpolation data S
Discrimination data VD (CY (0), CX of (0, 1)
It is prohibited to enter the discrimination result ZD in (1)) (S5
9) Write the discrimination result ZD only when they do not match (S
60). The main scanning and sub-scanning magnifications are both 1 or more (LX =
1, LY = 1), the previous main scan character core address (CX
P) does not match the current main-scan character interpolation address CX (0), and the previous sub-scan character core address (CYP) does not match the current sub-scan character interpolation address CY (0). Discrimination result ZD to discrimination data VD (CY (0), CX (0)) of data S (0, 0)
Is written (S62), and in other cases, writing is prohibited (S61).

Further, when both the main scanning magnification and the sub-scanning magnification are 1 or more (LX = 1, LY = 1) (S63), or at the time of reduction, the previous and present character nucleus addresses match, and the previous character nucleus data is obtained. Discrimination result RD (CY (1), CX
Only when (1)) is halftone (S64), the discrimination data VD (CY) of the character core data S (1,1) of this time
The discrimination result ZD is written in (1), CX (1)) (S6).
5). In other cases, writing of the determination result is prohibited.

The control signals LX and LY representing the magnification generated in the control circuit 15 are supplied to the discrimination data writing control circuit 5 and the character data writing control circuit 9, and the character addresses CX and CY are the discrimination data writing control circuit 5 and the discrimination data writing control circuit 5. It is input to the discrimination result memory 6, the character data write control circuit 9 and the character image memory 11.

Next, description will be made with reference to FIG. Since the processing of one pixel in the main scanning direction of the input pixels is completed as described above, the operations of the projection processing circuit 12, the character correction processing circuit 7, the image area separation circuit 3, and the control circuit 15 are performed by the number of input pixels in the main scanning direction. Repeat (S66, S67). The projection image memory 14, character image memory 11,
As a result of the determination, the line in which the writing in the memory 6 is completed is determined, and the line data is sequentially determined from the respective memories in parallel by the projection addresses AY, AX and the character addresses CY, CX at the same timing. CD (n) and projection image data TD (n) are read. The white level is written as an initial state in the projection image memory 14 and the character image memory 11 that have been read out, and the halftone state is written in the determination result memory 6. If it is determined that there is no write end line, the read process is not performed and the process for the next line of the input image memory is performed.

When the reduction ratio is large, the sub-scanning projection address AY (2) by the input pixel of this line may be the same as the sub-scanning projection address AYP by the previous line input pixel (S68). That is, as shown in FIG. 13, the previous and current sub-scanning input pixels may be projected on the same output pixel. In this case, it is determined that there is no write end line and the input pixel of the next line is processed (S72, S).
73). If AYP and AY (2) are not equal (S
68), the sub-scanning addresses AY and CY of the projection image memory 14, the character image memory 11 and the discrimination result memory 6 are sequentially read from the main scanning addresses AX = 0 and CX = 0 of the line where the sub-scanning addresses AY and CY are equal to AYP, and the image signal synthesis circuit 18 is read. Input (S6
9). When the sub-scanning address AYP by the input pixel of the previous line and the address AY (1) immediately before the sub-scanning address AY (2) by the input pixel of the current line are the same (S70), the writing end line is set. If it is determined that there is no input pixel, the input pixel of the next line is processed (S72, S).
73). If they are not the same (S70), the line in which the sub-scanning addresses AY and CX of the projection image memory 14, the character image memory 11, and the discrimination result memory 6 are equal to AY (1) is A.
The data is sequentially read from X = 0 and CX = 0 and input to the image signal synthesizing circuit 18 (S71). In this way, all the sub-scanning input pixels are read in (S72, S73).

Projection image data TD read in parallel
If the image area discrimination result in the image signal synthesizing circuit 18 is a character and the character correction processing output data is black, the character image data CD and the image area discrimination result RD are the minimum "0" of the composite multilevel output level. When the image area discrimination result is a character and the character correction processing output data is white, the maximum level of the composite multilevel output level is "63", and when the image area discrimination result is a halftone, , The projection image data TD is output.

The multi-valued data synthesized by the image signal synthesizing circuit 18 is subjected to error diffusion processing by the error diffusion processing circuit 19 and 2
Value data is converted. The black-and-white forcing circuit 25 outputs a black level when the error diffusion processing input data is at the minimum level, outputs a white level when the error diffusion processing input data is at the maximum level, and outputs error diffusion processing data at other levels. The character information synthesized by this processing surely becomes white or black, so that black-and-white inversion which rarely occurs due to error propagation in the error diffusion processing does not occur in the character portion.

[0059]

As is apparent from the above description, according to the present invention, the smoothing process or the black thin line saving process for each direction with respect to the input image is selectively executed by the main scanning magnification and the sub-scanning magnification in the character correction processing circuit. Is
In the projection processing circuit, the projection data is calculated from the input image data and the projection coefficient in the main scanning direction and the sub scanning direction, which are obtained from the main scanning magnification and the sub scanning magnification, and the image area discrimination unit inputs the characteristics of the image in the input image area. Whether the image data is character data or halftone data is discriminated. If the image area discrimination result is a character to the image signal synthesizing circuit, character correction processing data is output. If it is halftone data, the projection image processing data is transmitted. Output. The image signal synthesizing circuit synthesizes both of these data, and the error diffusion processing unit performs error diffusion processing to obtain 2
Convert to value data. The black level is output when the error diffusion processing input data is at the minimum level, the white level is output when the maximum level, and the error diffusion processing result is output otherwise.

By the above processing, the halftone portion of the input image data is projected, and the local reflectance of the projected image matches the local reflectance of the input image regardless of the main scanning magnification and the sub scanning magnification. The gradation of the halftone image is maintained. On the other hand, the character portion is smoothed and interpolated during the enlargement process, and the black thin line storage process prevents the thin line from being lost during the reduction process. In addition, smoothing interpolation data insertion and black thin line saving processing can operate independently in the main scanning direction and sub-scanning direction, and smoothing interpolation and black thin line saving can be performed even when one direction is enlarged and the other direction is reduced. The quality of the character image is improved. Also, by controlling the projection image memory, the character image memory, and the discrimination result memory at the same timing, the positions of the projection image data and the character image data match at any magnification setting, and when the image area discrimination circuit makes an erroneous discrimination. However, it is possible to minimize image quality deterioration.

[Brief description of the drawings]

FIG. 1 is a block diagram illustrating a configuration of an embodiment.

FIG. 2 is an operation flowchart of a control circuit

FIG. 3 is an operation flowchart of a control circuit and image memory reading.

FIG. 4 is an operation flowchart of projection data calculation processing and character correction processing.

FIG. 5 is an operation flowchart of character correction processing and image area separation processing.

FIG. 6 is a flowchart of a read operation of a projection image memory, a character image memory, and a discrimination result memory.

FIG. 7 is a diagram showing a relationship between an input pixel and an output pixel having a main scanning and sub-scanning magnification of 1.8 times.

FIG. 8 is a diagram showing a relationship between an input pixel and an output pixel having a main scanning and sub-scanning magnification of 1.2 times.

FIG. 9 is a diagram showing a relationship between an input pixel and an output pixel having a main scanning and sub-scanning magnification of 0.8 times.

FIG. 10 is a diagram illustrating a projection address and a projection coefficient.

FIG. 11 is a diagram illustrating a character address, character core data, and smoothing interpolation data.

FIG. 12 is a diagram for explaining calculation of projection coefficients.

FIG. 13 is a diagram illustrating a black thin line saving process.

[Explanation of symbols]

 2 image memory 3 image area separation circuit 4 image area discrimination circuit 5 discrimination data writing control circuit 6 discrimination result memory 7 character correction processing circuit 8 black fine line discrimination circuit 9 character data writing control circuit 10 smoothing processing circuit 11 character image memory 12 projection processing Circuit 13 Projection data operation circuit 14 Projection image memory 15 Control circuit 18 Image signal composition circuit 19 Error diffusion processing circuit 25 Black and white compulsory circuit

Claims (8)

[Claims] 1. An image area discriminating means for discriminating a character image and a halftone image by inputting binary image data from an image memory and referring to a predetermined area of the image data, and a binary image from the image memory. A character correction processing means for inputting data, generating smoothing image data around an output pixel whose center of the input pixel is projected during enlargement processing and its surroundings, and performing saving processing of a black thin line projected on the output image during reduction When,
Control means for generating a projection coefficient indicating the ratio of the projection of the input pixel to the output pixel according to the set magnification of the output image with respect to the input image, and the projection coefficient and the binary image data input from the image memory are projected on the output pixel. In the case of a character image based on the output data of the image area discrimination means, the output data of the character correction processing means is input, and the projection processing means for generating multivalued projection image data according to the ratio of the input black pixels In the case of a halftone image, the output data of the projection processing means is input, and a combining means for combining these into a multi-valued image and an error diffusion processing means for performing error diffusion processing of the data output from this combining means are provided. Image processing device. 2. A discrimination result memory for storing a discrimination result of the image area discrimination means, a character image memory for storing output data of the character correction processing means, and a projection image memory for storing output data of the projection processing means. The control means generates a projection address of data to be stored in the projection image memory and also generates a character address corresponding to the projection address in the discrimination result memory and the character image memory. 1. The image processing device according to 1. 3. The control means obtains a main scanning grid point at which a grid point representing a position in the main scanning direction of a pixel, which is input from a main scanning magnification of a set magnification, is projected on an output pixel, and the projection point is calculated from this value. Main scanning projection coefficient in the main scanning direction, main scanning projection address indicating the position of the main scanning grid point on the output pixel, and output in the main scanning direction at the position where the center of the input pixel in the main scanning direction is projected. The main scanning character address based on the pixel address is generated, and the grid point representing the position in the sub scanning direction of the input pixel is obtained from the set sub scanning magnification, and the sub scanning grid point on which the output pixel is projected is obtained. From the value, the sub-scanning projection coefficient of the projection coefficient in the sub-scanning direction, the sub-scanning projection address indicating the position of the sub-scanning grid point on the output pixel, and the sub-position of the position where the center of the input pixel in the sub-scanning direction is projected. To output address in scanning direction 3. The image processing apparatus according to claim 2, wherein the sub-scanning character address is generated based on the sub-scanning character address. 4. The projection processing means sets a white level as an initial level of an output image of the projection image memory, and when an input pixel is black, it is calculated by a projection coefficient for an output pixel indicated by a projection address of the pixel. The image processing apparatus according to claim 2, wherein the multilevel value is set according to the projected area. 5. When the image area discrimination result is a character image and the output data of the character correction processing means is black, the synthesizing means sets the composite multilevel output level to the minimum level, and the image area discrimination result is If it is a character image and the output data of the character correction processing means is white, the composite multilevel output level is set to the maximum level, and if the image area discrimination result is a halftone image, the projection image data of the projection processing means. The image processing apparatus according to claim 1, wherein the image processing apparatus outputs. 6. Binary image data is input from an image memory, an output pixel having a center of an input pixel projected during enlargement processing is used as a character core pixel, and a smoothing interpolation pixel for correction is provided around the pixel. Smoothing interpolation in the main scanning direction when the pixel address is the character address and the main scanning character address of the pixel is the same as the main scanning character address of the previously written character core pixel. If the sub-scanning character address of the smoothing interpolation pixel in the sub-scanning direction is the same as the sub-scanning character address of the previously written character core pixel, the writing of the pixel to the image memory is prohibited. An image processing apparatus comprising character correction processing means for inhibiting writing of smoothing interpolation pixels to an image memory. 7. A reference including binary image data input from an image memory, an output pixel having a center of an input pixel projected as a core pixel of a character, and a previous pixel of interest in a main scanning direction within a predetermined area of the input image. When the pixel pattern and the thin line pattern match and the main scanning magnification is less than 1, the storage of the character core pixel corresponding to the pixel of interest this time in the image memory is prohibited, and the sub scanning direction within the predetermined area of the input image is prohibited. The reference pixel pattern including the previous pixel of interest and the thin line pattern match,
Further, when the sub-scanning magnification is less than 1, an image processing device provided with a character correction processing means for prohibiting the storage of the pixel core pixel corresponding to the pixel of interest this time in the image memory. 8. The black level is output when the input data to the error diffusion processing means is the minimum level, the white level is output when the input data is the maximum level, and the output data of the error diffusion processing means is output otherwise. The image processing apparatus according to claim 5, further comprising a black-and-white forcing circuit for outputting.