JP2920635B2 - Image processing device - Google Patents

Description

Description: BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image processing apparatus, and particularly to an image processing apparatus suitable for obtaining a high-quality reproduced image with a simple configuration.

[Conventional technology]

2. Description of the Related Art Conventionally, an image processing apparatus performs image processing such as enlargement, reduction, rotation, density conversion, and conversion of a black-and-white pixel pattern on an image including density information represented by binary data of white and black to perform binary processing. Has been obtained.

First, a conventional method will be described by taking a linear density conversion in such an image processing apparatus as an example. Now, the original image is 2/3
Considering the case where the image is reduced to
It will be allocated at 3/2 pixel intervals. FIG. 2 shows the positional relationship between the pixels of the original image and the reproduced image. Here, each pixel of the original image is P (x, y), and each pixel of the reproduced image is Q (x, y).
y).

In this method, Q (0,0) and Q (2,0) are the values of the original image values P (0,0) and P (3,0), respectively.
The value of a pixel located between pixels of the original image such as (1,1) is determined using an interpolation process such as a logical sum method or a nearest neighbor method. Related literature, for example, Masashima
Hiroshi: "Performance evaluation and improvement of processing speed of various scaling methods for binary images", Transactions of Information Processing Society of Japan Vol.26 No5
(September 1985) Pages 920 to 925. Here, the logical sum method is a method in which four pixels P (m, n) surrounding Q (x, y) are observed, and if there is a black pixel among them, Q (x, y) is also determined to be black. The nearest neighbor method is a method in which the value of the closest pixel among the four pixels is used as the value of Q (x, y).

Similarly, interpolation processing such as a logical sum method or a nearest neighbor method has been used for image rotation processing.

On the other hand, the following known examples exist as enlargement / reduction processing methods for organized dither images (Kim et al .: A study of dithered image enlargement / reduction, 1987 IEICE Information and Systems Division National Convention) Proceedings (1987.11) Vol.1
94). In this example, a multi-valued image is estimated by inversely transforming the dither image using the dither matrix used when generating the pseudo halftone image. Then, the reproduced image is obtained by enlarging / reducing the estimated multi-valued image and binarizing it again by the systematic dither method.

Next, processing for converting the density of an image will be described. Conventionally, for a multi-valued image, there is an example in which a specific non-linear process is executed on input multi-valued data in order to correct nonlinearity of an input / output device. For example,
JP-A-61-123275 can be mentioned. But once 2
There is no effective means for the binarized image.

[Problems to be solved by the invention]

2. Description of the Related Art Conventionally, devices that handle binary image data target images originally represented by binary, such as characters and line figures. Therefore, the conventional image processing such as enlargement, reduction, rotation, and the like also aims at processing a normal binary image such as a character or a figure, and does not consider the case of processing a pseudo halftone image. Therefore, in the above-mentioned OR method, when white and black pixels are frequently switched as in a pseudo halftone image, most of the image is painted black. Also, in the nearest neighbor method, a drop is noticeable at the time of reduction, and the smoothness of the image is deteriorated at the time of enlargement.

In addition, the method of estimating a multi-valued image by inversely transforming a dither image requires information of a dither matrix used when generating the dither image, and also causes a difference between the original multi-valued image and an estimated value. The target halftone image is limited to only an image generated by the systematic dither method, and is not considered for an image in which line figures such as characters are mixed.

On the other hand, with respect to the process of converting the density of an image, the conventional method targets a multivalued image and cannot be applied to an image after the binarization process has been executed.

In addition, a halftone image may have a different appropriate binarization method depending on the output device. For example, many devices such as an LBP (laser beam printer) cannot perform isolated printing of one pixel. When outputting an image in which a large number of isolated white or black pixels exist in one-pixel units, such as a pseudo halftone image generated by a Bayer-type dither matrix, to this apparatus, many black pixels are missing or blurred. . To date, there has been no effective countermeasure.

An object of the present invention is to solve such a conventional problem and to simplify a simple halftone image generated by various pseudo halftone processes as well as a normal binary image obtained by binarizing a line figure. Can perform conversion processing such as enlargement, reduction, and rotation of an arbitrary conversion rate, and can also execute density conversion processing for pseudo halftone images and change black-and-white patterns according to the output device, if necessary. It is another object of the present invention to provide an image processing apparatus capable of obtaining a high-quality reproduced image.

[Means for solving the problem]

In order to solve the above object, an image processing apparatus according to the present invention includes:
An image storage means for temporarily storing binary digital image data P, and M pixels (M
≧ 1) a means for scanning using a scanning window, means for restoring density data Sp of each pixel from binary image data of M pixels in the scanning window, and the restored density Performs image processing on data Sp and multi-valued data
Means for obtaining Sq, and means for re-binarizing the multi-valued data Sq to determine binary image data Q. The means for restoring density data Sp depends on the position of each pixel in the scanning window. Has a plurality of types of weighting coefficients, and has means for selecting an asymmetrical one of the weighting coefficients with respect to the linear portion of the density data Sp, and means for performing image processing to obtain multi-valued data Sq. Means for multiplying the value of the restored image data of several pixels at the position surrounding the image data by a reciprocal of the distance to each pixel as a weight and determining the converted image data from the sum of the pixels; Means for restoring density data Sp of each pixel from binary image data of M pixels, wherein a ROM having a plurality of address lines is used as a density data restoring unit, and a switching signal is converted to a plurality of binary data or multi-value data. By doing, three or more weights And means for performing image processing on the restored density data Sp to obtain multi-value data Sq, and means for re-binarizing the multi-value data Sq to determine binary image data Q. Has concentration data Sp
Means has a plurality of types of weighting factors according to the position of each pixel in the scanning window, has means for selecting one of the various types of weighting factors, performs image processing, and converts the multi-valued data Sq The obtaining means has means for multiplying the value of the restored image data of several pixels at the position surrounding the converted image data by a reciprocal of the distance to each pixel as a weight, and determining the converted image data from the sum value of each pixel. Means for restoring the density data Sp of each pixel from the binary image data of M pixels in the scanning window, means for performing image processing on the restored density data Sp to obtain multi-value data Sq, Means for re-binarizing the multi-valued data Sq to determine binary image data Q;
The means for restoring has a plurality of types of weighting factors according to the position of each pixel in the scanning window, and selects one of the various weighting factors based on an external instruction or a determination result based on image characteristics. Restoring means for restoring the density data Sp of each pixel, arbitrarily determines a range to be referred to when restoring the density data, prepares a plurality of matrixes of weight coefficients, and uses a switching signal according to an image to be processed. One of the above matrices
It is characterized by switching to one.

[Action]

In a binary image of a line figure such as a character, information is expressed by white and black patterns, and the position of each point itself is information. Therefore, in the present invention, the binary image of the line figure is subjected to the affine transformation as it is.

However, in contrast, the binary image generated by the pseudo halftone is obtained by converting the density into the density of black pixels. Therefore, in a pseudo halftone image, information is represented not by a black and white pattern of individual pixels but by the density of black pixels in a certain area. It is not appropriate to perform processing such as affine transformation on individual pixels of the pseudo halftone image in the same manner as in a normal binary image.

Therefore, it is necessary to handle a pseudo halftone image represented by binary data as density information instead of pattern information. However, the method of estimating a multi-valued image by the inverse transformation shown in the conventional example requires information on the dither matrix used when generating the pseudo halftone image. Also, the difference between the original multi-valued image and the estimated value is large.

Therefore, in the present invention, the density of each pixel of the pseudo halftone image is restored in the same manner as the mechanism in which a human senses the shade from the pseudo halftone image.

Humans perceive shading from a pseudo halftone image without being conscious of the size of the dither matrix. In this case, the density of a certain point on the pseudo halftone image is visually perceived by combining the values of a plurality of pixels around the point. Therefore, when determining the density of a certain pixel, the contribution of the density may be determined according to the distance between the point of interest and each of the surrounding pixels. Specifically, for each pixel, the black and white distribution of surrounding pixels is checked, weighted according to the distance between each pixel and the point of interest, and arithmetic processing is performed to determine the density.

As a result, it is possible to obtain a multi-valued image close to human vision regardless of the method of the pseudo halftone processing.

Once the grayscale information of the image is restored, it is possible to affine-transform the grayscale information.

Further, since the multi-valued gray image is restored, the gray image can be binarized by a method suitable for each output device. Therefore, it is also possible to convert the monochrome pattern of the pseudo halftone image according to the output device.

〔Example〕

Hereinafter, an embodiment of the present invention will be described in detail with reference to the drawings.

First, the principle of the present invention will be described with reference to FIGS. Now, for example, for two values of the original image of M _p pixels * N _p pixels, the case of obtaining a reproduced image of the M _q * N _q pixels perform image processing, the present invention performs the following processes 3 steps.

(1) for each point M _p * N _p 2 value data P of pixels, to restore the density data by referring to the distribution of black pixels around. As a result, with the density information of M _p * N _p pixels, the restored image data S _p is obtained.

(2) larger relative to the restored M _p * N _p restored image data S _p of the pixel, reduction, rotation, or performs conversion such necessary image processing of the density, the density information of M _q * N _q pixels To obtain converted image data _Sq .

(3) by binarizing by an M _q * N _q converted image data S _q of pixels halftoning, obtaining M _q * N _q reproduced image data Q binary pixels. The method of the pseudo halftone processing can be switched according to the output device and the image.

First, the first step (1) of the present invention will be described.

The process (1) is for restoring density information for each point of a binary image with reference to neighboring binary data. As an example, a case where the density information of each pixel is restored from the binary image shown in FIG. In FIG. 3 (a), each small rectangle represents each pixel, of which the shaded portions represent black pixels.

Here, the restoration of the density of the coordinates (x ₁ , y ₁ ) in FIG. 3 (a) is basically performed by rewriting the values in the m * n pixels centered on the pixel (x ₁ , y ₁ ). It can be obtained by weighting and adding according to the distance from the center. For example,
m = 3, n = 3, then each weighted value of 9 pixels surrounded by a dotted line in FIG. 3 (a) and, S _p by adding it
(X ₁ , y ₁ ) is determined. An example of the weight coefficient α is shown in FIG. When using this weighting factor, the possible values of concentration S _p of each pixel becomes 17 steps from 0 to 16, obtained by the following equation.

However, it is determined as x ₃ = x ₁ −1 y ₃ = y ₁ −1.

FIG. 3 (b) shows the result of restoring the density information of each pixel in this manner with respect to the binary data of FIG. 3 (a). The weighting coefficient α used in the restoration of the density can be asymmetrical in the vertical and horizontal directions. This is particularly effective when the processing target includes a line graphic.

By the processing of the present invention described above, an image having density information can be restored from a pseudo halftone image regardless of the binarization method of the image.

Next, the process (2) which is the second stage of the present invention will be described.

This is a stage in which various types of image processing are executed using the obtained density information. The processing here conventionally performed on multi-valued image data can be applied as it is.

As an example, a case where a 2/3 reduction process is executed will be described. In this case, the density information S _p underlying, the positional relationship of each pixel of the density information S _q multivalued obtained as a result of processing the fifth
As shown in the figure. Hereinafter, the image that is the basis of the image processing will be referred to as a restored image, and its data _Sp will be referred to as restored image data. Further, an image obtained as a result of the processing is called a converted image, and its data _Sq is called converted image data. Here, the starting points S _p (0,0) and S _q (0,0) of both images are at the same position. 2 /
3, the converted image data S _q (x
₂ −1, y ₂ −1) and S _q (x ₂ −1, y ₂ +1) are respectively S _p (x ₁ −1, y ₁ −1) and S _p (x ₁ −1) of the restored image data. , y ₁ +
Corresponds to 2). However, a value such as S _q (x ₂ , y ₂ ) located between pixels of the restored image data needs to be determined from peripheral restored image data. As an example of a method of determining the converted image data _Sq (x, y), a distance inverse proportional method will be described. In this method, the value of the restored image data of several pixels at the position surrounding the converted image data S _q (x, y) is multiplied by the reciprocal of the distance to each pixel as a weight, and the converted image is calculated from the sum of the pixels. The data S _q (x, y) is determined. For example, S _q (x
₂ , y ₂ ) is determined by the following equation.

_{S q (x 2, y 2} ) = 1/4 ((β 1 S p (x 1, y 1) + β 2 S p (x 1 + 1, y 1) + β 3 S p (x 1, y 1 +1) + Β ₄ S _p (x ₁ +1, y ₁ +1)) where the coefficient β is the distance Δl ₁ , Δl ₂ ,
It is a weight coefficient determined by the reciprocal of Δl ₃ and Δl ₄ . In this case, since l ₁ = l ₂ = l ₃ = l ₄ , β ₁ = β ₂ = β ₃ = β
₄ = 1. By performing this process, the multivalued restored image shown in FIG. 6 (a) is subjected to a 2/3 reduction process, and the resulting multivalued converted image is shown in FIG. 6 (b).

On the other hand, the density conversion processing can be realized by linearly or non-linearly converting the density for each pixel. That is,
In this process _{(2), M p = M} q and _{N p = N q S q (} x, y) = f {S p (x, y)} executes processing that is.

Here, the function f (x) is a linear or non-linear function, for _{example, f (x) S max *} (x / S max) Γ _{However, S max ≧ S p (x} , y) and the like.

As a result, according to the present invention, density conversion, which has been applied only to multi-valued images, for example, Δ correction, can be applied to a pseudo halftone image that has already been binarized. .

Finally, the process (3) corresponding to the third stage of the present invention will be described. This is a process of determining binary reproduced image data Q from the converted image data _Sq . Here, the principle will be described for one example. In the present invention, various types of binarization processing methods conventionally used can be applied as the processing of this portion. Therefore, a plurality of binarization processing methods may be provided in the apparatus, and switching may be performed according to a target image or an output apparatus.

This method is a method in which binary data is determined such that the cumulative sum of rounding errors generated when binarizing converted image data is always minimized. FIG. 7A shows an example of converted image data S _q (x, y) of 17 gradations of density 0 to 16. When binarizing the pixels in the fixed threshold _{T, S q (x, y} ) = 0 or S _q
When (x, y) is other than 16, a rounding error ε occurs. The rounding error epsilon, the maximum possible value of the converted image S _q When S _max,
It is obtained by the following equation.

_{ε = S q (x, y} ): S q (x, y) <T ε = S q (x, y) -S max: is either _{S q (x, y) ≧} T. Therefore, by extracting this error ε and adding it to the converted image data of the pixel that has not been subjected to the binarization processing, the total error can be minimized. For example, when each pixel in FIG. 7A is binarized at a threshold value T = 8, the pixel at coordinates (1, 1) is S _q (1, 1) = 7 and 7 ≦ 8 (=
T), the binary data Q (1,1) = 0, and a rounding error ε = 7 occurs. Therefore, this ε is added to the converted image data S _q (2, 1) when binarizing the next pixel (2, 1).
Then, _Sq (2,1) = 5 originally, but becomes 12 by adding the error 7. Therefore, it is determined that Q (2,1) = 1. That is, if a pixel having a density of 7 is set to white, the rounding error is added as the density of the next pixel, so that the next pixel is easily determined to be black. Therefore, by repeating this process, the density can be converted to the density of black pixels by dispersing the rounding error caused by the binarization.
FIG. 7 (b) shows the result of performing this processing on the whole of FIG. 7 (a). Note that an error generated by the binarization can be two-dimensionally dispersed in converted image data of a plurality of neighboring pixels. Specifically, for example, the error 7 generated at the pixel number (1,1) in FIG. 7 (a) is replaced with the neighboring (2,1) (3,
1) It is a method to distribute to (1,2) (2,2) and so on. In this case, although the processing is slightly complicated, a higher-quality reproduced image can be obtained.

Through the above processing, the rounding error caused by binarization can always be minimized while converting the density into the density of black pixels.

On the other hand, the resolution of a line graphic such as a character is reduced when pseudo halftone processing is performed during the binarization processing. In this case, a binarization process using a fixed threshold may be performed as the process (3). Specifically, the threshold value T is defined as T = _Smax / 2, and Q (x, y) = 1: T ≦ _Sq (x, y) Q (x, y) = 0: T> _Sq (x , y) to determine the reproduced image Q (x, y).

On the other hand, in the process (3), by executing the binarization process according to the output device, the image quality can be improved by converting the black and white pattern. For example, in the case of an output device that cannot execute the printing of one pixel, the image quality of the output image can be improved by executing the pseudo halftone processing by the systematic dither method using the centralized dither matrix in which isolated pixels are less likely to occur. .

By executing the above processes (1), (2) and (3), it is possible to perform image processing on the original image p (x, y) to obtain a high-quality reproduced image Q (x, y). it can.

As described above, according to the present invention, it is possible to perform image processing for obtaining a high-quality reproduced image with respect to any binary image including a pseudo halftone image. Further, conversion of density, conversion of a black and white pattern, and the like, which could not be performed on a pseudo halftone image, can be realized.

FIG. 1 is a basic configuration diagram of an image processing apparatus showing one embodiment of the present invention.

In FIG. 1, 101 is a signal line for inputting binary original image data, and 100 is an original image data temporary storage unit for storing the original image data for several scanning lines. Reference numeral 200 denotes a density data restoring unit that obtains restored image data representing density information of each image from the original image data in the original image data temporary storage unit 100;
Denotes an image processing unit that performs necessary image processing such as enlargement, reduction, or density conversion on the restored image data and obtains converted image data. 400 denotes a unit that binarizes the converted image data output from the image processing unit 300 It is a value processing unit. The binarized data is output from a signal line 491 as reproduced image data. That is, the original image data temporary storage unit 100 and the density data restoring unit 200 execute the process of restoring multi-valued density data from the binary original image data (the process (1) described in the above-described principle). The necessary image processing (the processing (2) described in the above principle) is executed by the processing unit 300, and
The binarization processing unit 400 executes binarization processing of the converted image data (the processing (3) described in the above-described principle). Further, the entire operation is controlled by the control unit 500.

Next, each part of the apparatus will be described in detail. First, the operation of the portion for restoring the restored image data _Sp (x, y) indicating the density from the binary original image data P (x, y) will be described. 8th
The figure is a block diagram of a processing part for obtaining multi-value data _Sp (x, y) by this method. This portion corresponds to the original image data temporary storage unit 100 and the density data restoration unit 200 in FIG. The number of pixels to be referred to when restoring the density data is arbitrary, but here, an example will be described in which binary data in a range of 3 pixels * 3 pixels is referred to. Restored image data S
_When calculating _p (x ₁ , y ₁ ), the binary data to be referred to is the third
As shown in FIG. 7A, the data is 9-pixel data centered on the coordinates (x ₁ , y ₁ ) of the original image. Now, when the original image data P (x ₁ +1 and y ₁ +1) is input from the signal line 101, P (x ₁ , y ₁ +
1), P (x ₁ −1, y ₁ +1) are recorded. From the three output lines 131, the binary data for the two pixels and the original image data P (x ₁ +1 and y ₁ +1) are output. The line buffers 110 and 115 output the previous data one line at a time, and the three signal lines 132 and 133 output binary data for three pixels to the density data restoration unit 200. Here, the signal line bundle 130 is a total of 9 signal lines, 9 pixels centered on the coordinates (x ₁ , y ₁ ),
That coordinates (x ₁ -1, y _1-1) data in the range from _{(x 1 + 1, y 1} +1) is output, respectively. As the data of the nine pixels is shown in FIG. 4, the data for determining the restored image data _{S p (x 1, y 1} ). Therefore, the data output from the signal line 130 is input to the density data restoration unit 200.
Density data recovery unit 200 and the binary data of nine pixels which are input from the signal line 130, 3 * 3 restored by the weighting factor of the pixel α image data _{S p (x 1, y 1} ) values of the determined output This is the part to do. Here, the binary data of 9 pixels input from the signal line 130 is multiplied by a weight coefficient α for each pixel by the multiplier 210, and then added by the adder 220.
It is output as restored image data _{S p (x 1, y 1} ). Also,
If the signal line 130 is an address line,
It can also be realized with a single ROM (Read Only Memory). The capacity of this case, ROM is a 4bit * 2 ^9. In addition,
The range to be referred to in order to obtain the restored image data is arbitrary as described above.
As shown in the figure, a parallelogram corresponding to θ can be used.

In addition, by preparing a plurality of weight coefficient matrices and switching according to the image to be processed, the image quality of the output image can be further improved. For example, when processing a line figure such as a character as an image, for example, as shown in FIG. 10, the weight of the central pixel is increased. as a result,
Deterioration of the resolution of the output image can be prevented. And
When processing a pseudo halftone image, the weighting factors shown in FIG. 4 are used as described above.

An example for realizing this method will be described with reference to FIG. Now, a case where two types of weighting factors are used will be described for simplicity. The density data restoring unit 200 is configured by a ROM having ten address lines. Nine of the ten address lines, as in the case of FIG.
Binary image data for nine pixels to be referred to is input from the signal line bundle 130. Then, the signal line 139 is connected to the remaining one of the address lines, and the switching signal FLG is input. Here, the switching signal FLG is instructed by the external switch 230 as follows.

FLG = 1: Line figure image FLG = 0: Pseudo halftone image Also, for example, by using the output of the image determination unit 240 that performs real-time area determination of an image by a known method as the switching signal FLG, The weighting factor can be automatically switched accordingly. The configuration diagram in this case is
This is shown in FIG. The switching signal FLG can be determined according to not only the type of the image to be processed, but also the density of the image, the characteristics of the output device, and the like.

Further, a ROM having 11 or more address lines is used as the density data restoring unit 200, and the switching signal FLG is transmitted to a plurality of 2
By using value data or multi-value data, three or more types of weighting factors can be switched.

On the other hand, as the weighting factors, values symmetrical in the vertical and horizontal directions are usually used as shown in FIGS. However, depending on the image, a more desirable output image may be obtained by using an asymmetric value as shown in FIG. 12, for example. For example, when density restoration is performed on the straight line portion shown in FIG. 13A using the weighting coefficients shown in FIG. 4, the resulting restored density image becomes FIG. 13B. Here, coordinates x = x ₁ -1 and x = x ₁ +1 in the figure
The density of the pixel at the position is equal. Therefore, when the restored density image is binarized with a specific threshold value, one or three pixels become 1, and a line having a width of two pixels cannot be output. On the other hand, when the weighting coefficients shown in FIG. 12 are used, the obtained restored density image is as shown in FIG. 13 (c). By binarizing this image using an appropriate threshold, it is possible to output a line having an arbitrary width of one to three images. The threshold used here can be determined from the weight coefficient.

Next, as an example of the operation of the image processing unit 300 and the control unit 500,
First, for the restored image of X ₁ pixel * Y ₁ pixel, the x direction is represented by r ₁ / R
_1, the y-direction is converted by the reduction ratio r ₂ / R _2, will be described an example in which to perform a reduction process to obtain a converted image of X ₂ pixels * Y ₂ pixels. In processing such as enlargement, reduction, and rotation, the present apparatus can apply the same method as various kinds of interpolation processing on multi-valued image data. FIG. 14 shows a configuration example of a processing apparatus to which the inverse distance proportional method is applied as an example of the interpolation processing. Let x direction be r ₁ / R ₁ , y
In the reduction process of the direction and r ₂ / R _2, the restored image data S
If the positional relationship _p and the converted image data S ₂ is FIG. 5, the coordinates of the restored image data corresponding to the coordinates of the converted image data _{(x 2, y 2) (} x 1, y 1) determined by the following formula Can be

x ₁ = [x ₂ * R ₁ / r ₁ ] y ₁ = [y ₂ * R ₂ / r ₂ ] Here, [x] represents an integer not exceeding x. Also, in normal processing, the reduction ratios r ₁ / R ₁ and r ₂ / R ₂ are often r ₁ / R ₁ = r ₂ / R ₂ .

When determining the converted image data S _q (x ₂ , y ₂ ), the restored image data to be referred to is data _Sp (x ₁ , y ₁ ) for four images surrounding S _q (x ₂ , y ₂ ). S _p (x ₁ +1, y ₁ ), S _p (x ₁ , y ₁ +1), S _p
(X ₁ +1 and y ₁ +1). Hereinafter, the operation of each unit will be described with reference to FIG. If the restored image data S _p (x ₁ +1, y ₁ +1) is now input from the signal line 301, the restored image data S _p (x ₁ , y ₁ +1) is input from the latch 311 and the restored image data S _p (x ₁ , y ₁ +1) is input from the line buffer 320. S _p (x ₁ +1 and y ₁ )
_p (x ₁ , y ₁ ) is output. The multi-valued data for four pixels is multiplied by a weighting coefficient α by multipliers 321 to 324, respectively, and sent to an adder 330. The adder 330 adds the input multi-valued data of four pixels and converts the converted image data S
Outputs _q (x ₂ , y ₂ ). Here the weighting factor α is determined by the positional relationship of the restored image data S _p and the converted image data S _q,
This is a 2 * 2 determinant. Each term of the matrix is determined by the weight coefficient determination unit 340. In the inverse distance method,
The coefficient is determined in proportion to the reciprocal of Δl ₁ , Δl ₂ , Δl ₃ , Δl ₄ in FIG. In this apparatus, since the interval between input pixels is always constant, each coefficient can be determined by Δx and Δy in FIG. Δx and Δy are obtained in the control unit 500, and the signal line 34
The weight coefficient is input to the weight coefficient determination unit 340 by 1,342. Therefore, the weight coefficient determination unit 340 can be realized by one ROM using the signal lines 341 and 342 as address lines. Further, the restored image data S _p and [Delta] x, if all Δy address lines, the entire image processing unit 300 employed in the reduction process of one 1 R
It is also possible to configure with OM.

Here, the operation timing and Δx, Δy of each unit are determined by the control unit 50.
Determined by 0.

Therefore, next, the operation of the control unit 500 will be described with reference to FIG. In FIG. 15, reference numeral 510 denotes a clock for outputting a reference pulse, and reference numerals 520 and 530 denote address counters for counting the addresses x ₂ and y ₂ of the output multilevel data to be output. On the other hand, the address x ₁ , y of the input multi-valued data
₁ is counted by the address counters 540 and 550, respectively. Now, an example will be described in which the x direction is converted by r ₁ / R ₁ and the y direction is converted by r ₂ / R ₂ . In this case, the input of the input multi-value data is performed once a reference pulse r ₁ times, determination of the output multi-level data is performed once at _a time R. Therefore, the address counters 520 and 540 operate once _per reference pulse r ₁ and R ₁ respectively. Each time the input multi-valued data for one scanning line is read, the address of the output multi-valued data in the sub-scanning line direction is set to R _2.
/ r ₂ needs to be advanced. Therefore, the address counter 520
When processing for one scanning line is completed, the end detection unit 515 outputs r ₂
The pulse signal is output once. Hereinafter, the pulse signal output from the end detection unit 515 is referred to as a line pulse. The address counters 530 and 550 operate once every _two line pulses r ₂ and R ₂ , respectively. However, in order to determine the output multi-level data _{S q (x 2, y 2} ) , it is necessary to input multivalue data _{S p (x 1 + 1,} y 1 +1) has already been entered. Therefore, it is necessary to read the input multi-valued data one pixel before the output of the address counter 520 and one line before the output of the address counter 530. Further, in order to perform processing at an arbitrary conversion rate, it is necessary that all of r ₁ , R ₁ , r ₂ , and R ₂ can be externally designated. However, in actual use, there is no problem even if r ₁ and r ₂ are set to relatively large fixed values. Therefore, in the present embodiment, a case where only R ₁ and R ₂ are externally input will be described as an example. The counters 560 and 570 increase the count by the reference pulse and the line pulse, respectively, and return the output to 0 with r ₁ and r ₂ pulse inputs, respectively. As a result, the output of counters 560 and 570
Assuming that x ₄ and y ₄ , Δx and Δy are obtained as Δx = x ₄ / r ₁ Δy = y ₄ / r ₂ . Here, if r ₁ and r ₂ are constants, the weighting coefficient β (Δx, Δy) is determined by outputting x ₄ and y ₄ to the weighting factor determination unit 340 through the signal lines 341 and 342 as they are. Can be. In the method of this example, each time R ₁ pulse signals are output, the output multi-level data for one pixel is obtained. Therefore, high-speed processing is difficult in this state. During the processing of one image, the conversion rates r ₁ , r ₂ , R ₁ , R ₂ are usually invariant, so that the addresses x ₂ , y _{2 of the} original image are compared with the addresses x ₁ , y _{1 of the} reproduced image. And Δx, Δ
y fluctuates periodically. Therefore, x ₂ , y ₂ ,
This problem can be solved by determining Δx, Δy or x ₂ , y ₂ and β (Δx, Δy) and storing them in, for example, a line buffer and periodically outputting them. An example using this method is shown in FIG. In this scheme, the devices are x ₂ , y ₂ , Δx,
There are two types of modes, a preparation mode for determining Δy and an image processing mode for actually performing image processing. When the process starts,
First, a signal indicating the preparation mode is sent from the signal line 501. Therefore, the clock 510 is the reference clock to one scanning line output, the reference clock R _every time the address counter 540 and the line buffer 54 the output of the counter 560
Record at 5 and 565.

On the other hand, the same output number of lines pulses corresponding to the number of scanning lines of the image from the line pulse generator 511 simultaneously, the address counter 550 and the counter 570 for each line pulse R ₂
Is output to the line buffers 555 and 575. The line pulse generator 511 operates only in the preparation stage, and the clock 510
Can also be used. Thereafter, a signal indicating the image processing mode is output from the signal line 501 in the actual image processing. Then, the subsequent, entirely controlled by only the reference clock output from the clock 510, the address counter 520 is operated by input of a number less than _one R clock pulses. Therefore, higher-speed processing can be performed without changing the frequency of the reference clock. Then, at the same time when the output of the address counter 520 changes,
X ₂ corresponding thereto from the line buffers 545,555,565,575,
y ₂ , Δx, Δy are read out and sent to the weight coefficient determining unit 340. Here, the address of the data output from the line buffers 545 and 565 is controlled by the address counter 520. On the other hand, the address of the data output from the line buffers 555 and 575 is controlled by the address counter 530,
The address counter 530 operates every time the address counter 520 operates for one scanning line in the image processing mode. A selector 535 in the figure selects one of the output of the address counter 520 and the output of the line pulse generator 511 as a line pulse, and is controlled by a signal on a signal line 501.

With the above configuration, the reduction processing can be executed. The configuration and operation of the enlargement process are exactly the same as those of the reduction process.

Next, the image processing unit 30 for executing the density conversion process
One example of the configuration and operation of the control unit 500 will be described. In this case, the configuration of the control unit 500 is exactly the same as the above-described reduction processing. When the number of pixels does not change, r ₁ / R ₁ = 1, r ₂ / R ₂ = 1, and Δx = 0, Δy = 0 Becomes

In the density conversion, non-linear converted image data S _q (x, y) is output for certain restored image data S _p (x, y). The relationship between the two can be determined in advance by the characteristics of each input / output device. An example of the conversion formula is shown below.

_{S q (x, y) =} f {S p (x, y)} f (u) = v * (u / v) r Here, v in ≦ S _{p max} (u) wherein when the variable r with constant This processing is, for example, 1
It can also be realized with one ROM. Also, the variable r can be externally designated according to the processing. In this case, to calculate the output S _p over the entire range of possible pre-S _p,
This processing can be realized by storing the result in, for example, a RAM.

On the other hand, this processing can be executed simultaneously with the enlargement / reduction processing.

Next, the operation of the binarization processing unit 400 will be described in detail. This part is where the process of binarizing the multi-valued data to obtain the binary reproduced image data (the process (3) described in the above principle) is executed. This configuration is a conventional 2
A number of methods including the value processing method can be applied. Here, the case where the method described in the above process (3) is used will be described with reference to FIG. Converted image data S _q (x,
y) is input to the adder 430 through the signal line 401. The adder 430 adds S _q (x, y) and error data E (x, y) calculated by a method described later, and outputs multi-value data F (x, y). Here, the error data E (x, y) is input via the signal line 431. F (x, y) passes through a signal line 411 to a comparator 41
The threshold value is input to 0 and compared with a predetermined threshold value T by the threshold value determination unit 420. Based on the result of this comparison, the binary data Q (x, y) of the reproduced image is determined. Note that the error data E (x, y) used here is previously determined as an error ε generated when the multi-valued data F near the binarized (x, y) is binarized. Is the sum of the values multiplied by the weight coefficient δ. FIG. 18 shows an example of the weight coefficient δ. * In the figure
Is a pixel to be binarized at that time, and corresponds to the pixel at the coordinates (x, y) described here. In this case, E (x, y) is obtained by the following equation.

E (x, y) = 1/8 ｛ε (x−1, y−1) + ε (x + 1, y−1) + ε (x−2, y) + ε (x, y−2) +2 (ε (x , y−1) + ε (x−1, y−1))｝ On the other hand, the error ε of each pixel is the difference between the multi-valued data F and 0 or the difference between F and the maximum value F _max that F can take. One of them is one. This value is obtained as follows, at some point, the multi-level data F (x of the adder 430 is the coordinates (x ₂ -1, y _2)
2 -1, when the output _{y 2), F (x 2} -1, y 2) comparators 410
Are input to the differentiator 435 and the selector 440. Differentiator 435 outputs the difference between the converted image data F (x, y) maximum value of the possible values of F _max and _{F (x 2 -1, y 2} ), and sends to the selector 440. For example, in this example, since S (x, y) takes a value from 0 to 16, F _max = 16, and the differentiator 435 calculates 16−S _q (x ₂ −1, y
₂ ) output. Also here, if F (x ₂ −1, y ₂ )> F
_{When the value} becomes _max , the differentiator 435 outputs 0. selector
440 is F (x ₂ -1, y ₂ ) or F _max according to the following conditions:
One of −F (x ₂ −1, y ₂ ) is output as an error ε (x ₁ −1, y ₁ ).

ε (x, y) = F (x, y): Q (x, y) = 0 = F _max −F (x, y): Q (x, y) = 1 The output ε (x ₂ −1 , y ₂ ) is sent to a line buffer 450 or the like through a signal line 451. The process of obtaining E (x, y) from the error ε temporarily stored in the line buffer 450 is as follows.
This is performed by the latches 446, 447, 448 and the shift registers 471, 472. Next, the operation of each part will be described with reference to an example in which Q (x, y) is obtained.

When the binary data Q (x−1, y) is determined by the binarization processing unit, errors ε (x−1, y) and ε ( x−2, y), ε (x + 1, y−1),
ε (x, y−1), ε (x−1, y−1), ε (x, y−2) are output. Here, the data ε of the signal lines 451 and 454
(X−1, y) and ε (x, y−1) are shift registers 471,472
And 2 * ε (x−1, y) and 2 * ε (x, y−
1) is output. These are input to the adder 480.
The adder 480 adds the six types of input multi-value data, calculates 8 * E (x, y), and sends the result to the shift register 473. The shift register 473 obtains E (x, y) by shifting the input multi-value data. The obtained E (x, y) is input to the selector 485 via the signal line 431. Here, the selector 485 outputs E (x, y) or 0 to the adder 430 according to the data E flag sent from the external signal line 481. If the output of the selector 485 is E (x, y), the final output Q
Is a pseudo halftone image. On the other hand, if the output of the selector 485 is 0, the final output Q is a simple binarized image.

When the pseudo halftone processing method shown in this example is applied to a line figure such as a character, the image quality is less deteriorated as compared with the systematic dither method. However, by adding a mode of the simple binarization processing in such a manner, a higher-quality reproduced image can be obtained even when image processing is performed on a line figure. Moreover, the configuration of the device need only be increased by one selector 485.

When x = 1 or y = 1, a part of ε required to obtain E (x, y) may not exist. In this case, E (x, y) is determined using the values recorded in the line buffers 450 and 460 in advance.

According to the binarization processing method described above, it is possible to obtain a high-quality reproduced image free from moiré even if the original image is represented by halftone dots.

On the other hand, when the original image is not a halftone dot, a binarization process using the systematic dither method can be executed. For example, in the case of an output device that cannot display 1 * 1 pixel dots, binarization processing is performed using a halftone type dither matrix. Then, Bayer where dots of 1 pixel * 1 pixel originally tend to occur
An image binarized by the type dither matrix can also be displayed clearly. Further, as described above, when a line graphic is targeted, by using a simple binarization process with a fixed threshold as a re-binarization process method,
A high quality reproduced image can be obtained.

As described above, in this embodiment, since various methods can be applied as the binarization processing of the converted image data, higher image quality can be obtained by switching the re-binarization processing method according to each output device or image. Can be output. FIG. 19 shows an example of a basic configuration of the image processing apparatus in that case. Here, the image processing apparatus includes an original image data temporary storage unit 100, a density data restoration unit 200, an image processing unit 300, and a control unit 50.
0, a plurality of binarization processing units 710, 720, 730 and their two
It is composed of a selector 740 that switches and selects a value processing unit. Then, one of a plurality of binarization processing methods is switched by the signal G input from the signal line 701. The signal G is an external instruction, a part of image data,
Alternatively, it is determined by the output of the device.

As described above, with the above-described apparatus, a high-quality reproduced image can be obtained with a relatively simple configuration.

Although the present invention has been described for a monochrome image, it can be extended to a color image.

〔The invention's effect〕

As described above, according to the present invention, enlargement of an image,
When processing such as reduction and rotation is performed, a high-quality reproduced image can be obtained. Further, the process of converting the density of an image, which was conventionally applicable only to multi-valued data, can also be executed for a pseudo halftone image represented by binary data. Further, the monochrome pattern of the pseudo halftone image can be converted according to the output device.

[Brief description of the drawings]

FIG. 1 is a diagram showing a basic configuration of an image processing apparatus according to an embodiment of the present invention, FIG. 2 is a diagram for explaining a positional relationship between an original image and a reproduced image by linear density conversion, and FIG. FIG. 4 is a view showing an example of processing for restoring multi-valued image data, FIG.
FIG. 6 is a diagram showing an example of the relationship between the positions of the restored image data and the converted image data. FIG. 6 is a diagram showing an example of a multivalued data reduction process by the inverse distance proportional method. FIG. 7 is a grayscale image used in the present invention. FIG. 8 is a view for explaining the principle of an example of a data binarization processing method, FIG. 8 is a block diagram showing an example of the configuration of a density data restoring unit in FIG. 1, and FIG. FIG. 10 is a diagram showing an example, FIG. 10 is a diagram showing an example of a weight coefficient used for restoring density data, FIG. 11 is a diagram showing an example of a method for restoring density data, and FIG. 12 is an asymmetric weight used for restoring density data. FIG. 13 is a diagram showing an example of a coefficient, FIG. 13 is a diagram showing an example of restored data when a weight coefficient is changed, FIG. 14 is a diagram showing an example of a configuration of an image processing unit in FIG. 1, and FIG. FIG. 1 shows an example of a control unit in FIG. 1. FIG. 16 shows a control unit for performing high-speed processing. Figure showing an application example, Figure 17 is a diagram showing an example of the configuration of the binarization processing unit in FIG. 1, the
FIG. 18 is a diagram showing an example of a weight coefficient for determining an error correction amount, and FIG. 19 is a block diagram showing an example of a basic configuration of an image processing apparatus having a plurality of binarization processing units. 100: Image data temporary storage unit, 101: Original image data input line,
110: Line buffer, 115: Line buffer, 121, 122, 12
3: Latch row, 130: signal line bundle, 200: density data restoration unit, 21
0: multiplier, 220: adder, 230: external switch, 240: image determination unit, 300: image processing unit, 301: restored image data input line, 31
1,312: Latch, 320: Line buffer, 321,322,323,324:
Multiplier, 330: adder, 340: weight coefficient determination unit, 400: binarization processing unit, 410: comparator, 420: threshold value determination unit, 430: adder, 43
5: Differentiator, 440: Selector, 446, 447, 448: Latch, 450, 46
0: Line buffer, 471, 472, 473: Shift register, 480:
Adder, 485: Selector, 491: Playback image data output line, 50
0: control unit, 510: clock, 511: line pulse generator, 51
5: End detection unit, 520, 530, 540, 550: Address counter, 56
0,570: counter, 525: selector, 545, 555, 565, 575: line buffer, 701: signal line, 710, 720, 730: binarization processing unit, 7
40: Selector.

──────────────────────────────────────────────────続 き Continuation of the front page (56) References JP-A-62-114377 (JP, A) JP-A-63-13578 (JP, A) JP-A-61-131683 (JP, A) (58) Investigation Field (Int.Cl. ⁶ , DB name) G06T 1/00

Claims (3)

(57) [Claims] 1. An image storage means for temporarily storing binary digital image data P, and image data P stored in the image storage means.
Using a scanning window of M pixels (M ≧ 1) in the image processing apparatus, comprising: means for restoring density data Sp of each pixel from binary image data of M pixels in the scanning window; Means for performing image processing on the restored density data Sp to obtain multi-value data Sq; and means for re-binarizing the multi-value data Sq to determine binary image data Q, The means for restoring the density data Sp has a plurality of types of weighting factors corresponding to the positions of the respective pixels in the scanning window.
Means for selecting an asymmetrical one of the weighting coefficients for the straight line portion of Sp; means for performing the image processing to obtain multi-valued data Sq; An image processing apparatus comprising means for multiplying a value of image data by a reciprocal of a distance to each pixel as a weight and determining converted image data from a value of a sum of each pixel. 2. An image storage means for temporarily storing binary digital image data P, and image data P stored in the image storage means.
Using a scanning window of M pixels (M ≧ 1). A means for restoring density data Sp of each pixel from binary image data of M pixels in the scanning window. A ROM having a plurality of address lines is used as a density data restoring unit, and a switching signal is changed to a plurality of binary data or multi-value data, thereby switching between three or more weighting factors. Means for performing multi-value data Sq by performing image processing on Sp; means for re-binarizing the multi-value data Sq to determine binary image data Q; means for restoring the density data Sp Has a plurality of types of weighting factors according to the position of each pixel in the scanning window, has means for selecting one of the various weighting factors, and means for performing the above image processing to obtain multi-valued data Sq , The number at the position surrounding the converted image data The value of the restored image data of the unit, multiplies the inverse of the distance to each pixel as a weight, an image processing apparatus characterized by comprising means for determining a converted image data than the value of the sum of each pixel. 3. Image storage means for temporarily storing binary digital image data P, and image data P stored in the image storage means.
Using a scanning window of M pixels (M ≧ 1) in the image processing apparatus, comprising: means for restoring density data Sp of each pixel from binary image data of M pixels in the scanning window; Means for performing image processing on the restored density data Sp to obtain multi-value data Sq; and means for re-binarizing the multi-value data Sq to determine binary image data Q, The means for restoring the density data Sp has a plurality of weighting factors according to the position of each pixel in the scanning window, and selects one of the various weighting factors based on an external instruction or a determination result based on the characteristics of the image. The restoration means for restoring the density data Sp of each pixel arbitrarily determines a range to be referred to when restoring the density data, prepares a plurality of weight coefficient matrices, and responds to the image to be processed. Of the multiple matrices according to the switching signal The image processing apparatus characterized by switching to One.