JPH08307678A - Image processing unit - Google Patents

Abstract

PURPOSE: To prevent deterioration in image quality by setting initially error data less than a threshold level to each picture element position of an error storage means prior to starting binarization or multi-value processing so as to quicken incidence of a black (white) picture element at a low (high) density area of an image. CONSTITUTION: An error setting section 17 sets initially a prescribed value to each picture element position storing accumulatingly a weight error in an error storage circuit 16 prior to start of binarization processing. When a uniform input image signal Si is given to a correction circuit 11 in this state, the circuit 11 at first adds an image correction signal Sc from the circuit 16 to the signal Si for the correction to provide an output of a correction signal Sic. The signal Sic is compared with a threshold level Th at a binarization circuit 12 and the circuit 12 provides an output of a binarized image signal So. On the other hand, a binarization error calculation circuit 13 calculates the difference between the signals Sic and So and provides it as a binarization error signal Ser to a weight error calculation circuit 14. The circuit 14 calculates a weight error based on a weight coefficient of a noted picture element with respect to peripheral picture elements and applies accumulation processing to the level set initially at the picture element position of the circuit 16.

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 which binarizes or multivalues a multivalued image obtained by raster scanning, for example, characters or photographs.

[0002]

2. Description of the Related Art Generally, in document image processing capable of handling image information as well, in the case of image information read by an image reading means such as a scanner, image information such as characters and line drawings is simply binarized by a fixed threshold. The image information having gradation such as a photograph is binarized by a pseudo gradation means such as a systematic dither method.

This is because if the read image information is subjected to a simple binarization process with a fixed threshold value, the image quality is not deteriorated because the resolution is preserved in the character or line drawing area, but the gradation is applied in the photographic image area. There is a problem that the image quality is deteriorated because the image quality is not preserved, and when the read image information is subjected to pseudo gradation processing such as the systematic dither method, the gradation property is preserved in the area of the photographic image, and thus the image quality deterioration does not occur. This is because, although it does not occur, there is a problem that the image quality is deteriorated because the resolution is lowered in the area of characters and line drawings.

On the other hand, there is an error diffusion method as an image processing which can satisfy the gradation property of the area of the photographic image and can maintain the resolution even in the area of the character and the line drawing as compared with the systematic dither method. The error diffusion method has already added 2 to the density of the pixel of interest.
The binarization error of the binarized peripheral pixels is multiplied by a certain weighting coefficient, and binarized with a fixed threshold.

As an image processing apparatus using such an error diffusion method, the one shown in FIG. 3 is conventionally known.
That is, the input image signal Si and the image correction signal Sc are input to the correction circuit 1, the image information of the target pixel is corrected by the correction circuit 1, and the corrected correction image signal Sic is supplied to the binarization circuit 2. There is. The binarization circuit 2 also receives the binarization threshold Th. The binarization circuit 2 compares the corrected image signal Sic with the binarization threshold Th, and if Sic> Th, then 2
A signal "1", that is, a black pixel is output as the binarized image signal So, and a signal "0", that is, a white pixel is output as the binarized image signal So when Sic≤Th. In addition, 2
As the threshold value Th, the input image signal Si is, for example, 8b.
If it is it, the value of the input image signal Si is from 0 to 255, so that 128, which is an intermediate value, is set.

On the other hand, the corrected image signal Sic from the correction circuit 1
And the binarized image signal So from the binarization circuit 2 is supplied to the binarization error calculation circuit 3, and the binarization error calculation circuit 3 calculates the binarization error of the target pixel binarized. ing. That is, when the binarized image signal So is "1", the value is "25".
5 ”and when it is“ 0 ”, it is set to“ 0 ”to calculate the difference from the corrected image signal Sic, that is, Sic−So.

The binarization error signal Ser from the binarization error calculation circuit 3 is supplied to the weight error calculation circuit 4. The weighting coefficient from the weighting coefficient storage circuit 5 is also input to the weighting error calculation circuit 4. The weighting coefficient storage circuit 5 is a kind of error filter, and stores therein the weighting coefficients A, B, C, and D for the peripheral pixels of the pixel of interest “*”. The weighting factors A, B, C, D are, for example, A =
7/16, B = 1/16, C = 5/16, D = 3/1
It is set to 6.

The weight error calculation circuit 4 uses the binarization error signal S
er is multiplied by each of the weighting factors A, B, C, and D to calculate the weighting error of the position corresponding to the periphery of the pixel of interest, and the error is supplied to the error storage circuit 6. The error memory circuit 6 has the target pixel “*”.
The relevant weighting errors calculated for the relevant surroundings are accumulated, and the accumulated values eA, eB, eC, eD are stored. The error storage circuit 6 supplies the pixel corresponding to the input image signal Si, that is, the cumulative storage value of the pixel of interest "*" to the correction circuit 1 as the image correction signal Sc.

Further, in recent years, even in the error diffusion method when the number of gradations (the number of levels) of the output device is large, the above-mentioned 2
The binarization circuit 2 is replaced by a multi-value binarization circuit using a threshold value corresponding to the number of gradations. For example, if the number of output gradations is 4, three kinds of thresholds will be used.

Further, as seen in Japanese Patent Publication No. 6-66876, a plurality of weighting factors are provided to prevent texture, a random number is generated, and 1 is selected from these for each pixel.
There is also one that selects one to perform error diffusion.

[0011]

However, in any of these, when the density level of the image background is extremely small, the image density in the area is accurately reproduced even if the image density is constant. There was a problem I couldn't do. This is because the binarization error in the error diffusion method is diffused from the upper left of the image in the main scanning direction and the sub scanning direction and accumulated, and then the black pixel appears for the first time when the threshold value Th is reached. Even if there is, it is because the appearance rate of the black pixel is different between the start time of the error diffusion process and some time after the start. Therefore, the appearance delay of the black pixel occurs in the low density region.

Table 1 shows that when the input image signal is "8" in 256 gradations and the initial error of each pixel position of the error storage circuit 6 is "0", each pixel in the main scanning direction and the sub-scanning direction. The binarization error at the end of the processing is shown. The shaded areas in the table indicate the locations where black pixels occur. As can be seen from Table 1, black pixels are not output until the 10th line in the sub-scanning direction.

[0013]

[Table 1]

As described above, in any of the image processing apparatuses using the conventional error diffusion method, the output delay of the black pixel occurs in the low density region, which causes the shape of the image such as the background to be different, resulting in the deterioration of the image quality. There was a problem. This phenomenon occurs not only in the low density area of the image but also in the high density area of the image, and the appearance delay of white pixels occurs in the high density area. As described above, the image processing apparatus using the conventional error diffusion method has a problem in that the image quality is deteriorated due to the delay in the appearance of the black pixels and the white pixels in the low density area and the high density area of the image.

Therefore, the present invention provides an image processing apparatus capable of accelerating the appearance of black pixels in a low density area of an image and white pixels in a high density area to prevent image quality deterioration.

[0016]

The invention according to claim 1 is
Correction means for correcting the input image signal as a pixel of interest on the basis of the image correction signal; and means for comparing the image signal of the pixel of interest corrected by the correction means with a preset threshold value to binarize or multivalue Error calculating means for obtaining an error of the target pixel from the image signal of the ordered pixel binarized or multivalued by this means and the corrected image signal of the target pixel inputted to the binarized or multivalued means, Weighting coefficient storage means for storing the weighting coefficient of the peripheral pixel of the pixel of interest, and weighting for calculating the weighting error at the peripheral pixel position by multiplying the error obtained by the error calculating means by the weighting coefficient of the peripheral pixel stored in the weighting coefficient storage means. Error calculating means, error storing means for accumulating and storing the weighting errors at the peripheral pixel positions calculated by the weighting error calculating means, and each pixel position for accumulating and storing the weighting error by the error storing means. And an error setting means for initializing error data equal to or less than a threshold value used when binarizing or multi-valued in advance, and an image correction signal corresponding to a pixel of interest of an input image signal It is generated based on the accumulated storage contents of the position.

According to a second aspect of the invention, in the image processing apparatus according to the first aspect, the error setting means sets the error data at each pixel position of the error storage means so that the total sum of the error data in each line becomes zero. This is for initial setting.

According to a third aspect of the invention, in the image processing apparatus according to the first or second aspect, the error setting means uses the function generating device to generate error data for initial setting at each pixel position of the error storage means. To do.

[0019]

In the present invention having such a configuration, error data below the threshold value is initially set at each pixel position of the error storage means before the binarization or multi-value conversion is started.

When the input image signal is input, the correction means sets the input image signal as the pixel of interest, inputs the image correction signal corresponding to this pixel of interest from the error storage means, and outputs the corrected image signal. This corrected image signal is binarized or multi-valued by being compared with a threshold value in a binarization or multi-value conversion means.

On the other hand, the error of the pixel of interest is obtained from the binarized or multivalued image signal of the ordered pixel and the corrected image signal before binarized or multivalued, and the weighting coefficient of the peripheral pixels is added to the obtained error. The multiplication is performed to calculate the weighting error at each peripheral pixel position. Then, the calculated weighting error at the peripheral pixel position is accumulated at the corresponding pixel position of the error storage means.
Then, when an input image signal to be the next pixel of interest is input, an image correction signal corresponding to the pixel of interest is supplied from the error storage unit to the correction unit. The above processing is repeated for each pixel of one line, and this is further repeated for each line to perform the entire image processing.

[0022]

Embodiments of the present invention will be described below with reference to the drawings. First, an embodiment corresponding to claim 1 will be described. In FIG. 1, reference numeral 11 denotes a correction circuit as a correction means. The correction circuit 11 inputs an input image signal Si obtained by reading an image by an image reading device such as a scanner pixel by pixel, and an error as error storage means. An image correction signal Sc corresponding to each pixel is input from the storage circuit 16.
The input image signal Si is, for example, an 8-bit signal,
It is a multi-valued image signal that takes a value of 255.

The correction circuit 11 corrects the input image signal Si by adding the value of the image correction signal Sc to the value of the input image signal Si, and supplies the corrected image signal Sic to the binarization circuit 12. The binarization circuit 12 compares the corrected image signal Sic with the binarization threshold Th and outputs a signal "1", that is, a black pixel as the binarized image signal So if Sic> Th.
If Sic≤Th, a signal "0", that is, a white pixel is output as the binarized image signal So. As the binarization threshold Th, for example, 128 which is an intermediate value of 256 gradations is set.

Corrected image signal Sic from the correction circuit 11
And the binarized image signal So from the binarization circuit 12 is supplied to a binarization error calculation circuit 13 which is an error calculation means.
The binarization error calculation circuit 13 sets the value to “255” when the binarized image signal So is “1” and “0” when the binarized image signal So is “0”, that is, the difference from the corrected image signal Sic, that is, , S
ic-So is calculated to obtain the binarization error.

The binarized error signal Ser from the binarized error calculation circuit 13 is supplied to a weighted error calculation circuit 14 which is a weighted error calculation means. The weight error calculation circuit 14 also includes a weight coefficient storage circuit 15 which is a weight coefficient storage means.
The weighting factor is input from.

The weighting coefficient storage circuit 15 is a kind of error filter, and stores therein the weighting coefficients A, B, C and D for the peripheral pixels of the pixel of interest "*". The weighting factors are, for example, A = 7/16 and B = 1/1.
6, C = 5/16 and D = 3/16 are set.

The weighting error calculating circuit 14 multiplies the binarization error signal Ser by the weighting factors A, B, C and D, respectively, to calculate the weighting error at the position corresponding to the periphery of the pixel of interest, and outputs the result to the above. It is supplied to the error storage circuit 16. The error storage circuit 16 accumulates the corresponding weighting errors calculated for the corresponding periphery of the pixel of interest "*" and stores the accumulated values eA, eB, eC, eD. Then, the error storage circuit 16 determines the cumulative storage value of the pixel corresponding to the input image signal Si, that is, the pixel of interest "*", as the image correction signal Sc.
Is supplied to the correction circuit 11.

An error setting means 17 initializes error data below the threshold Th at each pixel position for cumulatively storing the weighting error in the error storage circuit 16 before starting the binarization processing of the input image signal Si. In the error setting part as
As the error data, for example, a constant value "32" is initially set at each pixel position.

In the embodiment having such a structure, as shown in FIG.
In the image space shown in (1), the error setting unit 17 initializes a fixed value "32" at each pixel position where the weight error in the error storage circuit 16 is cumulatively stored, before starting the binarization process.

In this state, when a uniform input image signal Si having a gradation value of "8" is input to the correction circuit 11, first, 1
The first pixel P11 of the first line becomes the target pixel, and the correction circuit 11 performs correction by adding the image correction signal Sc from the error storage circuit 16 to the input image signal Si. The value of the image correction signal Sc at this time is the first pixel P of the first line.
It corresponds to the value stored in the 11th pixel position, and since "32" is initially uniformly set in the error storage circuit 16, it becomes "32".

Therefore, the correction circuit 11 obtains 8 + 32 = 40 and outputs the corrected image signal Sic "40". This corrected image signal Sic is compared with the threshold Th in the binarization circuit 12, but since the threshold Th is "128", the binarization circuit 12
Outputs a white pixel "0" as the binarized image signal So.

On the other hand, the binarization error calculation circuit 13 determines the difference between the corrected image signal Sic and the binarized image signal So, that is, 40.
−0 = 40 is calculated, and this “40” is binarized error signal S
It is supplied to the weight error calculation circuit 14 as er. The weight error calculation circuit 14 is provided with the pixels P12, P21, P around the pixel of interest P11.
The weighting error is calculated by the weighting factors A, B, and C for 22. The weighting error corresponding to the pixel P12 is converted into the binarization error signal S
It is obtained by multiplying er by the weighting coefficient A. That is, 40 × 7 /
16 is used to obtain “18”. Similarly, the weighting error corresponding to the pixel P21 is obtained by multiplying the weighting coefficient C. That is, “13” is obtained by 40 × 5/16. Further, the weighting error corresponding to the pixel P22 is obtained by multiplying the weighting coefficient B. That is, “3” is obtained by 40 × 1/16.

Corresponding pixel position eA of the error storage circuit 16
Since "32" is initially set for (corresponding to P12), eC (corresponding to P21), and eB (corresponding to P22),
"18", "13" and "3" are accumulated in this respectively, and e
A = 50, eC = 45, eB = 35. The pixel of interest P11 becomes "40" by accumulating "8" in "32". In this way, the image processing of the first 1st pixel P11 of the 1st line is completed.

Next, the second pixel P12 of the first line becomes the target pixel, and the correction circuit 11 adds "8" of the input image signal Si and "50" of the image correction signal Sc. In this way, the corrected image signal Sic “58” is output from the correction circuit 11 to the binarization circuit 1.
2 is supplied. Since "58" is also smaller than the threshold Th,
The binarization circuit 12 outputs "0" of the white pixel as the binarized image signal So.

On the other hand, the binarization error calculation circuit 13 has a difference between the corrected image signal Sic and the binarized image signal So, that is, 58.
−0 = 58 is calculated, and this “58” is binarized error signal S
It is supplied to the weight error calculation circuit 14 as er. The weight error calculation circuit 14 includes pixels P13, P21, P around the target pixel P12.
The weighting error is calculated by the weighting factors A, B, C and D for 22 and P23. The weighting error corresponding to the pixel P13 is obtained by multiplying the binarization error signal Ser by the weighting coefficient A. That is,
“25” is calculated by 58 × 7/16. Similarly, the weighting error corresponding to the pixel P21 is obtained by multiplying the weighting coefficient D. That is, “11” is obtained by 58 × 3/16. Further, the weighting error corresponding to the pixel P22 is obtained by multiplying the weighting coefficient C. In other words, 58 × 5/16 gives "1
8 ”. Further, the weighting error corresponding to the pixel P23 is obtained by multiplying the weighting coefficient B. That is, “4” is obtained by 58 × 1/16.

Corresponding pixel position eA of the error storage circuit 16
“32” and “4” are assigned to (corresponding to P13), eD (corresponding to P21), eC (corresponding to P22), and eB (corresponding to P23).
Since "5", "35", and "32" are stored, "25", "11", "18", and "4" are accumulated respectively, and eA = 57, eD = 56, eC = 53, eB = 36. Become. In this way, the image processing of the second pixel P12 of the first line is completed. Thereafter, this image processing is sequentially performed for each pixel on one line, and this is sequentially performed for all lines, and the image processing for one page is completed.

Table 2 shows the value of the binarized error signal Ser generated when the processing of each pixel is completed. It should be noted that this table shows 13 pixels in the main scanning direction and a part of 18 lines in the sub scanning direction.

[0038]

[Table 2]

The shaded areas in this table indicate the locations where black pixels occur, and the first black pixel appears on the sixth line in the sub-scanning direction. As described above, the black pixels appear earlier than in the conventional case. That is, it is possible to prevent the deterioration of the image quality by accelerating the appearance of the black pixels in the low density region where the value of the input image signal Si is "8". Further, if this binarization process is performed, the appearance of white pixels in the high density area can be accelerated and the image quality deterioration can be prevented.

In the above-described embodiment, the error setting section 17 has been described in which a constant value is initially set as error data at each pixel position where the weight error in the error storage circuit 16 is cumulatively stored. For example, a random number generator may be provided in the error setting unit, and the random number generated from this random number generator may be initially set as error data at each pixel position in the error storage circuit 16.

For example, when a random number generated in the range of 0 to 32 is initially set as error data, the value of the binarized error signal Ser generated when the processing of each pixel is finished is shown in Table 3. become. The input image signal Si
The value of is "8" as described above.

[0042]

[Table 3]

In this table, the shaded areas indicate the locations where black pixels occur, and the first black pixel appears on the 7th line in the sub-scanning direction. In this way, even if a random number is used to set the error data, the appearance of black pixels can be accelerated and image quality deterioration can be prevented.

By the way, the error data used for the initial setting must be smaller than the threshold value Th. For example, when the error data of “144” larger than the threshold Th “128” is uniformly initialized to all the pixel positions of the error storage circuit 16, the binarization that occurs when the processing of each pixel is completed. The values of the error signal Ser are shown in Table 4. The value of the input image signal Si is "8" as described above.

[0045]

[Table 4]

As can be seen from this table, the appearance of black pixels can be accelerated, but black pixels are concentrated on the first line, and thereafter, no black pixels appear until the 11th line. This causes a loss of gradation in halftone processing.

Next, an embodiment corresponding to claim 2 will be described. The entire circuit configuration is the same as in FIG. In this embodiment, as the error setting unit, 1 of error data to be initialized is set.
Error data is generated so that the sum of the lines becomes zero.

That is, the value of the error data set before the binarization process becomes an error given to the entire image as it is. In other words, although black pixels appear earlier, the error causes the density of the entire image to change.

The sum of the input image signals Si is Di, the setting error is E, the main scanning direction of the image is n, and the output image signal So.
Let Do be the total sum of the following: Do = Di + E × n, and the output image becomes darker by E × n. Therefore, in order to solve this, the error is set so that the total sum of E × n becomes zero. That is,

[Equation 1]

It suffices to set

Table 5 shows that when the error setting unit is provided with a random number generator for generating random numbers of -64 to 64 and the error is set so that the sum of the error data becomes zero, when the processing of each pixel is completed. The value of the generated binarization error signal Ser is shown. The value of the input image signal Si is "8" as described above.
Is.

[0052]

[Table 5]

According to this, the appearance of the black pixel becomes faster on the 8th line, and the density of the entire image becomes constant, so that the reproducibility of the image can be improved.

In this embodiment, random numbers in the range of -64 to 64 are generated, but it is needless to say that the present invention is not limited to this.

Next, an embodiment corresponding to claim 3 will be described. The entire circuit configuration is the same as in FIG. In this embodiment, the error is set using a function that is periodically repeated so that the total sum of one line of error data to be initialized is zero, for example, a trigonometric function (sin). That is, a trigonometric function generator is provided in the error setting unit to initially set the error at each pixel position in the error storage circuit 16.

The trigonometric function has an error of E, a pixel of n, a period of the trigonometric function of t, a circular constant of π, and an amplitude (maximum value) of x. En = sin {(n / t) · 2π}・ It becomes x. For example, when an error is initially set at each pixel position in the error storage circuit 16 using a trigonometric function with a cycle t = 12 and an amplitude x = 64, a binarized error signal generated when the processing of each pixel is completed. Table 6 shows the value of Ser. The value of the input image signal Si is "8" as described above.

[0057]

[Table 6]

According to this, the appearance of black pixels is accelerated to the 7th line, and the density of the entire image is constant, so that the reproducibility of the image can be improved.

In this embodiment, the period t = 12 and the amplitude x =
Of course, 64 trigonometric functions were used, but the present invention is not limited to this.

In each of the above embodiments, the input image signal Si is binarized by the binarizing circuit and output. However, the binarizing circuit is replaced with a multi-valued circuit in which a plurality of threshold values are set. The present invention can also be applied to the case where the input image signal Si is converted into a multi-valued output.

[0061]

As described above, according to the present invention, the error storage means for accumulating and storing the weighting error is initialized with the error data equal to or less than the threshold value used when the error setting means performs binarization or multi-value conversion in advance. Since the setting is made, it is possible to prevent black pixels in the low density area of the image and white pixels in the high density area from appearing earlier and prevent the deterioration of the image quality.

Further, since the sum of the initially set error data in each line is set to zero, the density of the entire image becomes constant and the reproducibility of the image can be improved.

[Brief description of drawings]

FIG. 1 is a block diagram showing an embodiment of the present invention.

FIG. 2 is a diagram showing an image space in which binarization processing is performed in the same embodiment.

FIG. 3 is a block diagram showing a conventional example.

[Explanation of symbols]

 11 ... Correction circuit 12 ... Binarization circuit 13 ... Binarization error calculation circuit 14 ... Weight error calculation circuit 15 ... Weight coefficient storage circuit 16 ... Error storage circuit 17 ... Error setting unit

Claims (3)

[Claims] 1. A correction unit for correcting an input image signal as a pixel of interest based on an image correction signal, and an image signal of the pixel of interest corrected by the correction unit is compared with a preset threshold value to be binarized or multivalued. The error of the pixel of interest is converted from the image signal of the ordered pixel binarized or multivalued by this device and the corrected image signal of the pixel of interest input to the binarized or multivalued device. Error calculating means for obtaining, weighting coefficient storage means for storing weighting coefficients of peripheral pixels of the target pixel, and peripheral pixels by multiplying the error obtained by the error calculating means by the weighting coefficient of the peripheral pixels stored in the weighting coefficient storage means Weight error calculating means for calculating the weight error at the position, error storage means for accumulating and storing the weight errors at the peripheral pixel positions calculated by the weight error calculating means, and the error storage means An image correction signal corresponding to the pixel of interest of the input image signal, comprising error setting means for initially setting error data below a threshold value used when binarizing or multi-valued in advance at each pixel position for cumulatively storing the difference. Is generated based on the cumulative storage content of the corresponding pixel position of the error storage means. 2. The image processing apparatus according to claim 1, wherein the error setting means initializes the error data at each pixel position of the error storage means so that the sum of the error data in each line becomes zero. . 3. The image processing apparatus according to claim 1, wherein the error setting means generates error data which is initially set at each pixel position of the error storage means by using a function generating device.