JPH10200742A - Binarizing image processing method for multi-gradation data and image processing unit adopting the method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To smooth gradation and to enhance image quality by suppressing tailing and delayed starting of a highlight part of a gradation image and production of stripes to a transition part from the highlight part to a middle tone part which are caused by binarizing multi-gradation data by the error spread method. SOLUTION: An error distribution coefficient of a highlight part to be distributed in the main scanning direction is selected higher than the total sum of the coefficient in the subscanning direction, and the error distribution coefficient for a part from the end of the highlight part to a middle tone part is selected to be a median of the error distribution coefficient of the highlight part and the middle tone part. Moreover, the error distribution coefficient of the middle tone part is selected for the error distribution coefficient of the part after the middle tone part. Relating to this processing unit, an error distribution coefficient acquisition circuit 3 applies an error distribution coefficient to gradation data classified into three stages by a gradation share circuit based on gradation of received data 1 so as to conduct error distribution processing for the data and provides an output of binarized data.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

[0001] 1. Field of the Invention [0002] The present invention relates to a binarized image processing method for binarizing multi-tone data by an error diffusion method and a processing apparatus for executing the method.

[0002]

2. Description of the Related Art When outputting multi-gradation data to an imaging device capable of expressing only two gradations of 0 or 1, pseudo-expression is performed by gradation halftone processing. As this processing method, there are a collective dither method and a distributed dither method, which is a method of determining dot on / off by a simple comparison between a fixed two-dimensional threshold table and the pixel density. . Since this method reproduces the gradation by the number of dots within a certain area, it is possible to surely express the gradation corresponding to the area of the table. However, in this method, the resolution depends on the size of the table, that is, since a fixed two-dimensional table is used regardless of the pixel density, the horizontal resolution is the width of the table, and the vertical resolution is the height of the table. There was a problem that it decreased by minutes.

As another method, an error diffusion method (referred to in the document "loyd. RWand L. Steinberg: Adoptive algorithm for
spatial gray scale, SID Int.Sym.Digest of Tech Pap
ers36.37.1975 ") is known, but this method calculates the input pixel density and the density difference between the output pixel density output as a result of comparing the input pixel density with the threshold value. This difference is distributed to neighboring pixels after a specific weight is given, and this is a pseudo gradation expression method capable of high-quality output without substantially lowering the resolution. Table 1 shows that the pixel (x,
This is an actual example of the weight at the time of distributing the error of y) to the neighboring pixels, that is, the error diffusion coefficient, which has been proposed in the above-mentioned literature.

[0004]

[Table 1]

However, when using this method,
Since the distribution ratio of the density error to the neighboring pixels is always kept in a fixed positional relationship, there is a problem of creating a noise pattern.To solve this problem, conventionally, for example, a plurality of correction error distribution coefficient sets have been used. It is prepared such that the distribution ratio of the error to the peripheral pixels of the target pixel does not have a fixed relative relationship by randomly selecting the selected error and calculating the integration error (Japanese Patent Application Laid-Open No. 64-18369). Gazette,
(See U.S. Pat. No. 4,955,506.) A method in which a randomly selected variation value is added to a weighting coefficient to vary the coefficient, and the main scanning direction of pixel values is reversed left and right for each row.
Further, according to the level of the multi-valued data, the data area matrix size of the error diffusion coefficient is switched (for example, when the data is a gray level, a small matrix is selected. Otherwise, a large matrix is selected), and an error is diffused based on the matrix to prevent a periodic structure from being formed in the image (see JP-A-1-157166). .

[0006]

As described above, in the conventional error diffusion method, errors are accumulated to some extent in the occurrence of the noise pattern, that is, in a highlight portion of a gradation image whose density gradually increases or decreases. In addition to the above problem, in the main line, there is a problem that a band-like pattern (hereinafter, referred to as a tailing phenomenon) occurs in the main scanning direction because dots are easily turned on. In the area, dots are not turned on unless errors for several lines are accumulated, due to the characteristics of the error diffusion method.
There is a problem that an image looks like a frame (hereinafter, referred to as a trailing phenomenon).

Accordingly, an object of the present invention is to solve the above-mentioned problem of the conventional error diffusion method and improve the image quality. Further, when the error diffusion coefficient of the invention of claim 1 described below is directly applied to a gradation image including an image other than the highlight portion, a side effect of streaking at the boundary with the middle tone portion is recognized, Another object of the present invention is to eliminate this side effect, and to achieve a smooth gradation and an improved image quality.

[0008]

In order to solve the above-mentioned problem, the present invention is different from the conventional one in that an error diffusion coefficient is applied in which the weighting ratio is changed according to the density of the original image. The error diffusion process is performed.

According to a first aspect of the present invention, there is provided a binarized image processing method for binarizing multi-tone data by an error diffusion method, wherein an error diffusion coefficient of a highlight portion in a main scanning direction is changed to an error diffusion coefficient of a sub scanning direction. The image quality of the highlight portion is efficiently improved by performing the error diffusion process by applying the error diffusion coefficient distributed more than the sum of the above.

According to a second aspect of the present invention, in the binarized image processing method for binarizing multi-tone data by an error diffusion method, an error diffusion coefficient in a portion from the end of a highlight portion to the start of a middle tone portion is calculated by: The error diffusion coefficient is set to an intermediate value between the highlight part and the middle tone part, and the error diffusion coefficient is applied to perform the error diffusion processing, so that the gradation is smoothly expressed and the image quality is efficiently improved. is there.

According to a third aspect of the present invention, there is provided a binary image processing method for binarizing multi-tone data by an error diffusion method, wherein the tone data of the original image is highlighted in accordance with the tone value. From the end, the beginning part of the middle tone part and the three steps after the middle tone part, the highlight part distributes the error diffusion coefficient in the main scanning direction larger than the sum of the error diffusion coefficients in the sub-scanning direction, In the part from the end of the highlight part to the beginning of the middle tone part, the error diffusion coefficient of the intermediate value between the error diffusion coefficient of the highlight part and the middle tone part, and the error diffusion coefficient of the middle tone part after the middle tone part The obtained gradation data is subjected to error diffusion processing by applying the obtained error diffusion coefficients to the divided gradation data, so that the gradation is smoothly expressed and displayed. The is obtained so as to improve efficiency.

According to a fourth aspect of the present invention, there is provided a binarized image processing apparatus for binarizing multi-tone data by an error diffusion method, wherein a tone portion of an original image is highlighted in accordance with a tone value. From the end to the beginning of the middle tone section,
And a tone distribution circuit for distributing into three stages after the middle tone portion, and in the highlight portion, the error diffusion coefficient in the main scanning direction is distributed more than the sum of the error diffusion coefficients in the sub-scanning direction. The error diffusion coefficient, the error diffusion coefficient of the intermediate value of the error diffusion coefficient of the highlight part and the middle tone part in the part from the end of the highlight part to the beginning of the middle tone part, and the middle tone part after the middle tone part An error diffusion coefficient acquisition circuit for acquiring the error diffusion coefficient of each of the above, and an error diffusion process is performed by applying the error diffusion coefficient acquired by the error diffusion coefficient acquisition circuit to each of the gradation data distributed by the gradation distribution circuit. An error diffusion processing unit is provided so that gradation is smoothly expressed and image quality is efficiently improved.

[0013]

Embodiments of the present invention will be described below in detail. Table 2 shows a specific example of the accumulated error when the error diffusion coefficient (Table 1) proposed in the literature for the error diffusion method is applied.
5, the accumulation error of a certain 16 × 16 area when the area is uniformly filled with a density of 3% (8/256). The shaded portion in the table is the portion where the error value exceeds the threshold, and this pixel is actually drawn.

[0014]

[Table 2]

From Table 2, the following matters become clear. That is, (1) Pixels for which the accumulated error exceeds the threshold value and are turned on are concentrated in the third row (this is the cause of the tailing phenomenon). (2) The pixels are not turned on from the third row to the ninth row (this is the cause of the lag phenomenon). Here, when the accumulated error exceeds the threshold value, that is, when the pixel is turned on, the maximum density is output, and the error is a value obtained by subtracting the maximum density on the output side from the density on the input side to the comparator to be compared with the threshold value. Therefore, this error tends to be a large negative value in the highlight portion. And if the error is negative,
From the viewpoint of the coefficients proposed above, it can be seen that, after being distributed in the sub-scanning direction at a ratio of 9/16 as a whole, the error of the next row becomes much smaller than the previous error, which is again set to the threshold value. Several rows of errors must be accumulated before reaching. This is considered to be the cause of the lag phenomenon. Tables 2 to 4 show typical examples of this.

On the other hand, the error is 7/16 in the right pixel.
, The negative effect is weakened a few pixels ahead, and is offset by the positive error that has become stationary near the threshold value, so that the pixel is turned on again after exceeding the threshold value. This is considered to be the cause of the tailing phenomenon. In other words, it can be said that the occurrence of the trailing phenomenon and the lagging phenomenon are both caused by the distribution ratio of the error diffusion coefficient in the main scanning direction and the sub scanning direction.

From the above, (1) If a large amount of negative error after the threshold value is exceeded is distributed to the right side, the negative effect is maintained, so that the probability of neighboring pixels exceeding the threshold value is reduced, and ( 2) If the distribution ratio of the error in the sub-scanning direction is reduced, the timing at which the negative effect fades and the threshold value is reached is advanced, and the invention of claim 1 focuses on this point and describes the conventional weighting of the error distribution coefficient. The above-mentioned problem is solved by making the weight in the main scanning direction larger than that in the sub-scanning direction.

Table 3 shows an embodiment of the error distribution coefficient according to the first aspect of the present invention.

[0019]

[Table 3]

As shown in Table 3, the conventional 3 × 2 error diffusion coefficient has a main scanning direction (7/16) and a sub-scanning direction (3/1).
6 + 5/16 + 1/16 = 9/16), the relationship between (10/16) and (1/16 + 4/16) in the present embodiment.
+ 1/16 = 6/16), which is reversed.

Table 4 shows the accumulated errors when this coefficient is applied and uniform drawing is performed in the highlight portion as before. As is clear from Table 4, according to the first aspect of the present invention, pixels whose accumulated error value exceeds the threshold value, that is,
The distribution of the pixels in the ON state is uniform, so that the tailing phenomenon and the lagging phenomenon are suppressed. Note that, for the middle tone portion and thereafter, the conventional coefficients shown in Table 1 can be applied.

[0022]

[Table 4]

According to the first aspect of the present invention, the number of coefficients is 3 × 2 as in the prior art, so that a special memory circuit and arithmetic processing for the error diffusion processing of the present invention are not required. The image quality can be improved extremely efficiently.

By the way, the invention of claim 1 relates to the error diffusion coefficient of the highlight portion, but when this is applied to an image other than the highlight portion as it is, for example, to a gradation image, the difference between the image and the middle tone is obtained. Side effects of streaking at the border occur.

The reasons are as follows: (1) Since the error distribution ratio in the main scanning direction is different from that proposed in the above-mentioned document and is larger than that, the error on the boundary when moving from the highlight portion to the middle tone portion is reduced. (2) When shifting from the middle tone portion to the highlight portion, errors on the boundary tend to increase and horizontal streaks are likely to occur. It is believed that there is.

Therefore, in order to solve this problem, it is effective to apply the intermediate value of the two error diffusion coefficients as the error diffusion coefficient from the end of the highlight part to the start of the middle tone part. I can say.

According to the invention of claim 2, paying attention to this point, in the highlight portion, the error diffusion coefficient corresponding to the highlight portion is applied, and the middle portion of the middle tone portion is started from the end of the highlight portion. For the part, the intermediate value of both error diffusion coefficients was applied as the error diffusion coefficient. As a result, even in an image including a gradation other than the highlight, it is possible to obtain a high-quality output in which the gradation is smooth and uniform, regardless of the characteristics of the image.

Table 5 shows an embodiment of the error diffusion coefficient according to the second aspect of the present invention. The coefficient shown in Table 3 is used for a highlight part, and the coefficient shown in Table 1 is used for a middle tone part and thereafter. 9 shows an example of an error diffusion coefficient effective in a portion where a middle tone portion starts from the end of a highlight portion when a coefficient is applied. It goes without saying that things other than those exemplified here are also conceivable.

[0029]

[Table 5]

Next, an embodiment of the image processing apparatus according to the present invention will be described with reference to FIG. Reference numeral 1 denotes gradation data of the original image. Reference numeral 2 denotes a gradation distribution circuit which distributes the gradation data into three parts, a highlight part, a part from the end of the highlight part to the beginning of the middle tone part, and three parts after the middle tone part according to the gradation value. For example, when the maximum density is 255, the circuit distributes a density of 8 or less to a highlight portion, a density of 16 or less from the end of the highlight portion to the start of the middle tone portion, and the higher portion to a middle tone portion. Reference numeral 3 denotes an error diffusion coefficient acquisition circuit, and the grayscale data distributed according to the grayscale values here are different error diffusion coefficients, specifically,
The coefficients shown in Table 3, Table 5, and Table 1 are obtained. An error diffusion processing unit 4 performs a conventional error diffusion process by applying the obtained error diffusion coefficient.

In this image processing apparatus, the gradation data 1 of the original image
Is input, the data 1 is sorted by the sorting circuit 2 into three from the highlight portion, the end of the highlight portion to the start portion of the middle tone portion, and the middle tone portion according to the gradation value. The error diffusion coefficient obtaining circuit 3 obtains the error diffusion coefficients shown in Tables 3, 5 and 1 for the allocated gradation data in accordance with the allocated stages. The sorted gradation data is subjected to conventional error diffusion processing by applying error diffusion coefficients corresponding to the respective densities in the error diffusion processing unit 4 to obtain high-quality binary data. Can be

[0032]

According to the first aspect of the present invention, the weighting of the conventional error distribution coefficient in the main scanning direction is made larger than that in the sub-scanning direction in the error diffusion processing in the highlight portion of the gradation image. It is possible to suppress the occurrence of a tailing phenomenon and a rise delay phenomenon in a highlight portion which occurs when the method is performed, and
According to the invention of claim 2, since the intermediate value between the two error diffusion coefficients is applied as the error diffusion coefficient from the end of the highlight part to the start of the middle tone part, the gradation other than the highlight part is also included. Regardless of the image characteristics, it is possible to obtain a smooth and uniform high-quality output regardless of the image characteristics. According to the third and fourth aspects of the present invention, according to the gradation data of the original image, the highlight portion, the end portion of the highlight portion, the start portion of the middle tone portion, and the middle tone portion and the subsequent stages are assigned to different stages. The error diffusion process is performed by applying the error diffusion coefficient, which suppresses the tailing phenomenon and lag that occur in the conventional error diffusion method, and also creates a streak pattern at the boundary between the highlight and middle tone areas. It is possible to obtain high-quality binarized data with no image quality.

[Brief description of the drawings]

FIG. 1 is a block diagram of a main part showing an embodiment of an image processing apparatus according to the present invention.

[Explanation of symbols]

1 ... gradation data of original image, 2 ... gradation distribution circuit, 3
... Error diffusion coefficient acquisition circuit, 4 ... Error diffusion processing unit, 5 ... 2
Value data.

Claims (4)

[Claims] In a binarized image processing method for binarizing multi-tone data by an error diffusion method, an error diffusion coefficient in a main scanning direction of a highlight portion is distributed to be larger than a sum of error diffusion coefficients in a sub scanning direction. A binarized image processing method characterized in that the image quality of a highlight portion is efficiently improved by performing an error diffusion process by applying the error diffusion coefficient thus obtained. 2. A binary image processing method for binarizing multi-tone data by an error diffusion method, wherein an error diffusion coefficient in a portion from the end of a highlight portion to the start of a middle tone portion is determined by comparing the error diffusion coefficient between the highlight portion and the middle portion. A two-valued binary image processing method wherein an intermediate value of an error diffusion coefficient of a tone portion is used, and the error diffusion coefficient is applied to perform an error diffusion process, so that gradation is smoothly expressed and image quality is efficiently improved. Image processing method. 3. A binarized image processing method for binarizing multi-tone data by an error diffusion method, wherein a tone portion of an original image is highlighted according to a tone value, and a middle tone portion from the end of the highlight portion. At the beginning of the image and the middle tone part, and in the highlight part, the error diffusion coefficient obtained by distributing the error diffusion coefficient in the main scanning direction larger than the sum of the error diffusion coefficients in the sub-scanning direction is assigned to the highlight part. From the end to the beginning of the middle tone portion, obtain the error diffusion coefficient of the intermediate value of the error diffusion coefficient of the highlight portion and the middle tone portion, and obtain the error diffusion coefficient of the middle tone portion after the middle tone portion. The obtained error diffusion coefficient is applied to the divided gradation data to perform an error diffusion process, so that the gradation is smoothly expressed and the image quality is efficiently improved. A binarized image processing method, characterized in that: 4. A binary image processing apparatus for binarizing multi-tone data by an error diffusion method, wherein a tone portion of an original image is highlighted according to a tone value, and a middle tone portion from the end of the highlight portion. And a tone distribution circuit that distributes the error diffusion coefficients in the main scanning direction to the three stages after the middle tone portion. The error diffusion coefficients in the main scanning direction are larger than the sum of the error diffusion coefficients in the sub-scanning direction. From the end of the highlight part to the beginning of the middle tone part, the error diffusion coefficient distributed larger than the sum is the error diffusion coefficient of the intermediate value of the error diffusion coefficient of the highlight part and the middle tone part, and the middle tone part and subsequent parts. The error diffusion coefficient acquisition circuit for acquiring the error diffusion coefficient of the middle tone portion, and the erroneous grayscale data allocated by the grayscale distribution circuit. Binarization characterized by comprising an error diffusion processing unit for performing an error diffusion process by applying an error diffusion coefficient acquired by a difference diffusion coefficient acquisition circuit, to smoothly express gradation and improve image quality efficiently. Image processing device.