JPH09214760A - Picture data binarizing device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To binarize picture data without deteriorating picture quality. SOLUTION: Error data Σep is read from an error memory 20 concerning a picture element P to be considered. An adder 22 adds error data to multi- valued picture data concerning the picture element P to be considered. A random number generator 32 generates a random number value. A random number value converting circuit 34 converts the random number value in accordance with the value of a multi-valued picture data. At this time, the random number is converted so as to narrow the dynamic range of the random number value after conversion as the value of multi-valued picture data comes close to the min. value or the max. value and to widen the dynamic range as it comes close to the intermediate value. A adder 36 adds the random number after conversion to the value of picture data from the adder 22. The comparator 24 binarizes picture data after addition and outputs binary picture data. The subtractor 26 derives error (e) between picture data before binarization and picture data after binarization. The error distributing circuit 30 distributes error (e) to the four picture elements A-D.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image data binarization device for converting multi-valued image data into binary image data.

[0002]

2. Description of the Related Art Conventionally, in an ink jet printer or the like, when binarizing multi-valued image data, an error diffusion method has been adopted to perform multi-gradation recording of an image.

FIG. 12 is a block diagram showing a conventional image data binarizing apparatus adopting such an error diffusion method. As shown in FIG. 12, this image data binarization apparatus includes an error memory 120, an adder 122, a comparator 124,
The subtracter 126, the bit converter 128, and the error distribution circuit 130 are provided.

Here, the error diffusion method means that, when the image data of a certain pixel is binarized, the error between the image data before the binarization and the image data after the binarization is set at a predetermined ratio. This is a method of distributing and diffusing to each pixel in the vicinity of the pixel. FIG. 13 is an explanatory diagram for explaining the principle of such an error diffusion method. In FIG. 13A, 230
Indicates an image, the arrow 231 indicates a scanning line when scanning the image 230, and 232 indicates a pixel in the image 230.

Pixel under scanning (hereinafter referred to as a target pixel) P
When the value of the pixel data before binarization is about X and the value of the image data after binarization is about Y, the error e between the two is expressed by the following equation.

E = X−Y (1)

The value of the error e is distributed to each pixel in the vicinity of the pixel of interest P and diffused. That is, the value of the error e is divided at a predetermined ratio (error distribution ratio) and added to the image data before binarization for the neighboring pixels. As an example of the conventional diffusion processing, as shown in FIG. 13A, the value of the error e is set to the pixel A on the right of the pixel of interest P, the pixel B on the lower right, the pixel C on the lower right, and the pixel on the lower left. As shown in FIG. 13B, the error distribution rate is divided into four pixels D,
7/16 for pixel A, 1/16 for pixel B, pixel C
There is known an example in which 5/16 is applied to the above and 3/16 is applied to the pixel D.

Pixel A: e _A = e × 7/16 Pixel B: e _B = e × 1/16 (2) Pixel C: e _C = e × 5/16 Pixel D: e _D = e × 3/16

In this way, the error diffused to the neighboring pixels is accumulated and accumulated in the error memory 120 as error data for each pixel.

Now, the image data binarizing apparatus shown in FIG. 12 will be described. If the pixel of interest being scanned is P,
Multivalued image data of the pixel P is input to the adder 122. Also, the error data Σe _P for the pixel of interest _P is read from the error memory 120, and the adder 12
Entered in 2. The adder 122 adds the error data to the input multivalued image data.

The comparator 124 inputs the image data from the adder 122, compares it with a reference value which is separately input, and
1-bit data according to the magnitude relationship of those values, that is,
Output binary image data. As a result, the adder 12
The image data output from 2 is binarized. Further, the bit converter 128 converts the 1-bit binary image data from the comparator 124 into 8-bit binary image data.

In the subtractor 126, the image data (image data before binarization) from the adder 122 and the bit converter 12
Image data from 8 (image data after binarization) is input, and the image data after binarization is subtracted from the image data before binarization to derive an error e between them.

The error distribution circuit 130 distributes the error e obtained by the subtracter 126 to the pixels A, B, C and D in the vicinity of the pixel of interest P and diffuses it. That is, in the error distribution circuit 130, the pixels A, B, and
After multiplying the coefficients corresponding to C and D, respectively, the respective calculation results e _A , e _B , e _C , and e _D are converted into pixels A, E in the error memory 120.
Error data Σe _A , Σe _B , Σ for B, C and D
e _C , Σe _D , respectively, and each error data Σe _A ,
Update the values of Σe _B , Σe _C , and Σe _D.

As described above, the image data 2 shown in FIG.
According to the binarization device, even if binarization is performed by using an error diffusion method when binarizing multi-valued image data,
It is possible to represent an image in multiple gradations.

However, in the above-described conventional image data binarizing apparatus, when image data of an intermediate gradation number is input, a texture is generated in the image obtained as the binarized image data. However, there is a problem that the image quality is deteriorated. Here, the texture refers to a pattern or a pattern having some regularity included in a planar grayscale distribution.

Now, when the image data binarizing apparatus shown in FIG. 12 is adopted in a printer or the like, the number of gradations of the image data corresponds to the density in the printed image. Therefore, when image data with a relatively low gradation number is input, a print image with a relatively low density is obtained as a print result, and when image data with a relatively high gradation number is input, comparison is performed. A printed image with an extremely high density can be obtained.

FIG. 14 shows the entire density area (0% to 100%) obtained by the printer when the image data binarizing apparatus shown in FIG.
FIG. 5 is an explanatory diagram showing a printing result for (%). That is, in FIG. 14, the density changes stepwise from 0% to 100% from left to right of the image.

When the image data binarizing apparatus shown in FIG. 12 is used, as shown in FIG. 14, texture is generated and the image quality is deteriorated at an intermediate density.

Therefore, conventionally, for example, page 443 of "Journal of the Institute of Image Electronics Engineers, Volume 20, No. 5 (1991)".
As disclosed in the article “Output image characteristics by the improved error diffusion (MED) method” on page 449,
The dither data is superimposed on the input image data before binarization, and the error distribution rate at the time of distributing the value of the error e is randomly changed according to a random number, thereby suppressing the occurrence of texture in the intermediate density. Was there.

[0020]

However, in the technique disclosed in the above document, the above-described processing for suppressing the occurrence of texture is performed regardless of the number of gradations of the input image data before binarization. It is like this. Therefore, the above-mentioned processing is performed even for image data having a relatively low gradation number or a relatively high gradation number, in which the generation of the texture is originally small, and therefore, the low density area and the high density area are high. There is a problem that the image quality in the density area is significantly impaired.

The present invention has been made in view of the above-mentioned problems of the prior art, and an object thereof is to provide an image data binarizing device capable of binarizing image data without deteriorating the image quality. To provide.

[0022]

Means for Solving the Problem and Its Action / Effect To achieve at least a part of the above-mentioned object, the first invention is an image data binarizing device for binarizing multivalued image data. There is a random number generating means for generating a random number value, a random value converting means for converting the random number value according to the value of the image data before binarization, and the random number value after conversion before binarization. Random number adding means for adding to the value of the image data,
Comparing the image data after the addition of the random number value with a reference value and binarizing the image data according to the magnitude relationship of those values, the random number converting means includes the image data. The gist of the present invention is to convert the random number value such that the dynamic range of the converted random number value becomes narrower as the value of becomes closer to the minimum value or the maximum value, and becomes wider as the value becomes closer to the intermediate value.

In the first invention, the random number value converted by the random value conversion means is added to the value of the image data before binarization by the random value addition means. Since the regularity can be broken by adding the converted random number value to the value of the image data before binarization as described above, it is possible that the image data of an intermediate gradation number is input. Also, it is possible to suppress the occurrence of texture in the image obtained as the image data after binarization.

Further, in the random number conversion means, the closer the value of the image data before binarization is to the minimum value or the maximum value, the narrower the dynamic range of the converted random number value, and the closer it is to the intermediate value, the wider it becomes. To convert the random value.
By converting the random number value in this way, the above-described effect of suppressing the texture generation can be obtained for the image data having a relatively low gradation number or a relatively high gradation number in which the generation of the texture is originally small. Since the size becomes smaller, the image quality is not deteriorated even when the image data having a relatively low gradation number is input.

A second invention is an image data binarizing device for binarizing multi-valued image data, which comprises a random number generating means for generating a random number value and the image before binarizing the random number value. Random value converting means for converting in accordance with the value of the data, and the random number value after conversion is added to one of the binary value of the image data and a predetermined fixed value to obtain a reference value. The random number conversion is provided with a numerical addition means and a comparison means for comparing the image data before binarization with the reference value and binarizing the image data according to the magnitude relation of those values. The means converts the random number value such that the closer the value of the image data is to the minimum value or the maximum value, the narrower the dynamic range of the converted random number value, and the closer it is to the intermediate value, the wider the dynamic range. Use as a summary.

In the second invention, the random number value converted by the random number converting means is added to one of the image data value before binarization and a predetermined fixed value by the random number adding means. , Get the reference value. In this way, the regularity can also be broken by adding the converted random number values to obtain the reference value used when binarizing the image data.
Therefore, similarly to the first aspect, it is possible to suppress the occurrence of texture in the image.

Further, in the random number conversion means, as in the first invention, the closer the value of the image data before binarization is to the minimum value or the maximum value, the narrower the dynamic range of the converted random number value becomes, The random number value is converted so that it becomes wider as it gets closer to the intermediate value. Therefore, like the first invention,
There is no loss of image quality.

A third invention is an image data binarizing device for binarizing multi-valued image data, comprising random number generating means for generating a random number value and the image before binarizing the random number value. Random value conversion means for converting according to the value of the data to obtain a reference value, the image data before binarization is compared with the reference value, and the image data is converted according to the magnitude relation of those values. Two
Comparing means for digitizing, the random value conversion means,
The dynamic range of the random number value obtained by subtracting one of the image data value before binarization and a predetermined fixed value from the converted random number value is the minimum value or the maximum value of the image data value. The gist is to convert the random number value such that the closer it is to the value, the narrower it becomes, and the closer it is to the intermediate value, the wider it becomes.

In the third invention, the random number converting means converts the random number value to obtain the reference value. In this way, the regularity can also be broken by converting the random number value to obtain the reference value used when binarizing the image data.
Therefore, similarly to the first aspect, it is possible to suppress the occurrence of texture in the image.

Further, in the random value conversion means, the dynamic range of the random value obtained by subtracting one of the value of the image data before binarization and a predetermined fixed value from the converted random value is the image. The random number value is converted such that the closer the data value is to the minimum value or the maximum value, the narrower it becomes, and the closer it is to the intermediate value, the wider it becomes. Even by converting the random number values in this way, similar to the first aspect of the invention, for image data of a relatively low gradation number or a relatively high gradation number, which originally causes a slight occurrence of texture, Since the above-described effect of suppressing the occurrence of texture is reduced, even when image data having a relatively low gradation number or a relatively high gradation number is input, the image quality is not deteriorated as in the conventional case.

Furthermore, in the third invention, the random value adding means is not required as compared with the first or second invention, so that the number of constituent elements can be reduced accordingly.

[0032]

Other Embodiments of the Invention The present invention can also adopt the following other embodiments. That is, in the first mode, multi-valued image data obtained by scanning the image pixel by pixel
An image data binarizing device for binarizing, wherein an error storage unit for storing error data for each pixel and the error data read out from the error storage unit for a pixel of interest in scanning in the image Error data addition means for adding to the image data before binarization, random number generation means for generating a random number value, random number conversion means for converting the random number value according to the value of the image data before addition, and conversion The random number value adding means for adding the subsequent random number value to the value of the image data after addition, and the image data after adding the random number value with a reference value, and depending on the magnitude relation between the values, A comparison means for binarizing the image data, an error deriving means for deriving an error between the image data before the addition of the random number value and the image data after the binarization, and the error of the target pixel at a predetermined ratio. Peripheral pixels Error distribution means for distributing and accumulating the error data in the error storage means for each pixel, wherein the random value conversion means converts as the value of the image data approaches a minimum value or a maximum value. The gist of the present invention is to convert the random number value so that the dynamic range of the random number value afterwards becomes narrower and becomes closer to an intermediate value.

A second aspect is an image data binarizing device for binarizing multi-valued image data obtained by scanning an image pixel by pixel, storing error data for each pixel. Error storage means, error data addition means for adding the error data read from the error storage means to the image data before binarization, and a random value for a pixel of interest in scanning in the image. Random number generating means,
Random value conversion means for converting the random number value according to the value of the image data before addition, and adding the random number value after conversion to one of the value of the image data before addition and a predetermined fixed value Then, a random number addition means for obtaining a reference value, a comparison means for comparing the image data after addition with the reference value, and binarizing the image data according to the magnitude relation of those values, and addition An error deriving means for deriving an error between the subsequent image data and the binarized image data, and the error is distributed to pixels around the pixel of interest at a predetermined ratio, and the error is calculated for each pixel. Error distribution means for integrating the error data in the storage means, the random number conversion means, the closer the value of the image data to the minimum value or the maximum value, the dynamic range of the random number value after conversion. Becomes narrower and close to the middle value Ruhodo, so that wide, and be required to convert the random number value.

Further, as a third aspect, an image data binarizing device for binarizing multi-valued image data obtained by scanning an image pixel by pixel, storing error data for each pixel. Error storage means, error data addition means for adding the error data read from the error storage means to the image data before binarization, and a random value for a pixel of interest in scanning in the image. Random number generating means, the random number value is converted according to the value of the image data before addition, a random value conversion means for obtaining a reference value, and the image data after addition is compared with the reference value, A comparing means for binarizing the image data according to the magnitude relation of the values, an error deriving means for deriving an error between the image data after the addition and the image data after the binarization, and the error is predetermined. Attention ratio Error distribution means that distributes to each of the peripheral pixels and accumulates the error data in the error storage means for each pixel, wherein the random value conversion means calculates from the converted random number value before addition. The dynamic range of the random number value obtained by subtracting one of the value of the image data and a predetermined fixed value is narrower as the value of the image data is closer to the minimum value or the maximum value, and is closer to the intermediate value. The gist is to convert the random number value so that it becomes wider.

As described above, after preparing the error storage means for storing the error data for each pixel, the error between the image data before binarization and the image data after binarization is derived, and the error is noted. The data is distributed to the pixels around the pixel, and the error data in the error storage means is integrated for each pixel, while the error data read from the error storage means is converted into image data before binarization for the target pixel. By adding, error diffusion can be realized, and even if binarized, the image can be expressed in multiple gradations.

[0036]

BEST MODE FOR CARRYING OUT THE INVENTION Embodiments of the present invention will be described below based on Examples. FIG. 1 is a block diagram showing an image data binarizing apparatus as a first embodiment of the present invention.
As shown in FIG. 1, this image data binarizing apparatus includes an error memory 20, an adder 22, a comparator 24, a subtractor 26, a bit converter 28, an error distribution circuit 30, a random number generator 32, and a random number generator 32.
A random number conversion circuit 34 and an adder 36 are provided.

Now, the image data binarization apparatus of this embodiment will be described in detail with reference to FIG. Error memory 20
Is a memory for accumulating error data for each pixel. Although the error memory 20 has a storage capacity of only two lines of an image as shown in FIG. 1, by using the storage area of these two lines by a toggle, the error data for each pixel of one image can be stored. It can be accumulated and accumulated in order.

On the other hand, in FIG. 1, when the pixel of interest being scanned is P, multivalued image data for the pixel P is input to the adder 22. The multi-valued image data is, for example, 8-bit data, and the data value (that is, the number of gradations) can take values of “0” to “255”. The multi-valued image data is also input to the random value conversion circuit 34.

From the error memory 20, the pixel of interest P
For, it the error data Sigma] e _P that has been dispensed from pixels near the integrated storage until are then input to the adder 22. This error data is, for example, 1 for 8-bit data.
It is a total of 9-bit data with bit code data added. Incidentally, the error data Sigma] e _P is read, the error data Sigma] e _P accumulated in the error memory 20 is updated to 0.

The adder 22 adds the error data to the input multi-valued image data to obtain 9-bit image data (8
(Bit data + 1 bit code data) is output.

On the other hand, in the random number generator 32, "0" to "1"
A random number value of 28 ″ is generated and input to the random number value conversion circuit 34. In the random number value conversion circuit 34, the input random number value is
It is converted according to the value of the separately input multi-valued image data and is output. At this time, the conversion is performed so that the dynamic range of the output random number value (that is, the range in which the random number value can be taken) changes in accordance with the value of the multi-valued image data.

For example, assuming that the input random number value is R, the value of multi-valued image data is S, and the output random number value after conversion is R ', the random number value R'is given by It is represented by (3).

[0043]

(Equation 1)

Here, MIN is a function that selects the minimum value of the two values separated by commas in parentheses.

By converting the random number value R according to the equation (3), the dynamic range of the converted random number value R'changes as follows according to the value of the multivalued image data.

FIG. 2 is a characteristic diagram showing a change in the dynamic range of the converted random number value R'in the case of the equation (3). In FIG. 2, the horizontal axis represents the value S of multivalued image data, and the vertical axis represents the dynamic range of the converted random number value R ′. As shown in FIG. 2, the dynamic range of the random number value R ′ after conversion is “0” to “0” depending on the value S of the image data.
It changes like a triangular wave between "128". That is, the closer the image data value S is to the minimum value "0" or the maximum value "255", the narrower the dynamic range of the random number value R ', and the closer it is to the intermediate value "128", the wider it becomes. Therefore, for example, the value S of the input multivalued image data is
, The dynamic range of the random number value R'is "0" to "β" (the range indicated by the vertical solid line), so that the converted random number value R'is , Any value in the range of “0” to “β” corresponding to the input random number R is output.

Note that such a random value conversion circuit 34 can be easily configured by a look-up table or the like. Specifically, a memory having the random number value R and the image data value S as address inputs is prepared, and the random number value R ′ corresponding to each address is stored.

Next, in the adder 36, the random number value R ′ output from the random value conversion circuit 34 is added to the value of the image data output from the adder 22 to output 9-bit image data. .

In the comparator 24, the image data from the adder 36 is input, compared with a reference value input separately, and 1-bit data, that is, binary image data, depending on the magnitude relation between these values. Is output. Here, "128" which is an intermediate value between "0" and "255" is adopted as the reference value. That is, the comparator 24 outputs "1" when the value of the image data from the adder 36 is "128" or more,
If it is smaller than "128", "0" is output. In this way, the image data output from the adder 36 is binarized.

The bit converter 28 inputs the 1-bit binary image data from the comparator 24, converts it into 8-bit binary image data, and outputs it. That is, when the value of the image data from the comparator 24 is “0”, the data of the value “0” is output as 8-bit image data, and when the value of the image data from the comparator 24 is “1”. Outputs data of value "255" as 8-bit image data. In addition,
Such a bit converter 28, for example, branches the 1-bit image data line branched from the output of the comparator 24 into eight lines, and connects them to the input of the subtractor 126 as an 8-bit image data line. This can be easily realized.

In the subtractor 26, the image data from the adder 22 (image data before binarization) and the image data from the bit converter 28 (image data after binarization) are input, and 2
The image data after binarization is subtracted from the image data before binarization to derive the error e between them.

In the error distribution circuit 30, the error e obtained by the subtractor 26 is distributed to the pixels A, B, C and D near the target pixel P and diffused, as described with reference to FIG. That is, in the error distribution circuit 30, for example, the pixels A, B, and
After multiplying the coefficients corresponding to C and D, respectively, the respective calculation results e _A , e _B , e _C , and e _D are converted into pixels A, E in the error memory 120.
Error data Σe _A , Σe _B , Σ for B, C and D
e _C , Σe _D , respectively, and each error data Σe _A ,
Update the values of Σe _B , Σe _C , and Σe _D.

In this way, when the series of processes for the pixel of interest P is completed, the scan moves from the pixel P to the next pixel A, and this time, the same process is repeated with the pixel A as the pixel of interest being scanned. .

In the above example, the random number conversion circuit 34 converts the random value R according to the equation (3), but it may be converted according to the following equation (4).

[0055]

(Equation 2)

However, in the equation (4), as in the equation (3), R is the input random number value, S is the value of multi-valued image data, and R'is the output random number. Numerical values are shown respectively.

As described above, by converting the random number value R according to the equation (4), the dynamic range of the converted random number value R'changes as shown in FIG. 3 according to the value of the multivalued image data. To do.

FIG. 3 is a characteristic diagram showing a change in the dynamic range of the converted random number value R'in the case of the equation (4). In FIG. 3, as in FIG. 2, the horizontal axis represents the value S of multivalued image data, and the vertical axis represents the dynamic range of the converted random number value R ′. As shown in FIG. 3, the dynamic range of the converted random number value R ′ changes in a sine wave form between “0” and “128” according to the value S of the image data. That is, as in the case of FIG. 2, the closer the value S of the image data is to the minimum value “0” or the maximum value “255”, the narrower the dynamic range of the random number value R ′ and the intermediate value “12”.
As the value S of the input multivalued image data is “α” as shown in FIG. 3, the converted random number value R ′ is
Any value in the range of “0” to “γ” corresponding to the input random number value R is output.

As described above, according to this embodiment,
Since the adder 36 can break the regularity by adding the converted random number value R ′ to the value of the image data, even when the image data of the intermediate gradation number is input, It is possible to suppress the occurrence of texture in an image obtained as image data after binarization.

FIGS. 4 and 5 are explanatory views showing the printing results for the entire density area (0% to 100%) by the printer when the image data binarizing apparatus of FIG. 1 is used. That is, in FIG. 4 or 5, the density changes stepwise from 0% to 100% from left to right of the image. Of these, FIG. 4 shows a case where the random number value R is converted according to the equation (3), and FIG. 5 shows a case where the random number value R is converted according to the equation (4).

The printing in FIG. 4 or FIG. 5 is performed by scanning each line laterally from left to right or from left to right starting from the upper left corner of the image as shown in FIG. Be done. This also applies to the print result described later.

As shown in FIG. 4 or FIG. 5, when the image data binarizing apparatus shown in FIG. 1 is used, the generation of texture at intermediate density as shown in FIG. 14 of the related art is hardly seen.

Further, according to the present embodiment, the random number generator 32
The dynamic range of the converted random number value R ′ is shown in FIG. 2 or FIG. 3 by converting the random number value R from ## EQU1 ## according to the expression (3) or the expression (4) according to the value S of the multivalued image data As shown, the closer the image data value (gradation number) S is to the minimum value "0" or the maximum value "255", the narrower it becomes. That is, for image data with a relatively low number of gradations or a relatively high number of gradations where the generation of texture is originally small,
Since the above-described effect of suppressing the occurrence of texture is reduced, even when image data having a relatively low gradation number is input, the image quality is not deteriorated as in the conventional case.

FIGS. 6 and 7 are explanatory views showing the printing results for the density of 10% by the printer when the image data binarizing apparatus of FIG. 1 is used. That is, in FIG. 6 or 7, the entire image has uniform density.
Of these, FIG. 6 shows a case where the random number value R is converted according to the equation (3), and FIG. 7 shows a case where the random number value R is converted according to the equation (4).

At relatively low densities, the black dots are evenly distributed as shown in FIG. 6 or 7, and the image distortion due to the texture removal processing is hardly recognized. Further, although relatively high densities are not shown as a print result as shown in FIG. 6 or 7, white dots are scattered evenly as in the case of low densities, and the image is disturbed by the texture removal process. Is hardly recognized. Therefore, the image quality is not impaired even at a relatively low density or a high density.

In order to further confirm the effect of the present embodiment described above, in FIG. 1, the dynamic range of the random number value is converted into multi-valued image data without converting the random number value R from the random number generator 32. The result when the value is added to the value S of the image data while being kept constant regardless of the value S of is shown below. FIG. 8 is an explanatory diagram showing a printing result for the entire density region (0% to 100%) by the printer when the dynamic range of random number values is fixed. That is, in FIG. 8, the density changes stepwise from 0% to 100% from left to right of the image. Further, FIG. 9 shows a printing result for a density of 10% among all the density areas shown in FIG. That is,
In FIG. 9, the entire image has a uniform density.

As shown in FIG. 8 or FIG. 9, the disturbance of the image due to the texture removal processing is noticeable at both the relatively low density and the relatively high density.

Therefore, in the case where the dynamic range of the random number value is fixed and the value is not changed according to the value of the image data as shown in FIG.
It can be seen that the image quality is deteriorated at a relatively low density or a high density, as in the conventional case.

Now, in the above-mentioned first embodiment,
Although the converted random number value R ′ is added to the value of the image data input to one input of the comparator 24, instead of adding the converted random number value R ′ to the value of the image data, the converted random number value R ′ is a fixed value. , And the obtained value may be input to the other input of the comparator 24 as a reference value. Such an embodiment will be described with reference to FIG.

FIG. 10 is a block diagram showing an image data binarizing apparatus as a second embodiment of the present invention. As shown in FIG. 10, this image data binarization device includes an error memory 20, an adder 22, a comparator 24, a subtractor 26, a bit converter 28, an error distribution circuit 30, a random number generator 32, and a random number value conversion. The circuit 34 and the adder 38 are provided.

In the image data binarizing apparatus shown in FIG. 10, only parts different from those in FIG. 1 will be described, and description of the same parts will be omitted. As shown in FIG. 10, in the present embodiment, the random number value R ′ converted and output by the random number value conversion circuit 34 is added to a fixed value in the adder 38, and the obtained value R ″ is used as a reference value. Input to the comparator 24. Here,
The fixed value is "1" which is the same as the reference value in the first embodiment.
28 "is adopted.

As described above, the converted random number value R ′ may be superimposed on the reference value input to the comparator 124 without being added to the image data value input to the comparator 124. The regularity can be broken as in the first embodiment. Therefore, even when image data with an intermediate number of gradations is input, it is possible to suppress the occurrence of texture in an image obtained as binarized image data.

Also in the present embodiment, the above-described effect of suppressing the texture generation is effective for the image data having a relatively low gradation number or a relatively high gradation number in which the generation of texture is originally small. Since the size is small, the image quality is not deteriorated as in the first embodiment.

Although the random number conversion circuit 34 and the adder 38 are configured as separate circuits in this embodiment, the random number conversion circuit 34 adds the random number value R'to the fixed value "128". The adder 38 can be omitted if it is included in the conversion process of the random number value in. That is, in the random number conversion circuit 34, the random number value R may be converted according to the following equation (5) or equation (6) to obtain the random number value R ″.

[0075]

(Equation 3)

[0076]

[Equation 4]

Further, in the second embodiment described above,
Although the converted random number value R'is added to the fixed value and the obtained value is input to the comparator 24 as the reference value, the converted random number value R'is not added to the fixed value. The value S may be added to the value S of the image data, and the obtained value may be input to the comparator 24 as a reference value. Such an embodiment will be described with reference to FIG.

FIG. 11 is a block diagram showing an image data binarizing apparatus as a third embodiment of the present invention. As shown in FIG. 11, this image data binarization device includes an error memory 20, an adder 22, a comparator 24, a subtractor 26, a bit converter 28, an error distribution circuit 30, a random number generator 32, and a random number value conversion. The circuit 34 and the adder 38 are provided.

In FIG. 11, the image data binarization apparatus will be described only with respect to portions different from those in FIG. 10, and description of the same portions will be omitted. As shown in FIG. 11, in the present embodiment, the random number value R ′ converted and output by the random number value conversion circuit 34 is not the fixed value in the adder 38, but the value of the input multivalued image data. It is added to S and the obtained value R ′ ″ is input to the comparator 24 as a reference value.

As described above, even if the value S of the image data is used as the reference value and the converted random number value R'is superposed on the reference value, the regularity is destroyed as in the second embodiment. You can Therefore, even when image data with an intermediate number of gradations is input, it is possible to suppress the occurrence of texture in an image obtained as binarized image data.

Also in the present embodiment, the above-described effect of suppressing the texture generation is effective for the image data having a relatively low gradation number or a relatively high gradation number in which the generation of texture is originally small. Since the size is small, the image quality is not deteriorated as in the first embodiment.

In the present embodiment as well, the random number value conversion circuit 34 and the adder 38 are configured as separate circuits, but the addition of the random number value R ′ and the image data value S is performed by the random number value conversion circuit 34. The adder 38 can be omitted if it is included in the conversion process of the random number value in. That is, in the random number conversion circuit 34, the random number value R may be converted according to the following equation (7) or equation (8) to obtain the random number value R ′ ″.

[0083]

(Equation 5)

[0084]

(Equation 6)

The present invention is not limited to the above-described examples and embodiments, but can be implemented in various modes without departing from the scope of the invention.

[Brief description of drawings]

FIG. 1 is a block diagram showing an image data binarizing apparatus as a first embodiment of the present invention.

FIG. 2 is a characteristic diagram showing a change in a dynamic range of a random number value R ′ after conversion in the case of following Expression (3).

FIG. 3 is a characteristic diagram showing a change in a dynamic range of a random number value R ′ after conversion in the case where the equation (4) is followed.

FIG. 4 is a total density area (0% to 100%) obtained by a printer when a random number value is converted according to Expression (3) in FIG.
FIG. 7 is an explanatory diagram showing a printing result of the.

5 is a total density area (0% to 100%) obtained by the printer when the random number value is converted according to the equation (4) in FIG.
FIG. 6 is an explanatory diagram showing a printing result of the.

FIG. 6 is an explanatory diagram showing a printing result for a density of 10% by the printer when the random number value is converted according to the equation (3) in FIG.

FIG. 7 is an explanatory diagram showing a printing result for a density of 10% by the printer when the random number value is converted according to the equation (4) in FIG.

FIG. 8 is an explanatory diagram showing a printing result for the entire density region (0% to 100%) by the printer when the dynamic range of random number values is fixed.

FIG. 9 is an explanatory diagram showing a printing result for a density of 10% by the printer when the dynamic range of random number values is fixed.

FIG. 10 is image data 2 as a second embodiment of the present invention.
It is a block diagram which shows a quantizer.

FIG. 11 is image data 2 as a third embodiment of the present invention.
It is a block diagram which shows a quantizer.

FIG. 12: Conventional image data 2 employing an error diffusion method
It is a block diagram which shows a quantizer.

FIG. 13 is an explanatory diagram for explaining the principle of the error diffusion method.

FIG. 14 is an explanatory diagram showing a printing result for the entire density area (0% to 100%) by the printer when the image data binarizing apparatus of FIG. 12 is used.

[Explanation of symbols]

20 ... Error memory 22 ... Adder 24 ... Comparator 26 ... Subtractor 28 ... Bit converter 30 ... Error distribution circuit 32 ... Random number generator 34 ... Random value conversion circuit 36 ... Adder 38 ... Adder 120 ... Error memory 122 Adder 124 Comparator 126 Subtractor 128 Bit converter 130 Error distribution circuit 230 Image 231 Scan line 232 Pixel A, B, C, D Pixel P Pixel of interest e _A , e _B e _C , e _D ... Distribution error e ... Error Σe _A , Σe _B , Σe _C , Σe _D・ ・ ・ Error data Σe _P・ ・ ・ Error data

Claims (3)

[Claims] 1. An image data binarization device for binarizing multi-valued image data, comprising: a random number generating means for generating a random number value; and the random number value being a value of the image data before binarization. Random value conversion means for converting the random number value according to the following: random number value adding means for adding the converted random number value to the value of the image data before binarization, and comparing the image data after adding the random number value with a reference value And a comparing means for binarizing the image data according to the magnitude relation of the values, and the random number converting means converts the image data closer to the minimum value or the maximum value. An image data binarization device, characterized in that the random number value is converted so that the dynamic range of the subsequent random number value becomes narrower and the dynamic range becomes wider as it approaches an intermediate value. 2. An image data binarizing device for binarizing multi-valued image data, comprising random number generating means for generating a random number value, and the random number value being the value of the image data before binarization. Random number conversion means for converting in accordance with the above, and the random number addition means for obtaining the reference value by adding the converted random number value to one of the binary value of the image data and a predetermined fixed value. Comparing means for comparing the image data before binarization with the reference value and binarizing the image data according to the magnitude relation of the values, the random number value converting means An image characterized by converting the random number value such that the closer the value of the image data is to the minimum value or the maximum value, the narrower the dynamic range of the converted random number value is, and the closer it is to the intermediate value, the wider the dynamic range is. Data binarizer. 3. An image data binarizing device for binarizing multi-valued image data, comprising random number generating means for generating a random number value, and the random number value being the value of the image data before binarization. Random value conversion means for converting the image data to obtain a reference value, and the image data before binarization are compared with the reference value, and the image data is binarized according to the magnitude relation of those values. Comparing means, wherein the random value conversion means is configured to subtract one of the value of the image data before binarization and a predetermined fixed value from the converted random value of the random value. Dynamic range
An image data binarization device, wherein the random number value is converted such that the value of the image data becomes narrower as it approaches a minimum value or a maximum value, and becomes narrower as it approaches an intermediate value.