JP3475425B2 - Image processing apparatus, image processing method thereof, and recording medium recording image processing program - Google Patents

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, an image processing method therefor, and a recording medium on which an image processing program is recorded. In particular, error diffusion processing is performed when original data has a gradation characteristic reduced by gamma correction. The present invention relates to an improved image processing method.

[0002]

2. Description of the Related Art Conventionally, in an output device such as a printer or a display, gamma correction for converting original image data according to its characteristics is performed. For example, in printers, the dot diameter of one dot is set to be larger than the pixel interval for the purpose of eliminating horizontal stripes called banding caused by problems of paper conveyance accuracy and dot landing accuracy.

Therefore, one dot that exists independently in the low-density area is projected to the surrounding pixels whose pixels are OFF (blank), and becomes larger than the ideal density for the original image data without correction. . In order to solve the problem, gamma correction is a correction of characteristics opposite to those of the output device.

Normally, gamma correction is performed prior to halftone processing such as an error diffusion method or a dither method, which is essential in an output device that cannot output all gradations of original image data, to reduce the number of gradations of pixels. .

In the error diffusion method, a quantization error caused by gradation reduction (hereinafter referred to as quantization) of each pixel is diffused and added to surrounding pixels which have not been quantized yet, and then compared with a threshold value. This is a halftone method that can reduce the quantization error in the entire image by quantizing.

However, if the normal halftone processing is performed after the gamma correction processing, there arises a problem that the number of gradations of the original image is not effectively utilized. When the output becomes larger than the ideal value as in the above example of the printer, gamma correction is performed, for example, "0, 1, 2" of the original image data becomes "0, 0, 0", and "3. , 4, 5 ”is“ 1, 1, 1 ”and“ 6, 7, 8 ”is“ 3, 3, 3 ”
, ..., "254" is converted to "250", and "255" is converted to "255".

Halftone in a state in which the information of the subtle gradation which the original image has is lost (for example, "1, 2, 3" is compressed into one output gradation value "0"). As a result, the problem arises that details are destroyed or pseudo contours are generated in the output image.

In order to solve this problem, in the error diffusion method, the output evaluation value (in this case, quantization error = pre-quantization data value−output evaluation value) corresponding to the two-gradation image data indicating dot ON is calculated. There is a method of responding by changing the gamma characteristic.

That is, if the output evaluation value is made larger than the actual quantized data value, the quantization error becomes smaller, and the pixel in which this reduced quantization error is diffused has a smaller value. And the dot O when quantized
It is possible to control the generation of dots so that N is less likely to occur.

By making the degree of change of the output evaluation value correspond to the gamma curve, it becomes possible to make the gamma characteristic correspond to the dot generation. For the above method,
It is disclosed in Japanese Patent Laid-Open No. 9-163141.

[0011]

In the above-described conventional image processing method, the characteristics of the gamma curve are reflected in the pixels in the vicinity of the pixel of interest because of the quantization error generated from the pixel of interest. That is, there is a problem that the pixel that is actually gamma-corrected is not the pixel of interest.

Although there are few problems in a region where pixels of uniform density are continuous, there is a problem that optimum gamma correction is not performed in a very small region or an edge portion. Further, the method described in the above publication only mentions the case where the dots are binary values of ON and OFF, and does not deal with the case where multivalued halftone is performed.

Therefore, an object of the present invention is to solve the above-mentioned problems, and to perform an error diffusion process that does not reduce the gradation characteristics by gamma correction, an image processing apparatus, an image processing method thereof, and an image processing program. It is to provide a recording medium in which is recorded.

[0014]

An image processing apparatus according to the present invention comprises an error weight matrix storage unit in which a matrix for weighting and diffusing a quantization error to neighboring pixels to diffuse is set and stored, and Quantization error buffer for sequentially storing the error when the pixel is quantized, weighted averaging means for weighting and averaging the quantized error of the peripheral pixel by the error weight matrix of the peripheral pixel, and setting a threshold value for quantizing the pixel of interest. And a threshold value setting means for comparing the threshold value with a weighted average value of quantization errors of pixels in the vicinity of the pixel of interest and an added value of the values of the pixel of interest to quantize into a multi-valued binary value or more. And an image processing device for performing multi-valued error diffusion processing, including: a gamma correction unit for gamma-converting original image data according to a predetermined gamma characteristic; And a threshold value correction means for correcting the threshold value by referring to the after gamma conversion by the gamma correction unit data and the original image data.

In the image processing method according to the present invention, an error weight matrix storage means in which a matrix for weighting and quantizing the quantization error to neighboring pixels and diffusing it is stored in advance, and an error when quantized. A quantization error buffer that sequentially stores, a weighted average processing unit that weights and averages the quantization error of the peripheral pixels by the error weight matrix of the peripheral pixels, and a threshold setting unit that sets a threshold value when the pixel of interest is quantized. A quantizing means for comparing the threshold value with a weighted average value of quantization errors of pixels in the vicinity of the pixel of interest and an added value of the values of the pixel of interest to quantize into a multivalued binary or higher value; An image processing method of an image processing device for performing multi-valued error diffusion processing, comprising the steps of gamma conversion of original image data according to a predetermined gamma characteristic, and the gamma conversion thereof. And a step of modifying the threshold value with reference to the 換後 data and the original image data.

The recording medium recording the image processing program according to the present invention includes an error weight matrix storage unit in which a matrix for weighting and diffusing a quantization error to neighboring pixels and diffusing it is stored, and a quantization. Quantization error buffer for sequentially storing the errors in the case of being processed, weighted averaging means for weighting and averaging the quantization errors of the peripheral pixels by the error weight matrix of the peripheral pixels, and a threshold value for quantizing the pixel of interest are set. The threshold value setting means compares the threshold value with the weighted average value of the quantization errors of the pixels in the vicinity of the target pixel, and the added value of the values of the target pixel, and quantizes to quantize into a binary value or more. A recording medium storing an image processing program for causing an image processing apparatus to perform a multi-valued error diffusion process, the image processing program comprising: Sense device, and according to the gamma characteristic which has been predetermined the original image data is gamma-converted, to correct the threshold value with reference to its after gamma conversion data and the original image data.

That is, in order to solve the above-mentioned problems, the image processing apparatus of the present invention controls the output state of dots by changing the threshold value in the error diffusion method according to the gamma characteristic, and performs gamma correction. Binary or higher multi-valued error diffusion processing that does not reduce gradation characteristics is performed.

More specifically, the image processing apparatus of the present invention is
Means for converting the original image data according to the gamma characteristic (hereinafter referred to as gamma conversion), means for setting a threshold value from the gamma converted data, and means for correcting the threshold value from the original image data and the gamma converted data. And have.

The means for correcting the threshold value from the original image data and the gamma-converted data refers to the gamma-converted data obtained by the gamma conversion means and the original image data, and the gamma-converted data has the same value. If the data values are different, the threshold value of the error diffusion process is modified so that the darker grayscale of the original data is quantized. by this,
By the gamma correction, it is possible to realize the error diffusion processing without the gradation characteristic being reduced.

[0020]

BEST MODE FOR CARRYING OUT THE INVENTION Next, an embodiment of the present invention will be described with reference to the drawings. FIG. 1 is a block diagram showing the arrangement of an image processing apparatus according to an embodiment of the present invention. 1, an image processing apparatus according to an embodiment of the present invention includes a gamma correction unit 1, a threshold value setting unit 2, a threshold value correction unit 3, a quantization error buffer 4, an error weight matrix storage unit 5, and a quantization unit. Unit 6, weighted average processing unit 7, subtraction processing unit 8,
It is composed of an addition processing unit 9.

The gamma correction unit 1 gamma-converts the original image data of the pixel of interest according to a predetermined gamma characteristic. The threshold setting unit 2 sets a threshold for quantizing the data based on the value of the data after gamma conversion. There are many possible methods for setting this threshold value, but the simplest method is the intermediate value of the two output levels that are closest to the value of the data after gamma conversion.

Further, in order to solve the drawbacks of the ordinary error diffusion method, such as the delay of dots appearing in the ultra-low density region and the ultra-high density region and the dot line, the data value and output after gamma conversion It is also possible to change the threshold depending on the difference from the level value. However, since the threshold setting method is not the most desired feature of the present invention,
Detailed description thereof will be omitted.

The threshold correction unit 3 refers to the gamma-converted data and the original image data to correct the threshold value of the error diffusion process.
The threshold is modified so that if the data after gamma conversion has the same value but the value of the original data is different, the darker original data is quantized into a deeper gradation. This threshold correction unit 3 is the most characteristic of the present invention.

The quantization error buffer 4 sequentially stores the pixel of interest and the error when the pixel is quantized. The error weight matrix storage unit 5 has a matrix (hereinafter referred to as an error weight matrix) for weighting and diffusing the quantization error in the same error diffusion processing as in the conventional case to neighboring pixels.
It is preset and stored.

The quantizer 6 compares the corrected threshold value with a value obtained by adding the weighted average of the quantization errors of the pixels in the vicinity of the pixel of interest and the value of the pixel of interest to obtain a binary value or more multivalue. Quantize. This is similar to the comparison process in the conventional error diffusion process.

The weighted average processing unit 7 performs a weighted average on the quantization error of the neighboring pixels stored in the quantization error buffer 4 by using the error weight matrix of the neighboring pixels stored in advance in the error weight matrix storage unit 5. Then, the weighted average value of neighboring pixel errors is calculated. The subtraction processing unit 8 performs a subtraction process between the quantized data of the neighboring pixels obtained by the quantizing unit 6 and the gamma-corrected data of the neighboring pixels from the gamma correction unit 1 to calculate the quantization error. The weighted average processing unit 9 calculates the weighted average of the quantization error of the peripheral pixels by using the error weight matrix of the corresponding peripheral pixels.

FIG. 2 is a flow chart showing the operation of the image processing apparatus according to one embodiment of the present invention, and FIG. 3 is a flow chart showing the quantization processing of neighboring pixel data of the image processing apparatus according to one embodiment of the present invention. . The operation of the image processing apparatus according to the embodiment of the present invention will be described with reference to FIGS.

The processing operations shown in FIGS. 2 and 3 are realized by each unit of the image processing apparatus according to the embodiment of the present invention executing a program recorded in a control memory (not shown). ROM (Read Only Memory)
An IC (integrated circuit) memory or the like can be used.

The data of the pixel of interest input to the gamma correction unit 1 is gamma-converted according to a predetermined gamma characteristic (hereinafter, the gamma-converted data is referred to as gamma-converted data) (steps S1 and S2 in FIG. 2). ), The gamma-converted data is sent to the threshold setting unit 2, the threshold correction unit 3, the subtraction processing unit 8, and the addition processing unit 9.

The threshold setting unit 2 sets a threshold for quantizing the data based on the value of the gamma-converted data (step S3 in FIG. 2) and sends the threshold to the threshold correcting unit 3. The threshold value input to the threshold value correction unit 3 is corrected by referring to the gamma-converted data from the gamma conversion unit 1 and the original image data, and a correction process is performed based on the above-mentioned correction reference (step S4 in FIG. 2). The threshold value is sent to the quantizer 6.

The weighted average processing unit 7 performs a weighted average on the quantization error of the neighboring pixels stored in the quantization error buffer 4 by using the error weight matrix of the neighboring pixels stored in advance in the error weight matrix storage unit 5. Then, the weighted average value of neighboring pixel errors is calculated (step S in FIG. 2).
5), and sends the calculation result to the addition processing unit 9.

The addition processing unit 9 adds the neighboring pixel error weighted average value from the weighted average unit 7 and the value of the gamma-corrected data of the pixel of interest from the gamma correction unit 1 (step S in FIG. 2).
6), and sends the addition result to the quantizer 6.

The quantizing unit 6 calculates the corrected threshold value from the threshold correcting unit 3 and the addition result from the addition processing unit 9, that is, the weighted average of the quantization error of the pixels in the vicinity of the pixel of interest from the addition processing unit 7. Compare the value of the pixel of interest with the added value,
It is quantized into multi-values of binary or higher and output as quantized data (step S7 in FIG. 2).

When the data of all the pixels are quantized by the quantization processing of the quantization unit 6 (step S8 in FIG. 2), the above processing is ended. In addition, if the data of all pixels are not quantized by the quantization processing of the quantization unit 6 (see FIG.
In step S8), the target pixel is switched to the adjacent pixel (step S9 in FIG. 2), the process returns to step S2, and the above processing operation is repeated. The operation other than the operation related to the threshold correction unit 3 is the same as that of the conventional error diffusion processing, and a detailed description thereof will be omitted.

In the image processing apparatus according to the embodiment of the present invention, prior to the above-described quantization processing of the data of the pixel of interest, the same quantization processing is executed on the data of the pixels in the vicinity of the pixel of interest, and the quantization is performed. The quantization error calculated in the process is stored in the quantization error buffer 4.

That is, the quantized data of the neighboring pixels obtained by the quantizing unit 6 is sent to the subtraction processing unit 8 and subjected to the subtraction process with the gamma-corrected data of the neighboring pixels from the gamma correction unit 1 to perform the quantization. Error is calculated (step S1 in FIG. 3).
1), the quantization error is sequentially stored in the quantization error buffer 4 (step S12 in FIG. 3). This quantization processing is repeatedly performed until the data of all pixels are quantized (steps S11 to S13 in FIG. 3).

As described above, in the threshold correction unit 3, if the value of the original data is different from the threshold value in the error diffusion processing even if the data after gamma conversion is the same value, the darker the original data is, the darker the data is. By performing so that the tone characteristics are quantized, it is possible to realize the error diffusion processing in which the gradation characteristics are not reduced by the gamma correction.

Next, the operation of the image processing apparatus according to one embodiment of the present invention will be described in detail with reference to the drawings using a specific example. In this embodiment, for example, image data (0-255, the larger the value, the darker) the image data obtained by a scanner or a digital camera is quantized into four values (0, 85, 170, 255) for a printer having four gradations. Is an image processing device.

In FIG. 1, the gamma correction unit 1 gamma-converts the original image data of the pixel of interest according to a predetermined gamma characteristic. The relationship between the original image data x and the gamma-converted data γ (x) in this embodiment is Suppose Such a conversion table (not shown) is stored in the gamma correction unit 1, and the image data is converted based on this conversion table.

In this case, after the gamma change, "1, 2, 3,
All the data of "4" becomes "0", "5, 6, 7, 8,
So that all the data "9" becomes "1",
The information of the subtle gradation is lost. If the normal halftone process is performed in this state, there arises a problem that details are crushed or a pseudo contour is generated in the output image.

The threshold setting unit 2 sets a threshold for quantizing the data based on the value of the data after gamma conversion. In the threshold value setting method of this embodiment, the intermediate value of the two output levels closest to the value of the data after gamma conversion is used.

That is, the gamma-converted data γ (x) and the threshold value th (γ (x)) are as follows: 0 ≦ γ (x) <85 th (γ (x)) = 43 85 ≦ γ (x) When 170 <th (γ (x)) = 128 When 170 ≦ γ (x) ≦ 255, th (γ (x)) = 213.

The threshold correction unit 3 refers to the gamma-converted data and the original image data to correct the threshold value of the error diffusion process.
When the gamma-converted data has the same value and the original data has a different value, the threshold value is corrected so that the darker original data is quantized into a darker gradation.

A specific example of this processing will be described. Hereinafter, the threshold value correction process for the original image data Xa (a = 1, 2, 3, ...) Will be described. The threshold correction unit 3
First, the value γ (Xa) of the gamma-converted data for the original image data Xa and the value γ (Xa + n) of the gamma-converted data for other image data Xa + n (n is an integer) adjacent to Xa other than Xa are the same. Certain data (hereinafter, this data is referred to as the same gamma-converted value original data) is obtained.

In the gamma conversion table of this embodiment, the original gamma conversion value original data for the original image data "7" is "5, 6, 7, 8, 9". If there is only one original gamma-converted value original data, it is not necessary to correct the threshold value. If there are two or more original data with the same gamma conversion value, it is possible to give a difference in the quantization result among them (so that the original data is quantized into darker gradation). Therefore, it is necessary to modify the threshold.

Therefore, the larger the value of the same gamma-converted value original data is, the smaller the threshold value is corrected. Intermediate value of the same gamma-converted value original data (hereinafter referred to as the same original data intermediate value. For example, “2” in “1, 2, 3, 4”,
For “5, 6, 7, 8, 9”, “7”) is used as a reference threshold without correction, and as the value increases from the same data intermediate value, the corresponding threshold decreases from the reference threshold and decreases from the same data intermediate value. The corresponding threshold value is increased from the reference threshold value.

When an example is described by a general formula, if the same gamma-converted value original data is Xa + n (n is an integer) and the same data intermediate value is XaM, the correction threshold f_th (Xa
+ N) is f_th (Xa + n) = th (Xa) −α × ((a +
n)-(aM)). Here, α is a constant indicating the degree of correction, and is normally set to 1. (A + n)-(aM) is the target pixel data X
It indicates how far a + n is from the same data intermediate value XaM.

The quantization error buffer 4 sequentially stores the quantization error Er (Xa) between the pixel of interest Xa and the value Out (Xa) obtained by quantizing the pixel. In the error weight matrix storage unit 5, an error weight matrix Wa-r in the same error diffusion process as in the conventional case (r represents the degree of the vicinity of the target pixel Xa) is preset and stored.

The weighted average processing unit 7 stores the quantization error Er (Xa) of the neighboring pixels stored in the quantization error buffer 4.
-R) is weighted and averaged by the error weight matrix Wr of the neighboring pixels stored in advance in the error weight matrix storage unit 5, the neighboring pixel error weight average value MEar is calculated, and the neighboring pixel error weight average value MEar is added. Send to Part 9.

In the addition processing unit 9, the neighboring pixel error weighted average value MEar from the weighted averaging unit 7 and the value γ (Xa) of the gamma-corrected data of the target pixel Xa from the gamma correction unit 1 are calculated.
And are added and the value Va is sent to the quantizer 6.

The quantizing unit 6 adds the corrected threshold value f_th (Xa) from the threshold correcting unit 3 and the value Va added by the addition processing unit 9, that is, the weighted average of neighboring pixels of the target pixel from the addition processing unit 7. The value MEar and the value Va obtained by adding the gamma-converted data value γ (Xa) of the target pixel are compared and quantized into any one of four values, and the quantized data Out (X
Output as a).

In this embodiment, when 0 ≦ Va <85: Out (Xa) = 85 if Va> f_th (Xa) Out (Xa) = 0 if Va ≦ f_th (Xa) 85 ≦ Va <170 When: Va> f_th (Xa) Out (Xa) = 170 If Va ≦ f_th (Xa) Out (Xa) = 85 When 170 ≦ Va <= 255: If Va> f_th (Xa) Out (Xa) ) = 255 Va ≦ f_th (Xa), Out (Xa) = 170. This is similar to the comparison process in the conventional error diffusion process.

When the data of all the pixels are quantized by the quantization processing of the quantization unit 6, the above processing is ended.
If the data of all the pixels are not quantized by the quantization processing of the quantization unit 6, the target pixel is switched to the adjacent pixel and the above processing operation is repeated. The threshold correction unit 3
Since the operation other than the operation related to is the same as that of the conventional error diffusion processing, detailed description thereof will be omitted.

In the image processing apparatus according to the embodiment of the present invention, prior to the quantization processing of the data Xa of the target pixel described above,
Similar quantization processing is performed on the data Xa-r of the pixel in the vicinity of the pixel of interest and is stored in the quantization error buffer 4.

That is, the quantized data Out of the neighboring pixels
(Xa-r) is sent to the subtraction processing unit 8 and the gamma correction unit 1
After the gamma correction of the neighboring pixels Xa-r from γ (Xa
-R) and the quantization error Er (Xa-
r) is calculated and stored in the successive quantization error buffer 4.

As described above, if the original data has different values even if the gamma-converted data has the same value, the threshold value in the error diffusion process is quantized into a darker gradation when the original data is darker. By performing the correction, it is possible to realize the error diffusion process without the gradation characteristic being reduced by the gamma correction.

[0057]

As described above, according to the present invention, an error weight matrix storage means in which a matrix for weighting and diffusing a quantization error to neighboring pixels and diffusing it is stored in advance, and a quantization Quantization error buffer for sequentially storing the errors in the case of being processed, weighted averaging means for weighting and averaging the quantization errors of the peripheral pixels by the error weight matrix of the peripheral pixels, and a threshold value for quantizing the pixel of interest are set. The threshold value setting means and the quantizing means for comparing the threshold value with the weighted average value of the quantization errors of the pixels in the vicinity of the pixel of interest and the added value of the values of the pixel of interest to quantize into a multivalued binary or higher value. In the image processing device that performs multi-valued error diffusion processing, the original image data is gamma-converted according to a predetermined gamma characteristic, and the gamma-converted data and the original image data are By modifying the threshold referring to, there is an effect that it is possible to perform the error diffusion processing gradation characteristic without reducing the gamma correction.

[Brief description of drawings]

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

FIG. 2 is a flowchart showing the operation of the image processing apparatus according to the embodiment of the present invention.

FIG. 3 is a flowchart showing a quantization process of neighboring pixel data of the image processing apparatus according to the embodiment of the present invention.

[Explanation of symbols]

1 Gamma correction section 2 Threshold setting section 3 Threshold correction unit 4 Quantization error buffer 5 Error weight matrix storage unit 6 Quantizer 7 Weighted average processing unit 8 Subtraction processing unit 9 Addition processing unit

Claims (6)

(57) [Claims] 1. An error weight matrix storage means in which a matrix for weighting and diffusing a quantization error to neighboring pixels to be diffused is stored in advance, and a quantization for sequentially storing errors when quantized. An error buffer, a weighted average processing means for weighting and averaging the quantization error of the peripheral pixels by an error weight matrix of the peripheral pixels, a threshold setting means for setting a threshold when quantizing the target pixel, the threshold and the target pixel And a quantization means for comparing the weighted average value of the quantization error of the pixels in the vicinity of and the added value of the values of the pixel of interest to quantize into a multi-valued binary or higher value. An image processing apparatus that performs gamma conversion of original image data according to a predetermined gamma characteristic, and data after gamma conversion by the gamma correction means. An image processing apparatus, comprising: a threshold correction unit that corrects a threshold with reference to original image data. 2. The threshold correction means is configured such that, if the data after gamma conversion by the gamma correction means has the same value but the value of the original data is different, the original data is quantized into a darker gradation. The image processing apparatus according to claim 1, wherein the image processing apparatus is configured to correct the threshold value. 3. An error weight matrix storage means in which a matrix for weighting and diffusing a quantization error to neighboring pixels to be diffused is stored in advance, and a quantization for sequentially storing the error when quantized. An error buffer, a weighted average processing means for weighting and averaging the quantization error of the peripheral pixels by an error weight matrix of the peripheral pixels, a threshold setting means for setting a threshold when quantizing the target pixel, the threshold and the target pixel And a quantization means for comparing the weighted average value of the quantization error of the pixels in the vicinity of and the added value of the values of the pixel of interest to quantize into a multi-valued binary or higher value. An image processing method of an image processing apparatus, comprising: a step of gamma-converting original image data according to a predetermined gamma characteristic, the gamma-converted data, and the original image data. And a step of correcting the threshold value with reference to. 4. If the data after gamma conversion has the same value and the value of the original data is different, the step of correcting the threshold is modified so that the darker original data is quantized into a darker gradation. The method according to claim 3, wherein
The described image processing method. 5. An error weight matrix storage means in which a matrix for weighting and diffusing a quantization error to neighboring pixels to be diffused is stored in advance, and a quantization for sequentially storing the error when quantized. An error buffer, a weighted average processing means for weighting and averaging the quantization error of the peripheral pixels by an error weight matrix of the peripheral pixels, a threshold setting means for setting a threshold when quantizing the target pixel, the threshold and the target pixel The image processing apparatus includes a quantizing means for comparing the weighted average value of the quantization errors of the pixels in the vicinity of and the added value of the values of the pixel of interest to quantize into a multivalued binary value or more. A recording medium recording an image processing program for performing a diffusion process, wherein the image processing program causes the image processing apparatus to set a gamma characteristic in which original image data is predetermined. A recording medium on which an image processing program is recorded, characterized by performing gamma conversion in accordance with the above and correcting the threshold value by referring to the data after the gamma conversion and the original image data. 6. When the image processing program causes the image processing apparatus to correct the threshold, if the data after gamma conversion has the same value but the value of the original data is different, the darker the original data is, the darker the original data is. 6. A recording medium recording an image processing program according to claim 5, wherein the threshold value is modified so that the image processing program is quantized.