JP2000050067A - Image processing method and device therefor - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide an image processing method and an image processor capable of obtaining the output images of high quality for the source images of various conditions. SOLUTION: In this image processing method, by executing an error propagation processing (3) to inputted M gradation image data, they are converted to N (M>N) gradation image data. Then, a first average value which is the average value of the pixel under consideration and the adjacent pixels of the converted image data converted from M gradation to N gradation by the error propagation processing is calculated (4) and a second average value which is the average value of input image data corresponding to the pixel under consideration and the adjacent pixels is calculated (5). The first and second average values are compared (6) and the final value of the pixel under consideration is decided corresponding to the compared result.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image processing method and apparatus for processing an input M gradation image and converting it into an N (N> M) gradation image. The present invention relates to an image processing method and apparatus for performing processing.

[0002]

2. Description of the Related Art Generally, in digital copying machines, digital printers, digital facsimile machines, etc., when digital processing is performed on an image including a shaded portion, that is, a multi-tone image, image degradation does not occur and the amount of memory used. Has been an important issue. Therefore, conventionally, it has been considered that an effective method is to reduce the number of gradations by quantizing a multi-gradation image. Among these methods, there is, for example, an error diffusion method as a pseudo halftone expression processing method for outputting a multi-tone image in a pseudo halftone.

In the error diffusion method, a value obtained by multiplying a multi-level (binarization) error of a multi-level (binarization) of a peripheral pixel, which has already been multi-level (binarization), by a weighting coefficient is added to a target pixel, and then multiplied by a fixed threshold value. This is a method of binarizing (binarizing) and distributing a multi-level (binarizing) error generated at this time to peripheral pixels. That is, the output level corresponding to the pixel unit is a predetermined multi-level (binary), but relates to the local area of the output image corresponding to the local area of the input original image (multi-level image data) having continuous gradation. This is a method of eliminating or minimizing the sum of errors to compensate for gradation. For this reason, it is possible to express an image input with a smaller number of gradations than the number of gradations of the input pixel signal.

However, this error diffusion method uses
It has the drawback that a peculiar pattern is generated and granular noise is conspicuous. The error diffusion method has the following disadvantages depending on the size of a diffusion matrix for diffusing an error. That is, when the main scanning direction is on the right side and the sub-scanning direction is on the lower side, and when the diffusion matrix size is large,
When the white area on the right side and the lower side of the edge part is conspicuous, and the diffusion matrix size is small, there is a problem that the pattern of the halftone part is conspicuous. Therefore, in the error diffusion process, there is a problem that the image quality is deteriorated due to the above-described inconvenience. Further, in the error diffusion processing for performing multi-level conversion using a fixed threshold, substantial gray-scale characteristics corresponding to each level of the multi-level signal and a macro gray-scale characteristic with little change with respect to a change in the number of gray levels are small. In this case, however, a decrease in resolution and an increase in the degree of conspicuousness of the texture occur, so that deterioration in image quality cannot be sufficiently suppressed.

Therefore, Japanese Patent Application Laid-Open No. 63-164570 discloses an image processing method in which an edge component of an image is detected and a diffusion matrix size of an error diffusion method is made variable in accordance with the size of the detected edge component. Have been. According to this image processing method, when the edge component is large, the diffusion matrix is made small, and when the edge component is small, the diffusion matrix is made large. The halftone portion in the case of using the matrix size can be reproduced well. This allows
When an image has different edge components depending on the type of image such as a character, a photograph, and a halftone dot, the quality of the output image is improved by changing the diffusion matrix size suitable for the size of the edge component of each image.

[0006]

However, when an image including a shaded portion is reproduced in a half tone by the error diffusion method, it is necessary to set many processing parameters. In addition, in order to obtain a high-quality output image by expressing a pseudo halftone from the original image, it is necessary to change the set values of the above processing parameters in accordance with the characteristics of the image.

Therefore, as described in the above publication, it is difficult to output an image with high quality only by considering the edge components of the image, and it is necessary to set other parameters according to the type of the original image. . For example, when the entire image is processed with the same parameters, in an original image in which image regions having different characteristics such as a photograph portion and a character portion are mixed, a portion where the original image is appropriately reproduced and a portion where the original image is not properly included in the output image. Occurs. Also, in the original image of the halftone dot, the parameters for obtaining the optimum output image are different depending on the density. Further, when the scaling process is performed, the optimum output image is obtained regardless of the type of the output image. Are obtained. Therefore, for example, when a pseudo halftone expression process such as an error diffusion method is performed on an original image to be scaled, depending on the algorithm of the scaling process, jaggies are conspicuous, particularly in a line drawing, and the quality of an output image is high. The problem of lowering the As described above, when reproducing the halftone by the error diffusion method, the parameters for obtaining the optimum output image differ depending on the characteristics of the original image and the conditions of the scaling process. On the other hand, it has been difficult to obtain an appropriate output image even if a single process is performed using the same parameters.

SUMMARY OF THE INVENTION The present invention has been made to solve the above problems, and has as its object to provide an image processing method capable of obtaining a high-quality output image with respect to an original image under various conditions. An image processing device is provided.

[0009]

The present invention has the following arrangement to solve the above-mentioned problems. That is, claim 1 of the present application
The described invention is an image processing method for converting input M gradation image data to N (M> N) gradation image data by performing an error diffusion process, wherein the M diffusion process is performed by the error diffusion process. Calculating a first average value, which is an average value of the pixel of interest and its neighboring pixels of the converted image data converted from the tone to the N gradation, and averaging the input image data corresponding to the pixel of interest and the neighboring pixels. A second average value, which is a value, is calculated, the first and second average values are compared, and a final value of the target pixel is determined according to the comparison result.

According to the invention described in claim 2 of the present application, the neighboring pixels are
The selection is made such that the number of pixels located in the vertical direction or the horizontal direction is larger than the number of pixels located in the oblique direction from the target pixel. In the invention according to claim 3 of the present application, when calculating the average value of the neighboring pixels, the value of the target pixel uses a pixel value converted at a specific gradation and a value convertible at other gradations. The average value of each is calculated. The invention described in claim 4 of the present application determines the pixel value of interest used as the average value having the smallest difference as an output value as a result of comparing the average values.

The invention according to claim 5 of the present application provides a
An image processing apparatus for converting gradation image data into N (N <M) gradation image data and performing an error diffusion process on the gradation-converted image data. Means, image storage means for storing inputted input image data, error diffusion processing means for reading out a pixel value from the image storage means and performing error diffusion processing, a target pixel of the converted image data subjected to error diffusion processing, and First average value calculating means for calculating an average value of neighboring pixels, and second average value for calculating the average value of the pixel of interest and the neighboring pixels read from the image storage means and calculated by the first average value calculating means Calculating means; average value comparing means for comparing at least two or more average values calculated by the first average value calculating means with the average value calculated by the second average value calculating means; Comparison A pixel value changing unit for appropriately changing a pixel value of interest as necessary based on a comparison result by the stage; and an image output unit for outputting a finally determined converted image. , N (N <M), the average value of the neighboring pixels including the target pixel of the converted image data converted by the error processing is compared with the average value of the input image data corresponding thereto. As a result, the target pixel value used for the average value having the smallest difference is determined as the final output value, and the value of the target pixel is changed.

[0012]

DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will be described below with reference to FIGS. FIG. 1 is a block diagram showing an embodiment of an image processing apparatus according to the present invention. FIG. 2 is a flowchart showing an image processing method executed by the image processing apparatus of FIG. In each figure,
The image data of the original image input by the image input means 1 is subjected to correction processing such as shading correction and input gradation correction (not shown) and stored in the image storage means (step 1). Performs an error diffusion process.

The error diffusion process is a pseudo halftone processing method for reproducing an image having a shaded portion (multi-tone image) in a pseudo halftone while maintaining image characteristics. That is, the original image is converted into an image having a smaller number of tones than the number of tones and output. Therefore, after comparing the pixel density with a certain threshold value, the pixel density is quantized to a certain value according to the result, and the difference value generated at this time, that is, the quantization error is distributed (diffused) to the neighboring pixels, and the pseudo halftone is used. That is to reproduce.

This will be described in more detail with reference to the drawings as follows. As shown in FIG. 3, when a raster scan in which the main scanning direction of the document is in the x direction and the sub scanning direction is in the y direction is performed, the density value of the pixel of interest Z is represented by g (x,
y), this pixel is quantized, and the error generated at that time is represented by g (x + 1, y) and g (x−1, y), which are neighboring pixels.
+1), g (x, y + 1), g (x + 1, y + 1)
(Step 2).

For example, assuming that the quantization value of the target pixel Z is P, the error E is expressed by the following equation (1). E = g (x, y) -P (1) From equation (1), the density of the four pixels shown in FIG. And distributed as follows: That is, g (x + 1, y) is g (x +
1, (y) + E × 7/16, g (x−1, y + 1) is
g (x−1, y + 1) + E × 3/16, g (x, y +
1) becomes g (x, y + 1) + E × 5/16 and g (x +
1, y + 1) is g (x + 1, y + 1) + E × 1/16
become. In this way, the target pixel Z is advanced in the scanning direction,
The same processing is performed for pixels after g (x + 1, y). Note that the above diffusion coefficient is a coefficient for weighting the distribution of the error to the neighboring pixels of the target pixel Z,
Here, a diffusion coefficient by a Floyd & Steinberg filter is used.

After the error diffusion process is performed as described above, the first average value calculating means 4 calculates the average value of the quantized neighboring pixels including the target pixel. FIG. 4 shows an example of the selection range of the neighboring pixels. In the figure, Z is the pixel of interest, and its surrounding 12 pixels filled with oblique lines are neighboring pixels. Here, the pixels are located more vertically or horizontally than the number of pixels located diagonally from the pixel of interest. The number of pixels is selected so as to be larger. That is, one pixel goes obliquely from the target pixel Z, and two pixels are located in the vertical and horizontal directions. Then, the converted density value of the target pixel Z is defined as g ′ (x, y), and the above neighboring pixels are defined as g ′ (x, y−2) and g ′ (x−y), respectively.
1, y-1), g '(x, y-1), g' (x + 1, y
-1), g '(x-2, y), g' (x-1, y),
g ′ (x + 1, y), g ′ (x + 2, y), g ′ (x−
1, y + 1), g '(x, y + 1), g' (x + 1, y
+1), g '(x + 1, y + 1), g' (x, y + 2)
In this case, the average value P ′ of these neighboring pixels is expressed by the following equation (2).

P ′ = (g ′ (x, y−2) + g ′ (x−1, y−1) + g ′ (x, y−1) + g ′ (x + 1, y−1) + g ′ (x−2, y) + g ′ (x−1, y) + g ′ (x + 1, y) + g ′ (x + 2, y) + g ′ (x−1, y + 1) + g '(x, y + 1) + g' (x + 1, y + 1) + g '(x, y + 2) + g' (x, y) / 13 ...・ ・ ・ ・ ・ ・ ・ ・ （２）

The average value of the converted image data is obtained from the equation (2). Similarly, a value that can be taken when the pixel of interest Z has been subjected to the pseudo halftone expression processing, that is, in the case of error diffusion processing for converting into N-values, values from 0 to N are g ′.
Substituting into (x, y), the formula (2) is calculated, and the average value at each value (0 to N) is obtained. Further, while the processing by the error diffusion processing means 3 and the first average value calculation means 4 is being performed, the average value of the input image is also calculated by the second average value calculation means 5. In that case, the neighboring pixels are the first
Using the pixels in the same selection range as the selection range of the original image pixels used in the average value calculation means 4, the same calculation as the expression (2) is performed.

As described above, the first average value calculating means 4
When a plurality of average value outputs and the corresponding average value output of the second average value calculation means 5 are obtained, each average value is input to the average value comparison means 6. The average value comparison means 6 compares each of the plurality of average value outputs of the first average value calculation means 4 with the corresponding average value output of the second average value calculation means 5 (step 3). Then, among the N average value outputs calculated by the first average value calculation means 4 using the values of 0 to N, the value closest to the average value output of the second average value calculation means 5, that is, The value of the smallest difference between the average value output value of the first average value calculation means 4 and the average value output value of the second average value calculation means is selected as the final value of the target pixel Z. At this time, if the value of the pixel of interest converted by the error diffusion processing means 8 is selected, the value is transferred to the image output means 8 (step 5), and if a different value is selected, the value is Is changed by the pixel value changing means 8 (step 4), and is passed to the image output means 8 (step 5), and the image output means 8 outputs a final output image. In the above embodiment, since the number of pixels located in the vertical and horizontal directions is set to be larger than the number of pixels located in the oblique direction from the pixel of interest, the neighboring pixels are often seen in line drawings. The vertical and horizontal edge portions can be easily recognized, and the white spots on the right and lower sides of the edge portion in the line drawing can be improved.

The above is the description of the image processing apparatus according to one embodiment of the present invention and the image processing method executed by the image processing apparatus. Next, the image processing method executed by the image processing apparatus will be described in more detail. This will be described using a specific operation example. Here, a case will be described in which a 256-gradation image is quaternized, that is, converted into a pseudo-gradation image in which N = 4. However, the value of N is an integer of M>N> 1. The same processing is possible with any value. The four values are (0, 85, 170, 255). FIG. 5 shows a part of input image data to be subjected to pseudo halftone processing in the present embodiment. An error diffusion process is performed on this input image. In the error diffusion processing here, the threshold is set to 12
It is assumed that the error is distributed using a fixed threshold value of 8 and the diffusion coefficient obtained by the Floyd & Steinberg filter described above. When a converted image is obtained by this error diffusion processing, FIG.
It becomes as shown in.

Here, as an example of obtaining the average value, an example in which the central portion in FIG. 6 is set as a target pixel will be described. The value of the pixel of interest converted by the error diffusion process is 170. The average value of the neighboring pixels when the target pixel is 170 is 92. Similarly, the average value of the neighboring pixels when the pixel of interest is 0 is 78, the average value of the neighboring pixels when the pixel of interest is 85 is 85, and the average value of the neighboring pixels when the pixel of interest is 255 is 98. The average value of the corresponding neighboring pixels of the input image is 97. From this result, the value closest to 97 is 98, and the value of the pixel of interest when calculating the average value is 255. Therefore, the pixel value of interest of the output image data is 255. By sequentially performing the above-described operations on all the pixels, the converted image data in FIG. 6 is corrected into output image data as shown in FIG. 7, and ordinary error diffusion for simply distributing the quantization error to neighboring pixels is performed. A higher quality output image closer to the original image can be obtained as compared with the method.

Note that, in the above embodiment, as shown in FIG.
The case where a total of 12 pixels are selected in the horizontal and diagonal directions has been described as an example, but it is also possible to set the number of neighboring pixels to 12 or more or less, and the present invention is particularly limited to the above embodiment. It is not something to be done.

[0023]

As described in detail above, the present invention has the following effects. That is, according to the first aspect of the invention, the input M gradation image is processed by the processing city, N (N <M).
An image processing method for converting into a gradation image, comprising comparing an average value of neighboring pixels including a pixel of interest of the changed image data converted by the error diffusion process with an average value of the input image data corresponding thereto. By determining the final value of the pixel of interest in accordance with the result, a converted image with a smaller error from the input image is obtained, and a single process is performed on the original image under various conditions using the same parameters. It is also possible to obtain an appropriate output image.

According to the second aspect of the present invention, by selecting the neighboring pixels so that the number of pixels located in the vertical and horizontal directions is larger than the number of pixels located in the oblique direction from the pixel of interest, the number of pixels in the line drawing is increased. The visible vertical and horizontal edge portions can be easily recognized, and the white spots on the right and lower sides of the edge portion in the line drawing can be improved. According to the third aspect of the present invention, when the average value of the neighboring pixels is obtained, the value of the target pixel is obtained by using the converted value and the other convertible values to obtain the average value. The part where the error was not successfully diffused becomes clear, and it becomes possible to obtain converted image data with a small error from the input image data. According to the fourth aspect of the present invention, as a result of the comparison of the average values, the pixel value of interest used for the average value having the smallest difference is determined as the final output value. Image data can be obtained.

According to the fifth aspect of the present invention, when the input M gradation image is processed and converted into an N (N <M) gradation image, the pixel of interest of the converted image data converted by the error diffusion processing is determined. The average value of the neighboring pixels including the target pixel is compared with the average value of the corresponding input image data. As a result, the target pixel value used for the average value having the smallest difference is determined as the final output value. By determining the value as the final output value and changing the value of the pixel of interest, white spots on the right and lower sides of the edge portion in the line drawing are improved,
Furthermore, converted image data with little error from the input image data can be obtained, and an appropriate output image can be obtained even if a single process is performed on the original image under various conditions using the same parameters.

[Brief description of the drawings]

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

FIG. 2 is a flowchart showing a flow of image processing by the one shown in FIG.

FIG. 3 is an explanatory diagram of an error diffusion process shown in FIG.

FIG. 4 is an explanatory diagram of how to calculate an average value using the one shown in FIG.

FIG. 5 is an example of input image data used for the one shown in FIG. 2;

FIG. 6 is an example of converted image data used for the one shown in FIG. 2;

FIG. 7 is an example of output image data used for the one shown in FIG. 2;

[Explanation of symbols]

 Reference Signs List 1 image input means 3 error diffusion processing means 4 first average value calculation means 5 second average value calculation means 6 average value comparison means 7 pixel changing means 8 image output means Z target pixel

Claims (5)

[Claims] 1. An image processing method for converting input M gradation image data into N (M> N) gradation image data by performing an error diffusion process, wherein the M-th gradation image data is subjected to an error diffusion process. Calculating a first average value, which is an average value of the pixel of interest and its neighboring pixels of the converted image data converted from the tone to the N gradation, and averaging the input image data corresponding to the pixel of interest and the neighboring pixels An image processing method comprising: calculating a second average value as a value; comparing the first and second average values; and determining a final value of the target pixel according to the comparison result. 2. The method according to claim 1, wherein the neighboring pixels are selected such that the number of pixels located in the vertical or horizontal direction is larger than the number of pixels located in the oblique direction from the target pixel. Image processing method. 3. When calculating an average value of neighboring pixels, a value of a target pixel is calculated using a pixel value converted at a specific gradation and a value convertible at other gradations. 2. The image processing method according to claim 1, wherein a value is obtained. 4. The image processing method according to claim 1, wherein as a result of comparing the average values, a pixel value of interest used as the average value having the smallest difference is determined as an output value. 5. The method according to claim 5, wherein the input M gradation image data is N (N <N
M) An image processing apparatus that converts the image data into gradation image data and performs an error diffusion process on the gradation-converted image data, wherein: an image input unit for inputting original image data; Image diffusion means for reading a pixel value from the image storage means and performing an error diffusion process; and calculating an average value of a target pixel of the converted image data subjected to the error diffusion process and a neighboring pixel thereof. An average value calculating means, an average value calculating means for calculating an average value of a pixel of interest and a neighboring pixel which is read from the image storage means and calculated by the first average value calculating means, and a first average value. Average value comparing means for comparing at least two or more average values calculated by the calculating means with the average value calculated by the second average value calculating means, and a comparison result by the average value comparing means Pixel value changing means for appropriately changing the pixel value of interest as needed, and image output means for outputting the finally determined converted image. The input M gradation image is processed, and N (N <N M) When converting to a gradation image, the average value of neighboring pixels including the pixel of interest of the converted image data converted by the error processing is compared with the average value of the corresponding input image data. An image processing apparatus characterized in that a pixel value of interest used for an average value having a small difference is determined as a final output value, and the value of the pixel of interest is changed.