JP2002024817A - Image processing method and image processing device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide an image processing method and an image processing device, capable of suppressing generation of color shift or a pseudo-gradation in a color image. SOLUTION: Color conversion processing is executed relative to image data, in order to convert color image data having plural gradations into color image data having the smaller number of gradations and a smaller data quantity than the color image data; the histogram of a gradation level is obtained in a converted color space; a new gradation level is determined corresponding to the distribution state of gradation values on the histogram; and the gradation values of the new gradation level are selected relative to every pixel in local regions, based on the result obtained by dividing the image data into the local regions and detecting the color distribution state and existence of edge in every local position. Hereby, image processing is enabled by executing gradation number conversion having no color shift, no pseudo-gradation and having little deterioration in image quality.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

[0001] 1. Field of the Invention [0002] The present invention relates to an image processing method and an image processing apparatus capable of realizing tone number conversion while suppressing occurrence of color shift and pseudo tone.

[0002]

2. Description of the Related Art Conventionally, a dither method or an error diffusion method for performing conversion to two gradations or a multi-value conversion for three or more gradations has been conventionally used as a method for converting the number of gradations in image processing in a direction to reduce the number of gradations. There are a multi-value dither method and a multi-value error diffusion method.

In these methods, a matrix is defined in advance, and fixed matrix calculation and error diffusion are performed irrespective of the gradation distribution of the image to be processed. For this reason, tone values that do not actually exist in the original image are frequently used, and a large number of errors are generated, so that a pseudo tone or noise is generated. Or it will occur. There is also a method of switching between a plurality of matrices, but this method also has the same problem because the matrix is a fixed value.

[0004] As one method for solving these problems, Japanese Patent Publication No. 5-62863 discloses a technique relating to a halftone reproduction method. In the present invention, the number of gradations that can be expressed by the dither matrix is determined from the histogram of the entire image, and the average number of pixels is calculated from the total number of pixels and the number of gradations.
The histogram for each pixel density is sequentially added, the sum up to the closest to the average number of pixels is obtained, the histogram is smoothed, and the image density value serving as the boundary is set as the threshold of the dither matrix. It goes. As a result, an optimal halftone image can be obtained for each image.

Japanese Patent Application Laid-Open No. 9-154044 discloses a technique relating to an image display method and an apparatus therefor, and performs different gradation conversion for each pixel based on the luminance frequency distribution of the pixel near an edge. To prevent a decrease in contrast.

[0006]

As described above, in the conventional method, grayscale values that do not actually exist in the original image are frequently used, and a great deal of error occurs. And noise, and color misregistration and false colors occur in a color image.

Further, in the halftone reproducing method disclosed in Japanese Patent Publication No. 5-62863, reproduction can be performed using a binary recording device. However, since a color image is not mentioned, an intermediate color image is not reproduced. It is difficult to apply this method as it is when reproducing the key.

The image display method and apparatus disclosed in Japanese Patent Application Laid-Open No. 9-154044 is an invention characterized by basically improving the contrast, and does not refer to tint. There is insufficient support for gradation conversion from a natural image to a multi-valued color image.

In order to solve such a problem, the present invention prepares a histogram of the entire image from luminance information using a uniform color space, determines a new gradation level from the distribution state, and simultaneously determines a local level of the image. An image processing method and an image processing method that performs color information detection and edge detection for each region, and determines the tone value of a target pixel according to the state of neighboring pixels, thereby suppressing the occurrence of color shift and pseudo tone in a color image. It is an object to provide a processing device.

[0010]

The present invention has the following arrangement as means for solving the above-mentioned problems.

(1) An image processing method for converting color image data having a plurality of gradations into color image data having a smaller number of gradations and a smaller data amount than the color image data. A color conversion process is performed to obtain a histogram of gradation levels in the converted color space, and a new gradation level is determined according to the distribution state of the gradation values on the histogram. Is divided into local regions, the color distribution state and the presence / absence of edges are detected for each local region, and the new gray level value is selected for each pixel of the local region based on the detection result. And

In this configuration, in order to convert color image data having a plurality of gradations into color image data having a smaller number of gradations and a smaller data amount than the color image data, color conversion is performed on the image data. Processing is performed to obtain a histogram of the gradation levels in the converted color space, and a new gradation level is determined according to the distribution state of the gradation values on the histogram. Further, the image data subjected to the color conversion processing is divided into local regions, and based on the result of detecting the color distribution state and the presence / absence of an edge for each local region, a new gradation level for each pixel of the local region is obtained. Select a key value. Therefore, by executing the above-described image processing method, it is possible to obtain an image which is free from color shift, pseudo gradation and the like, and has little image quality deterioration.

(2) A uniform color space is used as a color space after color conversion of image data.

In this configuration, the uniform color space is the color space after the color conversion processing of the image data. Therefore,
The positional relationship of each peripheral pixel can be easily obtained from the center point of the uniform color space.

(3) A histogram is created from the luminance information of the entire image, and the current gradation level is divided into a plurality according to a predetermined new number of gradations and the frequency of each gradation value. It is characterized in that each representative value is determined from the group and the new gray level is determined.

In this configuration, a histogram is created from the luminance information of the entire image, and each representative value is obtained from a plurality of divided gradation levels according to a predetermined new number of gradations and the frequency of each gradation value. To determine a new gray level. Therefore, since a new gradation value is determined according to the frequency of each gradation value, it is possible to reduce the number of gradations according to the gradation characteristics of the image.

(4) The number of pixels obtained by dividing the total number of pixels of the image data by the number of gradations of the new gradation level and adding a certain number is defined as one gradation level range, While counting and dividing the number of pixels from the maximum value or the minimum value of the tonal value, and starting the next count from the position where the predetermined value equal to or less than the certain number is counted backward from the last position of the previous count, , The current gradation level is divided into a plurality of pixels, and the current gradation level is divided into a plurality of gradation values.

In this configuration, the range of one gradation level is the number of pixels of a value obtained by adding a fixed number to the value obtained by dividing the total number of pixels of the image data by the number of gradations of the new gradation level. The number of pixels is counted from the maximum value or the minimum value of the gradation value on the histogram, and the next count is started from the position where the predetermined value equal to or less than a predetermined number is counted in the reverse direction from the last count last position, and sequentially. Counting is performed, and the current gradation level is divided into a plurality of gradation levels by dividing the current gradation level into a plurality of gradation levels. Therefore, nearer grayscale values are more finely divided, while smaller grayscale values are more widely divided, and in the divided portions, adjacent grayscales have overlapping portions. Therefore, the gradation value can be selected according to the presence or absence of the edge, and the frequency of appearance of the gradation value in each image and the number of gradations can be converted according to the characteristics of the image. It becomes possible.

(5) The representative value is a weighted average value according to a distribution in each gradation level group.

In this configuration, a weighted average value according to the distribution in each gradation level group is set as a representative value in each of the divided gradation level groups. Therefore, the most average value in the divided gradation level group can be determined as the representative value.

(6) The representative value is a center value in each gradation level group.

In this configuration, the center value in each gradation level group is set as a representative value in each of the divided gradation level groups. Therefore, it can be determined as a representative value by a simple calculation.

(7) The representative value is a maximum frequency value in each gradation level group. In this configuration, the maximum frequency value in each gradation level group is set as a representative value in each of the divided gradation level groups. Therefore, it is possible to easily determine the value by selecting only the most frequent value as the representative value, and furthermore, it is possible to make the value appropriate as the representative value of the divided gradation level group.

(8) The representative value is added to the converted new gradation image data.

In this configuration, each representative value in the divided gradation level group is added to the converted new gradation image data. Therefore, by adding the relationship between the original luminance and the luminance of the converted gradation value, it is possible to re-convert to a color space suitable for each output device without changing the contrast.

(9) Using the chromaticity coordinate values of the converted color space, it is determined that the pixel is a chromatic color or an achromatic color based on the distance from the origin on the coordinates of each pixel. The color distribution state is detected by determining whether the colors are similar or not and grouping the colors.

In this configuration, in order to detect a color distribution state, each pixel is determined to be chromatic or achromatic according to the distance from the origin of each pixel on the chromaticity coordinates of the converted color space, and each pixel on the coordinates is determined. The pixels are grouped by judging whether or not the pixels have the same color based on the mutual positional relationship between the pixels. Therefore, chromatic colors, achromatic colors, and similar colors can be easily determined from the distance from the origin on the chromaticity coordinates and the mutual positional relationship.

(10) In each pixel, when the Euclidean distance from the origin on the chromaticity coordinates is equal to or smaller than a first specified value, it is determined that the pixel is achromatic, and the Euclidean distance from the origin is smaller than the first specified value. If they are separated from each other, the color is determined to be chromatic.

In this configuration, when the Euclidean distance from the origin on the chromaticity coordinates of the converted color space of each pixel is equal to or less than the first specified value, it is determined that the pixel is an achromatic color, and the position of each pixel from the origin is determined. If the Euclidean distance is longer than the first specified value, it is determined that the color is chromatic. Therefore, you can easily distinguish between achromatic and chromatic colors,
In addition, it is possible to group colors so that the difference in color is not so noticeable.

(11) When the Euclidean distance from the origin on the chromaticity coordinates between the pixels is equal to or less than a second specified value, it is determined that the pixels are of the same color.

In this configuration, if the Euclidean distance from the origin on the chromaticity coordinates between the pixels is equal to or smaller than the second specified value, the pixels are determined to be of the same color. Therefore, similar colors and non-similar colors can be distinguished by a simple method.

(12) The method is characterized in that the difference between the luminance of each pixel in the local area and the luminance of the neighboring pixels is obtained, and the presence or absence of the edge is detected by examining the distribution of the difference.

In this configuration, the presence or absence of an edge is detected by examining the distribution of the luminance difference between each pixel in the local area and the neighboring pixels. Therefore, there is a luminance difference between each pixel in the local region and a neighboring pixel, and if the luminance difference is continuous, an edge exists there. Becomes possible.

(13) When the peripheral pixels including the target pixel are achromatic and the gradation value of the target pixel belongs to only one gradation level group, the gradation value of the gradation level group is selected. I do. In this configuration, when the peripheral pixels including the target pixel are achromatic and the gradation value of the target pixel belongs to only one gradation level group, the gradation value of this gradation level group is selected as the gradation value. I do. Therefore, in the case of an achromatic color, there is no relation to color shift, and it is possible to simply select a gradation value.

(14) When the peripheral pixels including the target pixel are achromatic and the gradation values of the target pixel belong to two gradation level groups,
When there is an edge, the tone value with the larger tone difference is selected from the two new tone levels, and when there is no edge, the tone value closer to the neighboring new tone levels from the two new tone levels is selected. The method is characterized in that a key value is selected.

In this configuration, when the peripheral pixels including the target pixel are achromatic and the gradation values of the target pixel belong to two gradation level groups, if there is an edge, the gradation difference from the two new gradation levels is obtained. The larger gradation value is selected, and if there is no edge, the gradation value closer to the surrounding new gradation levels is selected from the two new gradation levels. Therefore, in the case of an achromatic color, there is no relationship between color shifts. Therefore, when the color belongs to two gradation level groups, an optimum gradation value can be selected only by the presence or absence of an edge.

(15) If the peripheral pixels including the target pixel are chromatic and the gray level of the target pixel belongs to only one gray level group, the gray level of the gray level group is selected as a new gray level. When the peripheral pixel including the target pixel is a chromatic color and the grayscale value of the target pixel belongs to two grayscale levels, the grayscale value is selected in accordance with the presence / absence of an edge and the presence / absence of a similar color. I do.

In this configuration, when the peripheral pixels including the target pixel are chromatic and the gradation value of the target pixel belongs to only one gradation level group, the gradation value of this gradation level group is changed to a new gradation level. Select as a value. When the peripheral pixel including the target pixel is a chromatic color and the target pixel belongs to two gradation level groups, the gradation value is selected according to the presence or absence of an edge and the presence or absence of a similar color. Therefore, in the case of a chromatic color, since a color shift is also involved, it is possible to select an optimum tone value by checking not only the presence or absence of an edge but also whether or not the color is the same.

(16) When the peripheral pixels including the pixel of interest are chromatic and the gradation values of the pixel of interest belong to two gradation level groups,
When there is an edge, a tone value with a larger tone difference is selected from the two new tone levels, and when there is no edge, it is checked whether or not the pixel of interest is the same color as its surrounding pixels,
If the colors are not the same, select the gradation value with the larger gradation difference from the two new gradation levels, and if the colors are the same, change the two new gradation levels to the surrounding new gradation levels. It is characterized in that a nearer gradation value is selected.

In this configuration, if the peripheral pixel including the target pixel is a chromatic color and the target pixel belongs to two gray level groups, if there is an edge, the gray level difference from the two new gray levels is obtained. Is selected, and if there is no edge, it is checked whether or not the pixel of interest is of the same color as its surrounding pixels. If not, the difference between the two new gray levels is large. The tone value that is closer is selected, and if the color is the same, the tone value closer to the surrounding new tone level is selected from the two new tone levels. Therefore, in the case of a chromatic color, a pseudo tone due to a color shift or a color is also involved. Therefore, it is necessary to examine the presence or absence of an edge and a similar color to select a tone value at which a pseudo tone is hardly generated. It becomes possible.

(17) An image processing apparatus which converts color image data of a plurality of gradations input from an input device into color image data having a smaller number of gradations and a smaller data amount than the color image data, and outputting the color image data to an output device. A color conversion processing section for performing color conversion processing on the image data input from the input device; and obtaining a histogram of gradation levels in the color space converted by the color conversion processing section. A tone level determining unit that determines a new tone level according to a tone value distribution state, a color information detecting unit that divides image data into local areas, and detects a color distribution state for each local area, An edge detection unit that divides the image into local regions and detects the presence or absence of an edge for each local region; and a new gradation determined by the gradation level determination unit based on detection results of the color information detection unit and the edge detection unit. And a gradation value selecting section for selecting a gradation value from the level and outputting the selected gradation value to the output device.

In this configuration, the image processing apparatus includes a color conversion processing section, a gradation level determination section, a color information detection section,
The image processing apparatus includes an edge detection unit and a gradation value selection unit, performs color conversion processing on image data input from the input device in the color conversion processing unit, and converts the image data in the color conversion processing unit in the gradation level determination unit. A histogram of the gradation level is obtained in the color space obtained, a new gradation level is determined in accordance with the distribution state of the gradation values on the histogram, and the image data is divided into local regions by the color information detecting unit, and each local region is To detect the color distribution,
The image data is divided into local regions by the edge detection unit, and the presence or absence of an edge is detected for each local region. The gradation value selection unit determines the gradation level based on the detection results of the color information detection unit and the edge detection unit. A tone value is selected from the new tone levels determined by the unit and output to the output device. Therefore, the gradation level is determined from the distribution state of the histogram of the entire image, and the color distribution state of the local region and the presence / absence of an edge are detected. By determining the above, it is possible to realize an image processing apparatus that performs tone number conversion without color shift and pseudo tone and with little image quality deterioration.

[0043]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS One embodiment of a gradation number conversion method as an image processing method will be described with reference to FIGS. FIG. 2 is a flowchart illustrating an embodiment of the gradation number conversion method. FIG. 3 is a diagram illustrating a method of obtaining a histogram of all pixels. FIG. 4 is a diagram illustrating a method for obtaining a new gradation level. FIG. 5 is a diagram illustrating a method of extracting an achromatic region. FIG. 6 is a diagram illustrating a method of extracting similar colors in a chromatic color area. FIG. 7 is a diagram illustrating a method of extracting an edge. FIG. 8 is a diagram illustrating a result of conversion to a new luminance level. In FIG. 2, RGB color image data of 8 bits for each color is used, and L ^* a ^* is used as a color space to be converted ^.
The case where a b ^* color space is used is shown.

First, an RGB color image is subjected to a color conversion process to convert an RGB color space into an L ^* a ^* b ^* color space (101). The L ^* a ^* b ^* color space is one of the uniform color spaces, and is a color space intended to allow color differences perceived to be equal in size to correspond to equal distances in the space. This makes it possible to make an equal comparison of how much the colors are different, regardless of the colors.

Thereafter, as shown in FIG. 3, in order to calculate a histogram of the entire image, each L ^* value as luminance information is counted for all image data (10).
2). In this case, the pixel data from the upper left pixel data to the lower right pixel data of the image are counted in order. L ^* is from 0 to 10
Since it is expressed by the value of each step up to 0, for example, 1
01 array variable areas are secured, and L ^* 0 to 100
Substitute the counted value into the array variable corresponding to.

Subsequently, as shown in FIG. 4, the range from the value obtained by dividing the total number of pixels by a new predetermined number of gradations, which is a predetermined value, to a value obtained by adding a certain number to the value is one gradation level. And the pixel is counted and divided from the maximum value or the minimum value on the histogram. In addition, counting is performed sequentially from the position where the previous count was completed to the position counted in the opposite direction by a value equal to or less than a certain number added at the time of the previous count as the next count start position (10).
3). As a result, the vicinity of a frequently used tone value is finely divided, while the vicinity of a less frequently used tone value is broadly divided. Further, in the divided portion, an overlapping portion is formed between adjacent gradations, so that a gradation value can be selected according to the presence or absence of an edge, and the appearance frequency of the gradation value and the image The conversion of the number of gradations according to the feature becomes possible, and the deterioration of the image quality can be suppressed.

In dividing the gradation level group, the number of pixels is counted, and when the specified value is reached in the middle of a certain gradation value, the current gradation is counted from the counted gradation value. The values up to the value are divided as one gradation level group. Further, when the specified value is the same as the number of pixels at a certain gradation value, if the number of pixels at the subsequent gradation values is 0, one gradation level group including the gradation value without the count is included. And divide until the end.

Next, a representative value is calculated for each of the divided gradation level groups (104). As the representative value to be calculated,
For example, in FIG. 2, a weighted average value in the gradation level group is used, but a center value of the gradation level group or a maximum frequency value may be used. The representative value obtained in this way is substituted into an array variable or the like as a new gradation level.

In parallel with the calculation of the new gradation level,
A local image is extracted from the image converted from the RGB color space to the L ^* a ^* b ^* color space (111). here,
As an example, a local region of 8 × 8 pixels is extracted.

Next, two processes are performed on the extracted image data of the local region. That is, one is detection of color information and the other is detection of edge information.

First, in the detection of color information, as shown in FIG. 5, the values of a ^* and b ^* of each pixel in the local area are called, and the Euclidean distance D ₁ from the origin is obtained by the following equation (113). . D ₁ (n) = {(a ^{* 2} + b ^{* 2} ) ^} (1) Here, n is the number of each pixel in the local area. If the local area is 8 × 8 pixels, the number of pixels is 64, so
Assuming that the upper left is 0, numbering is done in the right direction,
After the top right is set to 7 and the left end of the next line is numbered sequentially to 8, the bottom right becomes 63.

Then, for each pixel, it is determined whether or not this value exceeds a first specified value, and it is determined whether the color is chromatic or achromatic. For example, when the Euclidean distance is 30 or less, it is determined that the color is determined to be achromatic, and when it is greater than 30, it is determined to be chromatic.

Next, as shown in FIG. 6, the Euclidean distance between adjacent pixels on the a ^* -b ^* plane is determined (114). First, in order for each pixel in the local region, obtained by the following equation Euclidean distance D ₂ between the upper and lower left and right adjacent pixels.

D ₂ (n ₁ , n ₂ ) = {((a ₁ ^* −a ₂ ^* ) ² + (b ₁ ^* −b ₂ ^* ) ² ) ^} (2) In the equation (2), (a) ₁ ^* , b ₁ ^* ) represent the chromaticity coordinate values of the pixel of interest. (A ₂ ^* , b ₂ ^* ) represents the chromaticity coordinate values of the other pixels. Further, n ₁ represents the number of the target pixel, and n ₂ represents the numbers of the other pixels.

When the local area is 8 × 8 pixels, D ₂ (0,
1), D ₂ (0, 2), ‥‥‥‥, D ₂ (0, 8), D
₂ (1, 9), ‥‥‥, D ₂ (8, 9), D ₂ (9, 1
0), ‥‥‥‥, D ₂ (61, 62), D ₂ (62, 6)
3), and a total of 112 Euclidean distances are obtained.

Then, it is determined whether or not this value exceeds the second specified value for each pixel, and it is determined whether or not the pixel is of the same color (114). For example, when the Euclidean distance is 10 or less, it is determined that the color is the same color, and when the distance is more than 10, it is determined that the color is not the same color. In this way, color information is detected.

In the detection of another edge information, as shown in FIG. 7, first, the value of L ^* of each pixel in the local area is called, and the luminance difference between each pixel and each neighboring pixel is calculated by the following equation. (121).

DL (1) = dL (9) = P (x, y) −P (x−1, y−1) dL (2) = P (x, y) −P (x, y−1) dL (3) = P (x, y) -P (x + 1, y-1) dL (4) = P (x, y) -P (x + 1, y) (3) dL (5) = P ( x, y) -P (x + 1, y + 1) dL (6) = P (x, y) -P (x, y + 1) dL (7) = P (x, y) -P (x-1, y + 1) dL (8) = dL (0) = P (x, y) -P (x-1, y) And, among the values of dL (1) to dL (8), those having a certain value or more Is extracted and compared with the results before and after (121). For example, dL
If (3) exceeds a certain value, dL (2) and dL
(4) is compared with dL (3). If the difference between dL (2) and dL (4) is within a certain range, an edge exists in that direction. In this way, the presence or absence of an edge and the direction of the edge are detected in advance for each pixel by determining the difference in luminance between neighboring pixels and examining the distribution of the difference.

Then, as shown in FIG. 8, a new gradation value of each pixel in the local area is substituted into the array variable or the like from two detection results of color information detection and edge information detection. Select from levels (131). As a result, the representative value of each divided gradation level group is added to the new gradation image data. FIG. 8 shows an example in which a luminance value of 100 levels is converted into 16 levels.

Here, as a method of selecting a new gradation value, the peripheral pixels including the target pixel are achromatic, and the gradation value of the target pixel does not belong to the two gradation level groups, but one gradation level. If it belongs to only the group, the gradation value of the gradation level group is selected. Further, when the gradation value of the target pixel belongs to two gradation level groups, depending on the presence / absence of an edge, when there is an edge, the gradation value is increased from two new gradation levels so that a gradation difference appears at an edge portion. Select the tone value with the larger difference. On the other hand, if there is no edge, a tone value closer to the new tone levels in the vicinity is selected from the two new tone levels.

When the peripheral pixel including the target pixel is a chromatic color, and the gradation value of the target pixel does not belong to two gradation level groups but belongs to one gradation level group, the gradation value of the gradation level group Select Further, when the gradation value of the pixel of interest belongs to two gradation level groups, two new gradation levels are set according to the presence / absence of an edge so that when there is an edge, a gradation difference appears at an edge portion. Select the gradation value with the larger gradation difference. On the other hand, when there is no edge, it is checked whether the target pixel has the same color as the surrounding pixels. If the colors are not the same, the tone value with the larger tone difference is selected from the two new tone levels so that the tone difference is greater.
If the colors are the same, a tone value closer to the new tone levels in the vicinity is selected from the two new tone levels.

In this way, the same processing is repeated for the local area for all the images, and when the processing for all the pixels is completed, the image is output (132). Assuming that the output image is fixed to, for example, 8-bit data in the system, first, a header portion exists before the image data, and a new gradation level is included in the header portion as index data. Has been inserted. In particular,
If the new gradation level is, for example, 16 levels, a data size obtained by adding 1 to the value is assigned to the header, and the number of new gradation levels and the value of the original gradation level corresponding to the new gradation level are changed. It is in order.

In the example of FIG. 8, the header data has a numerical value of 16 in the first byte, followed by 0,
9,13,20,25,29,33,36,41,4
7, 49, 52, 55, 63, 70 and 79 are inserted. After that, a value that expresses the index data corresponding to the gradation-converted value in bits and divides the data into units of 8 bits, followed by a byte unit, follows.

For example, when there are gradation values of 33, 79, 20, 36, and 49, each gradation value is replaced with the corresponding index data of 6, 15, 3, 7, and 10, respectively. Then, when each is expressed in bit units, “0”
110 "," 1111 "," 0011 "," 011
1 "," 1010 ". When these are connected and arranged,"
011011111001101111010 ", the value is again divided into 8-bit units, and
01111 "= 0x6F = 111," 0011011
1 ″ = 0x37 = 55, ‥‥. This means that data for two pixels can be represented by one byte. Thus, image data is created.

Next, an embodiment of an image processing apparatus for executing the above image processing will be described. FIG. 1 is a block diagram illustrating a configuration of the image processing apparatus. The image processing device 2 performs image processing on the image data output from the input device 1 and outputs the processed image data to the output device 3. The image processing device 2 includes a color conversion processing unit 4,
Tone level determination unit 5, color information detection unit 6, edge detection unit 7
And a gradation value selection unit 8.

The original image data input from the input device 1 is converted into a uniform color space by the color conversion in the color conversion processing section 4 of the image processing device 2. Next, using the luminance information of the uniform color space in the gradation level determination unit 5,
The histogram of the entire image is obtained, and further, in order to determine a new gradation level from the obtained distribution state of the histogram, the histogram is divided into a predetermined number within a range corresponding to the distribution state. Thereafter, a new gradation level is determined for each of the divided ranges of the histogram. The color information detection unit 6 divides the image data into local areas, and detects the color distribution state using chromaticity information in a uniform color space for each local area. Similarly, the edge detection unit 7 detects the presence or absence of an edge using luminance information for each local region. The gradation value selection unit 8 selects an optimum gradation value for each pixel in the local area from a new gradation level according to these detection results. Further, detection and gradation value determination are sequentially performed until the processing of the entire image is completed for each local region, and finally, the output image is output to the output device 3. The output device 3 performs conversion to a color space suitable for the output device 3 based on the information of the header portion of the image data output from the tone value selection unit 8, and performs output processing.

Next, each part of the image processing apparatus 2 using the above-described gradation value conversion method will be described in detail using specific embodiments. FIG. 9 is a diagram illustrating table data of RGB and LAB. FIG. 10 is a diagram illustrating a relationship between the selected table data and the conversion target data.

Here, as an example, it is assumed that 512
An example will be described in which 8-bit RGB color data is used for each color with an image size of × 512 pixels and an L ^* a ^* b ^* color space is used as a color space to be converted.

First, in the color conversion processing section 4, the input R
Convert GB data to L ^* a ^* b ^* data. To convert the data in the RGB color space to L ^* a ^* b ^* color space, preliminarily a table the relationship between the RGB color space and the L ^* a ^* b ^* color space for a number of representative colors, The color space can be converted by performing an interpolation operation using table data of a representative color close to the input pixel value. Therefore, R
In the GB color space, the representative color is selected such that the R value, the G value, and the B value are arranged at regular intervals. Then, the color patch is measured in the L ^* a ^* b ^* color system as a color patch, and stored in a memory as a look-up table as shown in FIG.

More specifically, for example, when the color of the input pixel data whose coordinates in the RGB color space are represented by (r, g, b) is converted into the data of the L ^* a ^* b ^* color space, The representative color closest to the input pixel value is selected. The representative color is a point corresponding to a vertex of a cube obtained by dividing a three-dimensional space around each color of RGB at equal intervals. As shown in FIG. 10, the input pixel value is changed to a point p (r,
When represented by g, b), the vertices p ₀₀₀ , p _{111 of the} cube including the point p therein are selected as the representative colors. Here, assuming that (r, g, b) = (150, 104, 32), based on the table data in FIG.
In the coordinates of _{000 -p 111, r 0 = 128} , r 1 = 1
92, g ₀ = 64, g ₁ = 128, b ₀ = 0, b ₁ = 6
4.

The data of each representative color selected in this way is
The data is converted into data in the L ^* a ^* b ^* color space based on the table data shown in FIG. For example, the point p ₀₀₀ (r ₀ ,
g ₀ , b ₀ ) = (128,64,0) is (L ^* , a
^* , B ^* ) = (61.54, 13.78, 61.72)
Is converted to The other selected representative colors are similarly converted.

The converted data of the input pixel value is
It is expressed by the following equation (4). Here, q in the following equation indicates a corresponding point on the L ^* a ^* b ^* color space after the above p is color-converted, and is represented by coordinates (L ^* , a ^* , b ^* ). .

Q (L ^* , a ^* , b ^* ) = c ₀ + c ₁ Δr + c ₂ Δg + c ₃ Δb + c ₄ Δr △ g + c ₅ Δr △ b + c ₆ Δg △ b + c ₇ Δr △ g △ b ‥‥ (4) _{where, △ r = r-r 0} = 150-128 = 22, △ g = g-g 0 = 104-64 = 40, △ b = b-b 0 = 32-0 = 32, _{c 0 = p 000, c 1} = (p 100 -p 000) / (r 1 -r 0), c 2 = (p 010 -p 000) / (g 1 -g 0), c 3 = (p 001 −p ₀₀₀ ) / (b ₁ −b ₀ ), c ₄ =
_{(P 110 -p 010 -p 100 +} p 000) / [(r 1 -r
_{0) (g 1 -g 0)} ], c 5 = (p 101 -p 001 -p 100 + p 000) / [(r 1
−r ₀ ) (b ₁ −b ₀ )], c ₆ = (p ₀₁₁ −p ₀₀₁ −p ₀₁₀ + p ₀₀₀ ) / [(g ₁
−g ₀ ) (b ₁ −b ₀ )], c ₇ = (p ₁₁₁ −p ₀₁₁ −p ₁₀₁ −p ₁₁₀ + p ₁₀₀ + p
₀₀₁ -p ₀₀₀ ) / [(r ₁ -r ₀ ) (g ₁ -g ₀ ) (b
₁₋ b ₀ )]. Thereby, (150, 10) on the RGB color space
The input pixel value of (4, 32) is represented by coordinates (72.85, 8.44, 69.70) in the L ^* a ^* b ^* color space. Such processing is performed for all input pixels of the input image data.

The tone level determining section 5 first counts each L ^* value which is luminance information for all image data in order to calculate a histogram of the entire image. In this case, the pixel data from the upper left pixel data to the lower right pixel data of the image are counted in order. L ^* is 0 to 100 when expressed only by integer values
Up to 101 levels, one step at a time. Therefore, for example, an array variable area for 101 pieces is secured and L ^*
In the array variable address corresponding to 0 to 100 of
Each time the luminance value is counted, the value of the corresponding address is sequentially added.

Subsequently, the total number of pixels is divided by a predetermined new number of gradations, and a value obtained by adding a certain number to the value is defined as one gradation level range. Count and divide the minutes. Further, the next count start position is set as a position where the value of the added fixed number or less is counted in the reverse direction from the pixel next to the last pixel of the count, and counting is sequentially performed. At the time of dividing the gradation level group, the number of pixels is counted, and when the specified value is reached in the middle of a certain gradation value, the count from the gradation value at which counting started to the current gradation value is performed. It is divided as one gradation level group.

For example, the new gradation number is set to 51 levels, the fixed number is set to 1000, and the value equal to or smaller than the fixed number to be counted is set to 500 which is 1/2 of the fixed number. The value of the range of the gradation level is 512 × 512/51 + 1000 = 6140. Here, the calculation result is rounded off to the first decimal place. If the minimum value of L ^* is 0, counting from 0 to 6140 pixels on the histogram is performed, and 0 to 6140 pixels are counted.
A new gradation level of 1 up to the gradation level of the 140th pixel
Set as one range. Then, the 5641th pixel, which is a value obtained by subtracting the value 500 to be counted in reverse from the 6141th pixel, is set as the start position of the next gradation level.
The counting for 40 pixels is started. This is repeated to perform division as a new gradation level group.

After the division to the end, the representative value is calculated for each of the divided gradation level groups. When a weighted average value is used as the representative value calculated here, it is assumed that 45
To 51 are one new gradation level range, level 45 is 500 pixels, 46 is 2100 pixels, 47 is 11
00 pixels, 48 are 800 pixels, 49 is 900 pixels, 50
Is 500 pixels and 51 is 240 pixels, the representative value is
(45 × 500 + 46 × 2100 +
47 × 1100 + 48 × 800 + 49 × 900 + 50 ×
(500 + 51 × 240) / 6140 = 47. Here, the calculation result is rounded off to the first decimal place.

The representative value thus obtained is stored as a new gradation level at a predetermined memory address. In the case where the center value is used as the representative value,
(51−45) / 2 + 45 = 48. When the maximum frequency value is used as the representative value, the number is 46 of 2100 pixels.

Next, the color information detecting section 6 performs feature extraction on the color information for each extracted local image. Here, a part of an 8 × 8 pixel area is taken as an example of the local image area. First calling the value of a ^* and b ^* of each pixel in a local area, Euclidean distance D ₁ of the from the origin using the equation (1).

X ₁ (a ₁ ^* , b ₁ ^* ) = (11, 23), x ₂ (a ₂ ^* , b ₂ ^* ) = (39, −45), x ₃ (a ₃ ^* , b ₃ ^*) ) = (42, −40), D ₁ (x ₁ ) = √ (a ₁ ^{* 2} + b ₁ ^{* 2} ) = 25, D ₁ (x ₂ ) = 60, D ₁ (x ₃ ) = 58 Become. Then, it is determined whether or not this value exceeds the first specified value for each pixel, and whether it is a chromatic color or an achromatic color is determined. For example, if the Euclidean distance is 3
If it is 0 or less achromatic and chromatic colors is greater than 30, the previously decided to determine, x ₁ is achromatic, x _2, x ₃ will be referred to as achromatic.

Further, the Euclidean distance between pixels on the a ^* -b ^* plane is obtained. First in order for each pixel in the local region, the Euclidean distance D ₂ between the other pixel determined by the equation (2). Then, D ₂ (x ₁ , x ₂ ) = √ ((a ₁ ^* −a ₂ ^* ) ² + (b ₁ ^* −b ₂ ^* ) ² ) = 74, D ₂ (x ₂ , x ₃ ) = 6 , D ₂ (x ₁ , x ₃ ) = 70. And for each pixel,
The Euclidean distance D ₂ is to determine whether it exceeds a second predetermined value, it is determined whether similar color.

For example, if the Euclidean distance is 10 or less, it is determined that it is not the same color when the Euclidean distance is larger than 10, it is not the same color. the x ₁ is achromatic, x ₂ and x ₃ is the fact that similar colors in a chromatic color.

Next, the edge detection unit 7 performs feature extraction on edge information for each extracted local image. Here, a part of an area of 8 × 8 pixels is taken as an example of the local image area.

First, the value of L ^* of each pixel in the local area is called, and the luminance difference between each pixel and each neighboring pixel is obtained by the above equation (3).

L ^* (x0, y0), L ^* (x1, y
0), ‥‥, L ^* (x6, y7), L ^* (x7, y7)
Is as shown in FIG. 7, P (x,
y) is L ^* (x1, y1), dL (1) = d
L (9) = 59, dL (2) = 35, dL (3) = 3
5, dL (4) = 4, dL (5) = 13, dL (6) =
5, dL (7) = 31, dL (8) = dL (0) = 38
Becomes

Then, of the values dL (1) to dL (8), those having a certain value or more (for example, dL (n)) are extracted from the respective values, and the results before and after (dL (n)) are extracted.
-1) and dL (n + 1)).

For example, if the fixed value is 25 or more, dL
Since (1) exceeds a certain value, dL (0) and dL
(2) is compared with dL (1). The difference between dL (0) and dL (2) is
If each is within a certain range, an edge exists in that direction. Assuming that this fixed range is ± 15
And keep it. At this time, the difference between dL (0) and dL (1) is 21, the difference between dL (2) and dL (1) is 24,
It is determined that there is no edge in the direction of dL (2). Similarly, since dL (2) exceeds a certain value, dL (1) and dL (1)
Comparing the difference with L (3), dL (1) and dL (2)
Is 24, and the difference between dL (2) and dL (3) is 0, and it is determined that there is an edge in the direction dL (2) -dL (3). Similarly, it is determined that there is an edge in the dL (7) -dL (8) direction. Thus, the presence or absence of an edge and the direction of the edge are detected for each pixel.

Next, the tone value selecting section 8 assigns a new tone value of each pixel in the local area to the array variable or the like from the two detection results of color information detection and edge information detection. Select from key levels.

For example, as a method of selecting a new gradation value, when the peripheral pixels including the target pixel are achromatic, when the gradation value of the target pixel does not belong to the two gradation level groups, the corresponding gradation is selected. When the tone value of the pixel of interest belongs to two tone level groups, if there is an edge, two tone levels are set so that a tone difference appears at the edge portion. From the new gradation level, select the gradation value with the larger gradation difference, and if there is no edge, select the gradation value closer to the surrounding new gradation levels from the two new gradation levels. .

When the peripheral pixel including the target pixel is a chromatic color, and the gradation value of the target pixel does not belong to two gradation level groups, the corresponding gradation value is selected, and When the tone value belongs to two tone level groups, the tone difference is changed from two new tone levels so that a tone difference appears at the edge part when there is an edge, depending on the presence or absence of the edge. Is selected, and if there is no edge, it is checked whether or not the pixel of interest has the same color as its surrounding pixels. If the colors are not the same, the color difference is set to 2
A tone value with a larger tone difference is selected from the two new tone levels, and if the colors are similar, a tone value closer to the neighboring new tone levels from the two new tone levels is selected. select.

In this way, the same processing is repeated for the local area of all the images, and when the processing of all the pixels is completed, the image is output. Finally, index data of a new gradation level is inserted as header information at the head of the output image data.

Finally, the output device 3 refers to the index data, which is the header portion of the image data, converts the image data into an output format suitable for the output device 3, and outputs the converted data to an output medium.

In the present embodiment, the L ^* a ^* b ^* color space is used as the color space to be converted. However, it goes without saying that the color space other than the L ^* a ^* b ^* color space may be used as long as it is a uniform color space. . For edge detection, the pixels to be compared are in the neighborhood of 8
It is not limited to pixels, and may be 8 pixels or less or 8 pixels or more.

Further, it goes without saying that the determination of achromatic / chromatic colors and the determination of similar colors may be other than the numerical values given in the examples. The color space may be converted not by table data but by matrix operation. In addition, in the present embodiment, the color space is divided into cubes as a method of obtaining an interpolation value at the time of table conversion. However, it may be divided into other shapes such as a triangular prism or a triangular pyramid.

By the image processing method using the above-mentioned gradation number conversion method, there is no problem such as color shift and pseudo gradation.
It is possible to reduce the amount of data and convert the number of gradations with less deterioration in image quality.

In addition, by using an image processing apparatus that executes the above-described image processing method, the input image input from the input device is free from color shift, pseudo gradation, and the like, and is reduced in image quality while reducing the data amount. This makes it possible to perform the conversion of the number of gradations with less and output an image of higher image quality to the output device than before.

[0097]

According to the present invention, the following effects can be obtained.

(1) In order to convert color image data having a plurality of gradations into color image data having a smaller number of gradations and a smaller data amount than the color image data, color conversion processing is performed on the image data. To obtain a histogram of the gradation levels in the converted color space, determine a new gradation level according to the distribution state of the gradation values on the histogram, and convert the image data subjected to the color conversion processing. Since the image is divided into local areas and a new gray level value is selected for each pixel of the local area based on the result of detecting the color distribution state and the presence / absence of an edge for each local area, the above-described image processing is performed. By executing the method, it is possible to obtain an image having no image quality deterioration without color shift or pseudo gradation.

(2) Since the uniform color space is the color space after the color conversion processing of the image data, the positional relationship of each peripheral pixel can be easily obtained from the center point of the uniform color space.

(3) A histogram is created from the luminance information of the entire image, and respective representative values are determined from a plurality of divided gradation level groups according to a predetermined new number of gradations and the frequency of each gradation value. By doing so, a new gradation level is determined, so that a new gradation value is determined according to the frequency of each gradation value, so that the number of gradations can be reduced according to the gradation characteristics of the image.

(4) The range of one gradation level is the number of pixels of a value obtained by adding a fixed number to a value obtained by dividing the total number of pixels of image data by the number of gradations of the new gradation level. The minute is counted from the maximum value or the minimum value of the gradation value on the histogram, the next count starts from the position where the predetermined value less than a certain number is counted backward from the last position of the previous count, and the count is sequentially performed. Is performed, and the current gradation level is divided into a plurality of gradation levels by dividing the current gradation level into a plurality of gradation levels. In the vicinity of a small number of gradation values, the gradation is broadly divided, and in the divided portion, there is a portion where adjacent gradations overlap, so that the gradation value can be selected according to the presence or absence of an edge, Frequency of appearance of gradation values in each image In addition, the number of gradations can be converted according to the characteristics of the image, and the deterioration of the image quality can be suppressed.

(5) By setting the weighted average value according to the distribution in each gradation level group as a representative value in each of the divided gradation level groups, the most average value in the divided gradation level groups can be obtained. Value can be determined as a representative value.

(6) Since the central value in each gradation level group is set as a representative value in each of the divided gradation level groups, it can be determined as a representative value by a simple calculation.

(7) By setting the maximum frequency value in each gradation level group as a representative value in each of the divided gradation level groups, simply selecting the value with the highest frequency as the representative value is simple. It can be determined and can be set to a value appropriate as a representative value of the divided gradation level group.

(8) By adding each representative value of the divided gradation level group to the converted new gradation image data, the relationship between the original luminance and the luminance of the converted gradation value is obtained. Can be reconverted into a color space suitable for each output device without changing the contrast.

(9) In order to detect a color distribution state, each pixel is determined to be chromatic or achromatic according to the distance from the origin of each pixel on the chromaticity coordinates of the converted color space, and each pixel on the coordinates is determined. Since it is determined whether or not the colors are similar from each other based on their positional relationship, they are grouped, so it is easy to determine chromatic, achromatic, and similar colors from the distance from the origin on the chromaticity coordinates and their positional relationship. it can.

(10) If the Euclidean distance from the origin on the chromaticity coordinates of the converted color space of each pixel is less than or equal to the first specified value, it is determined that the pixel is achromatic, and the Euclidean distance of each pixel from the origin is determined. If it is more than the first specified value, it is determined to be a chromatic color, so it is possible to distinguish between an achromatic color and a chromatic color by a simple method. Can be grouped.

(11) If the Euclidean distance from the origin on the chromaticity coordinates between the pixels is equal to or smaller than the second specified value, each pixel is determined to be of the same color, so that the same color and non-similar color are determined by a simple method. Colors can be distinguished.

(12) The presence or absence of an edge is detected by examining the distribution of the luminance difference between each pixel in the local area and the neighboring pixels.
If there is a luminance difference between each pixel in the local region and a neighboring pixel, and if the luminance difference is continuous, an edge exists there, so that the presence or absence of the edge can be easily determined.

(13) If the peripheral pixels including the target pixel are achromatic and the gradation value of the target pixel belongs to only one gradation level group, the gradation value of this gradation level group is used as the gradation value. Since the color shift does not matter in the case of an achromatic color, the gradation value can be simply selected.

(14) When the peripheral pixel including the target pixel is an achromatic color and the gradation value of the target pixel belongs to two gradation level groups, the presence of an edge increases the gradation difference from the two new gradation levels. In the case of an achromatic color, there is no color misregistration because the grayscale value is selected, and if there is no edge, the grayscale value closer to the new grayscale level is selected from the two new grayscale levels. Therefore, when belonging to two gradation level groups, an optimum gradation value can be selected only by the presence or absence of an edge.

(15) If the peripheral pixels including the target pixel are chromatic and the gradation value of the target pixel belongs to only one gradation level group, the gradation value of this gradation level group is set as a new gradation value. If the peripheral pixel including the pixel of interest is a chromatic color and the gradation value of the pixel of interest belongs to two gradation level groups, the gradation value is selected according to the presence / absence of an edge and the presence / absence of a similar color. Therefore, in the case of a chromatic color, a color shift is also involved, so that not only the presence or absence of an edge but also whether or not it is a similar color can be checked to select an optimum tone value for each.

(16) When the peripheral pixels including the target pixel are chromatic and the gradation values of the target pixel belong to two gradation level groups,
If there is an edge, a tone value with a larger tone difference is selected from the two new tone levels, and if there is no edge, it is checked whether or not the target pixel is the same color as its surrounding pixels. If there is not, select the tone value with the larger tone difference from the two new tone levels, and if it is the same color, select the tone value closer to the surrounding new tone levels from the two new tone levels Since the selection is made, in the case of a chromatic color, a pseudo tone due to color misregistration or color is also involved. Therefore, the presence or absence of an edge and the presence of a similar color are checked, and a tone value that makes it difficult to generate a pseudo tone is selected. be able to.

(17) The image processing apparatus includes a color conversion processing section, a gradation level determination section, a color information detection section, an edge detection section,
A tone value selection unit, which performs color conversion processing on image data input from the input device in a color conversion processing unit, and performs a color conversion process on a color space converted in a color conversion processing unit in a tone level determination unit. A histogram of the gradation level is obtained, a new gradation level is determined according to the distribution state of the gradation values on the histogram, and the color information detecting unit divides the image data into local regions,
The color distribution state is detected for each local area, the image data is divided into local areas by the edge detection section, the presence or absence of an edge is detected for each local area, and the color information detection section and the edge detection are detected by the gradation value selection section. To select a gradation value from the new gradation level determined by the gradation level determination unit based on the detection result with the unit and output it to the output device, the gradation level is determined from the distribution state of the histogram of the entire image. By detecting the distribution state of the color of the local region and the presence or absence of an edge, and determining an optimum gradation value from the gradation level from each result,
It is possible to realize an image processing apparatus that performs tone number conversion without color shift or pseudo tone and with little image quality deterioration.

[Brief description of the drawings]

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

FIG. 2 is a flowchart showing an embodiment of a gradation number conversion method according to the present invention.

FIG. 3 is a diagram illustrating a method of obtaining histograms of all pixels.

FIG. 4 is a diagram illustrating a method for obtaining a new gradation level.

FIG. 5 is a diagram illustrating a method of extracting an achromatic region.

FIG. 6 is a diagram illustrating a method of extracting similar colors in a chromatic color area.

FIG. 7 is a diagram illustrating a method of extracting an edge.

FIG. 8 is a diagram illustrating a result of conversion to a new luminance level.

FIG. 9 is a diagram illustrating table data of RGB and LAB.

FIG. 10 is a diagram illustrating a relationship between selected table data and conversion target data.

[Explanation of symbols]

 Reference Signs List 1-input device 2-image processing device 3-output device 4-color conversion processing unit 5-gradation level determination unit 6-color information detection unit 7-edge detection unit 8-gradation value selection unit 9-conversion target data Surrounding table data 10-Conversion target data

Continued on front page F-term (reference) 5B057 CA01 CA08 CA12 CA16 CB01 CB08 CB12 CB16 CC01 CE11 CE17 DA17 DB02 DB06 DB09 DC16 DC19 5C077 MP01 MP08 NN02 PP36 PP37 PP39 PP47 PQ08 PQ12 PQ17 PQ18 PQ19 PQ23 LA01 LA10 LA01 LA01 LA10 LA01 LA10 LA01 LA10 LA01 LA31 LB11 NA02 5L096 AA02 AA06 FA06 FA15 FA37 FA66 GA28 GA41 GA43 MA07

Claims (17)

[Claims] 1. An image processing method for converting color image data having a plurality of gradations into color image data having a smaller number of gradations and a smaller amount of data than the color image data, comprising: A conversion process is performed to obtain a histogram of gradation levels in the converted color space, a new gradation level is determined according to the distribution state of the gradation values on the histogram, and the color-converted image data is Dividing into local areas, detecting a color distribution state and the presence or absence of an edge for each local area, and selecting the new gray level value for each pixel of the local area based on the detection result. Image processing method. 2. The image processing method according to claim 1, wherein a uniform color space is used as a color space after color conversion processing of the image data. 3. A histogram is created from luminance information of the entire image, and a current gradation level is divided into a plurality of parts according to a predetermined new number of gradations and a frequency of each gradation value. 2. The image processing method according to claim 1, wherein each of the representative values is determined from the following, and the new gradation level is determined. 4. A gradation on a histogram, wherein the number of pixels of a value obtained by adding a fixed number to a value obtained by dividing the total number of pixels of image data by the number of gradations of a new gradation level is defined as one gradation level range. While counting and dividing the number of pixels from the maximum value or the minimum value of the value, and starting the next count from the position where the predetermined value equal to or less than the predetermined number is counted backward from the last position of the previous count, 4. The image processing method according to claim 3, wherein counting is performed sequentially, and the current gradation level is divided into a plurality of gradation values corresponding to the number of pixels. 5. The image processing method according to claim 3, wherein the representative value is a weighted average value according to a distribution in each gradation level group. 6. The image processing method according to claim 3, wherein the representative value is a center value in each gradation level group. 7. The image processing method according to claim 3, wherein the representative value is a maximum frequency value in each gradation level group. 8. The image processing method according to claim 5, wherein the representative value is added to the converted new gradation image data. 9. Using the chromaticity coordinate values of the converted color space,
By judging a chromatic or achromatic color from the distance from the origin on the coordinates of each pixel, judging whether or not they are the same color from the mutual positional relationship on the coordinates of each pixel and grouping them, the color distribution The image processing method according to claim 1, wherein the state is detected. 10. In each pixel, when the Euclidean distance from the origin on the chromaticity coordinates is equal to or less than a first specified value, the pixel is determined to be achromatic, and the Euclid distance from the origin is smaller than the first specified value. The image processing method according to claim 9, wherein when it is distant, it is determined that the color is chromatic. 11. The image processing method according to claim 9, wherein when the Euclidean distance from the origin on the chromaticity coordinates between the pixels is equal to or smaller than a second specified value, the pixels are determined to be similar colors. 12. The image processing apparatus according to claim 1, wherein the presence or absence of the edge is detected by determining a difference between the luminance of each pixel in the local area and the luminance of a neighboring pixel, and examining the distribution of the difference. Method. 13. When the peripheral pixels including the target pixel are achromatic and the gradation value of the target pixel belongs to only one gradation level group, the gradation value of the gradation level group is selected. The image processing method according to claim 1. 14. When the peripheral pixels including the target pixel are achromatic and the gradation values of the target pixel belong to two gradation level groups,
When there is an edge, the tone value with the larger tone difference is selected from the two new tone levels, and when there is no edge, the tone value closer to the neighboring new tone levels from the two new tone levels is selected. The image processing method according to claim 1, wherein a tone value is selected. 15. When a peripheral pixel including a target pixel is a chromatic color and the target pixel belongs to only one gray level group, the gray level of the gray level group is selected as a new gray level value. When the peripheral pixel including the target pixel is a chromatic color and the target pixel belongs to two gray level groups, the gray value is selected according to the presence / absence of an edge and the presence / absence of a similar color. The image processing method according to claim 1. 16. When a peripheral pixel including the target pixel is a chromatic color and the gradation value of the target pixel belongs to two gradation level groups,
When there is an edge, a tone value with a larger tone difference is selected from the two new tone levels, and when there is no edge, it is checked whether or not the pixel of interest is the same color as its surrounding pixels,
If the colors are not the same, select the tone value with the larger tone difference from the two new tone levels, and if the colors are the same, the two new tone levels are closer to the surrounding new tone levels. 2. The image processing method according to claim 1, wherein one of the gradation values is selected. 17. An image processing apparatus for converting color image data of a plurality of gradations input from an input device into color image data having a smaller number of gradations and a smaller data amount than the color image data, and outputting the color image data to an output device. A color conversion processing unit for performing color conversion processing on the image data input from the input device; and obtaining a histogram of gradation levels in the color space converted by the color conversion processing unit. A gradation level determination unit that determines a new gradation level according to a distribution state of a tonal value, a color information detection unit that divides image data into local regions, and detects a color distribution state for each local region, An edge detection unit that divides the image into regions and detects the presence or absence of an edge for each local region; and a new floor determined by the gradation level determination unit based on detection results of the color information detection unit and the edge detection unit. The image processing apparatus characterized by comprising a gradation value selecting section that outputs from the level select the tone value to the output device.