JP2006074809A - Image processor and image processing method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To suppress deterioration in image quality caused by a quantization error caused by the conversion of the number of gradations in converting the number of gradations by assigning image data to a lattice point preset in a color space. <P>SOLUTION: A color space is divided in advance to set lattice points and in a gradation number pre-converting unit 140, original color image data ORG are assigned to near lattice points. In such a case, a quantization error is incurred by assigning the original image data to the lattice points however the data are assigned so that the error becomes less than a predetermined value in average, thereby preserving final image colors. When setting the lattice points, a density area is divided smaller as progressing toward a low density area, to minimize deterioration in image quality caused by the quantization error. Furthermore, smoothing is performed after the pre-conversion of the gradation number, thereby reducing influences of the quantization error and improving image quality. <P>COPYRIGHT: (C)2006,JPO&NCIPI

Description

  The present invention relates to an image processing apparatus and an image processing method, and more particularly to an image processing apparatus and method for performing color correction processing on input color image data and outputting the same.

  2. Description of the Related Art Conventionally, there has been known an image processing apparatus that reads a color original using an image input unit such as a scanner, and reproduces and displays the read image data using a display such as a CRT or a color printer. .

  Since image output devices such as displays and color printers have specific color reproduction characteristics, the colors used for the color image input using a scanner or the like are reproduced well regardless of the characteristics of the output device. A method of performing color correction processing in accordance with the color reproduction characteristics of the output device has been proposed. One such color correction technique is disclosed in Japanese Patent Application Laid-Open No. 63-2669. In this prior art, an RGB three-dimensional color correction table corresponding to all combinations of three color components of red (hereinafter referred to as R), green (hereinafter referred to as G), and blue (hereinafter referred to as B) is prepared. is doing. This color correction table stores in advance color correction contents for all positions in the color space represented by three-dimensional coordinates. The image processing apparatus performs color correction by referring to the color correction table.

  This color correction method is not practical enough because the storage capacity of the color correction table to be used becomes enormous. For example, when the input original color image data has a gradation number of 8 bits (256 gradations) for each of R, G, and B colors, the number of colors is 256 to the power of about 16.78 million colors. If the data after color correction is also 8 bits, a storage capacity of 48 megabytes is required as a color correction table for R, G, B3 colors.

  On the other hand, in order to avoid an increase in the capacity of the color correction table, the color correction table is composed only of data corresponding to limited grid points, and for the color data between grid points, the grid point data is used. There have also been proposed ones that perform color correction processing by performing an interpolation calculation (for example, Japanese Patent Laid-Open Nos. HEI 14414481 and H4-185075). According to this method, the storage capacity of the color correction table can be reduced, but considerable time is required for the interpolation calculation. In particular, when the resolution of image data handled as recently becomes higher and the gradation expression for each pixel becomes finer, the computation time required to output one image becomes longer, and the image output is completed. The waiting time becomes extremely long. However, if the interpolation operation is simplified, the color reproducibility is lowered.

  Accordingly, the applicant of the present application has proposed an image processing technique that does not require a large amount of time for color correction interpolation calculation without increasing the capacity of the color correction table (for example, Japanese Patent Laid-Open No. 7-30772). See the official gazette). This image processing method divides the color space at a predetermined interval, prepares color correction data only for the lattice points obtained by the division, and for image data other than the lattice points, These are assigned to grid points and color correction is performed without interpolation. In this case, since the original image data is allocated to the neighboring grid points, an error occurs every time the grid points are allocated. Therefore, when the image data of each pixel constituting the image is sequentially processed, the assignment to the grid points is performed so that this error becomes as small as possible on average.

  In the above image processing method, if attention is paid to each pixel, there is an error in its color, but the error is eliminated in a certain range, without reducing the quality of the output image, Although it is excellent in that the calculation time required for image processing is significantly shortened, when the number of gradations that can be expressed by the image output apparatus that outputs the image is low, for example, only binary expression like an ink jet printer is possible. If this is not possible, the image quality may be slightly degraded in areas where the image density is low. This is because, in a low density region where the dot density is small and the dots are sparsely distributed, even if a little noise (quantization error caused by allocation to the lattice points), the appearance position of the dots is greatly shifted.

  In the above image processing, even if pixels of the same color are close to each other in the original image, both pixels may be assigned to different grid points. Therefore, in order to reduce the storage capacity of the color correction table, If the division is made rough and the number of grid points is reduced, the color separation between the output pixels is large, and as a result, the quality of the image may be lowered.

  An object of the present invention is to solve these problems and further improve the quality of an output image centering on a low density area without increasing the number of arithmetic processes for color correction.

The first image processing apparatus of the present invention that achieves the above-described object provides:
An image processing apparatus for color-correcting and outputting a multicolor image expressed by coordinate values in a two-dimensional or higher color space,
For each pixel of the image, input means for inputting color image data expressing the coordinate value using a predetermined number of gradations;
The color space is divided by the number of gradations smaller than the number of gradations representing the coordinate value, and the color space is divided more finely than the other areas in a predetermined low density area of the color space, and the division is performed. Grid point information storage means for storing the coordinate values of grid points obtained by performing for each dimension for the color space;
A color correction table storing correction data related to the color of the color image data corresponding to each grid point;
The coordinate values of the input color image data in the color space are determined according to a method in which a distance from the grid point is an average value or less on the average of grid points stored in the grid point information storage unit. Grid point conversion means for converting to coordinate values;
The gist of the present invention is to provide color correction means for reading out correction data of lattice points corresponding to the converted coordinate values from the color correction table and outputting them as corrected color image data.

The first image processing method corresponding to this image processing apparatus is
An image processing method for color-correcting and outputting a multicolor image represented by coordinate values in a two-dimensional or higher color space,
For each pixel of the image, input color image data expressing the coordinate value using a predetermined number of gradations,
The color space is divided by the number of gradations smaller than the number of gradations representing the coordinate value, and the color space is divided more finely than the other areas in a predetermined low density area of the color space, and the division is performed. Storing the coordinate value of the grid point obtained by performing for each dimension for the color space,
Corresponding to each grid point, a color correction table storing correction data related to the color of the color image data is prepared,
The coordinate value in the color space of the input color image data is converted into the coordinate value of the stored grid point according to a method in which the distance from the grid point is equal to or less than a predetermined value on average,
The gist is to read out correction data of lattice points corresponding to the converted coordinate values from the color correction table and output the corrected color image data.

  In this image processing apparatus and image processing method, it is obtained by dividing the color sky angle by the number of gradations smaller than the number of gradations expressing the coordinate value in the color space of two or more dimensions, and performing the division for each dimension. The coordinate values of the grid points are stored, and color correction data storing correction data relating to the color of the color image data corresponding to each grid point is prepared. For each pixel of the image, color image data expressing coordinate values using a predetermined number of gradations is input, and this coordinate value is converted into a grid according to a method in which the distance from the grid point is less than or equal to the predetermined value on average. Convert to the coordinate value of a point. Such conversion eliminates the need for performing a complex interpolation operation using a plurality of grid point data. In addition, since the grid points are divided finely in the low density region, the interval of the data of the color correction table in the low density region is also fine. Therefore, the quantization error when converting the coordinate value of the input data into the coordinate value of the lattice point becomes small in the low density region where it becomes a problem, and deterioration of the image quality is suppressed. As a result, the color correction calculation adapted to the image output apparatus is performed at high speed, good color reproduction is obtained, and deterioration of the image quality is suppressed.

  The low density region in the present invention refers to a region where the density of dots output to the image output apparatus is low. For example, when the final image output is an ink jet printer that expresses gradation by turning dots on and off, it means an area where the density of ink dots such as CMY is low. In addition, when the output device is a CRT or the like, white dots are sparsely distributed if attention is paid to white dots (black dots are sparsely distributed if attention is paid to black dots). This is a region to be used (low density region in the entire image). In addition, as will be described later, if the printer has high density ink and low density ink for the same color and separates dark dots from light dots, the area where light dots are sparsely distributed (low in the entire image). Not only the density area) but also the area where dark dots are sparsely distributed corresponds to the low density area for the ink.

  Note that the color space of two or more dimensions includes not only RGB and CMY color spaces, but also a color space represented by an XYZ color system, an L * a * b * color system, and an L * C * h table. Various color spaces such as a color system and Munsell color system can be considered. As the number of gradations, when these coordinate values are expressed by digital information of n bits, they are often expressed by gradations of 2 n (for example, 256 for 8 bits). There may be 100 gradations, 17 gradations, or the number of gradations other than 2 to the power of n. In the above configuration, the coordinates of the grid points are finely divided in the low density region, but the gradation interval of each color component of the color image data after being corrected by the color correction unit is narrowed in the low density region. Thus, it is also preferable to select. As a result, the quantization error of the pre-gradation number conversion in the low density area in the data after color correction is reduced, so that the deterioration in image quality due to the division of the color space with a small number of gradations is more reliably suppressed. be able to.

  It is desirable that the number of gradations of the image data output by this image processing is larger than the final number of gradations suitable for the image output apparatus. In this case, this image data is further converted to the final number of gradations and then supplied to the image output device. However, the quantization error in the gradation number conversion due to the assignment to the grid points described above. Is smaller than the error that occurs during the conversion to the final number of gradations suitable for the output device. As a result, image quality deterioration due to the introduction of tone number conversion by assignment to grid points is suppressed.

  The above-described conversion of the image coordinate values into the coordinate values of the grid points can be performed using an error diffusion technique. It is also possible to convert the coordinates of lattice points by a systematic dither method using a distributed dither threshold matrix. According to these methods, the quantization error can be appropriately dispersed.

  In addition, as one of the methods for taking a color space, when color image data is configured by combining a plurality of basic color densities, such as RGB and CMY, the number of gradations expressing coordinate values in the color space Is a format that directly corresponds to the density of each color component of the color image data, and the color correction table stores information on the gradation, that is, density, of each color component of the color image data. In this case, the grid point information divides a predetermined low density area of the color space so that the gradation interval of each color component of the color image data is narrower than the other density areas in the predetermined low density area. What should I do?

  A color correction table for correcting the color of the color image data can store correction data that is suitable for the color reproduction characteristics of the image output apparatus that finally processes the color image data. Various purposes can be considered for color correction. Of these uses, correction in accordance with the color reproduction characteristics of the image output apparatus is one of the things that are often required in image processing.

The second image processing apparatus of the present invention is
An image processing apparatus for color-correcting and outputting a multicolor image expressed by coordinate values in a two-dimensional or higher color space,
For each pixel of the image, input means for inputting color image data expressing the coordinate value using a predetermined number of gradations;
The color space is divided by the number of gradations smaller than the number of gradations representing the coordinate value, and the color space is divided more finely than the other areas in a predetermined low density area of the color space, and the division is performed. Grid point information storage means for storing the coordinate values of grid points obtained by performing for each dimension for the color space;
A color correction table storing correction data related to the color of the color image data corresponding to each grid point;
The coordinate values of the input color image data in the color space are determined according to a method in which a distance from the grid point is an average value or less on the average of grid points stored in the grid point information storage unit. Grid point conversion means for converting to coordinate values;
Color correction data reading means for reading correction data of lattice points corresponding to the converted coordinate values from the color correction table;
A value corresponding to the main color is selected from the read correction data of the grid point, and for each main color, the correction data of the adjacent grid point in which only the coordinate value corresponding to the main color is shifted from the grid point. A value corresponding to the main color is read out, interpolation is performed using both values, and the correction data obtained by replacing the correction data for the main color with the interpolation result is output as final corrected color image data. And a color correcting means for performing the above.

An image processing method corresponding to this image processing apparatus is as follows:
An image processing method for color-correcting and outputting a multicolor image represented by coordinate values in a two-dimensional or higher color space,
For each pixel of the image, input color image data expressing the coordinate value using a predetermined number of gradations,
The color space is divided by the number of gradations smaller than the number of gradations representing the coordinate value, and the color space is divided more finely than the other areas in a predetermined low density area of the color space, and the division is performed. Storing the coordinate value of the grid point obtained by performing for each dimension for the color space,
Corresponding to each grid point, a color correction table storing correction data related to the color of the color image data is prepared,
The coordinate value in the color space of the input color image data is converted into the coordinate value of the stored grid point according to a method in which the distance from the grid point is equal to or less than a predetermined value on average,
Read out the correction data of the grid points corresponding to the converted coordinate values from the color correction table,
A value corresponding to the main color is selected from the read correction data of the grid point, and for each main color, the correction data of the adjacent grid point in which only the coordinate value corresponding to the main color is shifted from the grid point. A value corresponding to the main color is read out, interpolation is performed using both values, and the correction data obtained by replacing the correction data for the main color with the interpolation result is output as final corrected color image data. The gist is to do.

  According to the image processing apparatus and the image processing method, as in the first image processing apparatus and the image processing method, it is possible to obtain a good result with little deterioration of the image quality in the low density region. Furthermore, in the second technique, since only the main color is interpolated using the correction data read from the color correction table, the grid points of the coordinate values in the color space are not substantially caused to complicate the calculation. Degradation of image quality due to conversion to the coordinate value of can be suppressed. This interpolation processing may be performed only in the low density region where the influence of the quantization error of the pre-gradation number conversion is a problem.

  As an interpolation process performed only for the main color, a simple one-dimensional interpolation operation such as linear interpolation can be performed. This is because the conversion of the coordinate value is originally performed so that the distance from the grid point is on average equal to or less than a predetermined value, and thus even a simple interpolation operation can sufficiently obtain the effect. Of course, it is possible to perform a higher-order interpolation operation using the coordinate values of other adjacent grid points.

In such an image processing apparatus, the coordinate value of the color image data is defined as a value corresponding to the density of each color component of the color image data, and the lattice point conversion means converts the conversion of the lattice point coordinates as the gradation number conversion. First gradation number conversion means to perform,
Furthermore, a second gradation number conversion means for converting the gradation number for each color component of the corrected color image data output from the color correction means into a gradation number suitable for the image output apparatus. Prepared,
The number of gradations obtained by the coordinate value conversion by the first gradation number conversion means may be greater than the number of gradations after being converted by the second gradation number conversion means.

The third image processing apparatus of the present invention
An image processing apparatus for color-correcting and outputting a multicolor image expressed by coordinate values in a two-dimensional or higher color space,
For each pixel of the image, input means for inputting color image data expressing the coordinate value with a predetermined number of gradations;
A grid in which the color space is divided by the number of gradations smaller than the number of gradations expressing the coordinate value, and the coordinate values of the grid points obtained by performing the division for each dimension are stored for the color space. Point information storage means;
A color correction table storing correction data related to the color of the color image data corresponding to each grid point;
The coordinate values of the input color image data in the color space are determined according to a method in which a distance from the grid point is an average value or less on the average of grid points stored in the grid point information storage unit. Grid point conversion means for converting to coordinate values;
Color correction data reading means for reading correction data of lattice points corresponding to the converted coordinate values from the color correction table;
The gist of the invention is that it comprises an averaging processing means for performing a process of averaging the correction data of each pixel read by the color correction data reading means based on the correction data of the pixels in the vicinity of each pixel.

An image processing method corresponding to this image processing apparatus is as follows:
An image processing method for color-correcting and outputting a multicolor image represented by coordinate values in a two-dimensional or higher color space,
For each pixel of the image, input color image data expressing the coordinate value with a predetermined number of gradations,
Dividing the color space by the number of gradations smaller than the number of gradations representing the coordinate value, storing the coordinate values of grid points obtained by performing the division for each dimension for the color space;
Corresponding to each grid point, a color correction table storing correction data related to the color of the color image data is prepared,
The coordinate value in the color space of the input color image data is converted into the coordinate value of the stored grid point according to a method in which the distance from the grid point is equal to or less than a predetermined value on average,
Read out the correction data of the grid points corresponding to the converted coordinate values from the color correction table,
The gist is to perform the process of averaging the read correction data of each pixel based on the correction data of the pixels in the vicinity of each pixel.

  According to the image processing apparatus and the image processing method, the conventional complicated interpolation calculation can be omitted by the method of converting the coordinate value in the color space into the coordinate value of the grid point, and good color reproduction can be obtained. . Further, the image quality obtained by converting the coordinate value of each pixel into the coordinate value of the grid point by performing an averaging process (smoothing process) based on the correction data of the neighboring pixels with respect to the correction data after color correction. Can be prevented. This smoothing process may be performed only in the low density region where the influence of the quantization error of the coordinate value conversion is a problem.

  It is desirable that the number of gradations of the image data output by this image processing is larger than the final number of gradations suitable for the image output apparatus. In this case, this image data is further converted to the final number of gradations and then supplied to the image output device. However, the quantization error in the gradation number conversion due to the assignment to the grid points described above. Is reduced by the conversion to the final number of gradations suitable for the output device. As a result, image quality deterioration due to the introduction of tone number conversion by assignment to grid points is suppressed. This is the same as the first image processing apparatus and the image processing method.

  The above-described conversion of the image coordinate values into the coordinate values of the grid points can be performed using an error diffusion technique. It is also possible to convert the coordinates of lattice points using a threshold matrix of distributed dither. According to these methods, the quantization error can be appropriately dispersed. Furthermore, it is possible to adopt a configuration in which which method is used is switched based on the properties of the image. The error diffusion method is excellent in error dispersion, but generally requires more arithmetic processing. On the other hand, the distributed dither method is superior in terms of high-speed processing. Therefore, it is also desirable to switch between both methods and take advantage of their respective characteristics. In this case, the averaging process may be performed in a region where an error diffusion technique is used and the density of the image data is low.

  Various pixels can be considered as pixels to be averaged when the above averaging process is performed. For example, neighboring pixels when the averaging process is performed can be adjacent pixels along the direction in which the color image data is input. Normally, since the image processing is performed in the order of image data input, if the adjacent pixels are in the input direction, the processing becomes easy. As an adjacent pixel along the input direction, a pixel on the near side in the input direction can be used, or a pixel on the rear side can be used. Furthermore, it is possible to take an average of adjacent pixels on both sides in front and rear. In addition, as neighboring pixels on which the averaging process is performed, pixels adjacent to the direction in which color image data is input and pixels in a direction intersecting with this direction can also be used. As the averaging process, it is possible to use a weighted average value obtained by performing predetermined weighting in addition to a simple average value.

  Here, the coordinate value of the color image data is defined as a value corresponding to the density of each color component of the color image data, and the grid point conversion means performs the conversion of the grid point coordinates as the gradation number conversion. 1 gradation number conversion means, and further converts the gradation number for each color component of the corrected color image data output from the color correction means into a gradation number suitable for the image output apparatus. Second gradation number conversion means for converting the gradation number obtained by the coordinate value conversion by the first gradation number conversion means after the conversion by the second gradation number conversion means. It can be larger than the logarithm. In this case, the quantization error by the first gradation number conversion means is reduced by the conversion of the gradation number by the second gradation number conversion means, and the influence of image quality degradation due to the former quantization error is small. Become.

  Here, the second gradation number converting means is a means for performing binarization and expressing the gradation by the distribution density of the binarized dots, and the averaging processing means is for the binarized dots. When the density is equal to or lower than a predetermined value, it can be a means for performing an averaging process. In the case of an image output apparatus that performs binarization, the density is expressed by the density of the dots. Therefore, if the averaging is performed when the density is equal to or less than a predetermined value, the image quality in the low density area is degraded. Can be prevented.

  Further, in such an image processing apparatus, the averaging processing means is provided with means for comparing the color correction data read by the color correction data reading means with the color correction data of the adjacent pixels, and the difference between the compared two color correction data. When is equal to or less than a predetermined value, an averaging process can be performed. If the averaging process is performed uniformly, the sharpness of the image tends to be lost. Since the color of adjacent pixels is often largely separated at the place where there is an edge from the beginning, if averaging is performed when the difference in color correction data is less than or equal to a predetermined value, averaging is performed at the original edge of the image. This processing is not performed, and both smoothness of image quality and sharpness can be achieved.

  Whether or not to perform the averaging process is determined for at least one of the colors constituting the color image data, and the averaging process is performed for each color based on the determination of the color. It is also good.

  Alternatively, the averaging process may be performed when it is estimated that the distance in the color space of the color image data input for adjacent pixels is equal to or less than a predetermined distance.

  Various judgment methods can be considered for estimating whether or not the separation of the color image data in the color space is a predetermined distance or less. For example, when at least adjacent lattice points converted by the lattice point converting means for adjacent pixels are adjacent to each other, it can be estimated that the distance between the adjacent pixels is equal to or less than a predetermined distance. Further, when the color image data for adjacent pixels is included in the unit space formed by the lattice points stored in the lattice point information storage means, the distance between adjacent pixels is equal to or less than a predetermined distance. Can also be estimated. When pixels having the same or almost the same color in the original image are assigned to different grid points as a result of conversion by the grid point conversion means, the averaging process loses the fineness and resolution of the image. This is because the image quality can be improved without any problems.

  When it is estimated that the distance in the color space of adjacent pixels in the color image data is less than or equal to a predetermined distance, it is determined whether other execution conditions for the averaging process are satisfied, and the two adjacent pixels are The averaging process can be performed only when it is determined that the distance is equal to or less than the predetermined distance and other execution conditions are satisfied. As other execution conditions, various conditions can be considered such as when the image data is below a predetermined density.

  Also in an image processing apparatus that performs such averaging processing, the grid point information storage means stores the coordinates of grid points in a predetermined low density area of the color space by dividing the color space more finely than other areas. can do. As a result, the image quality in the low density region is further improved.

Furthermore, the fourth image processing apparatus of the present invention provides:
An image processing apparatus for color-correcting and outputting a multicolor image expressed by coordinate values in a two-dimensional or higher color space,
For each pixel of the image, input means for inputting color image data expressing the coordinate value with a predetermined number of gradations;
A grid in which the color space is divided by the number of gradations smaller than the number of gradations expressing the coordinate value, and the coordinate values of the grid points obtained by performing the division for each dimension are stored for the color space. Point information storage means;
A color correction table storing correction data related to the color of the color image data corresponding to each grid point;
The coordinate values of the input color image data in the color space are determined according to a method in which a distance from the grid point is an average value or less on the average of grid points stored in the grid point information storage unit. Grid point conversion means for converting to coordinate values;
Color correction data reading means for reading correction data of lattice points corresponding to the converted coordinate values from the color correction table;
A value corresponding to the main color is selected from the read correction data of the grid point, and for each main color, the correction data of the adjacent grid point in which only the coordinate value corresponding to the main color is shifted from the grid point. Color correction means for reading out a value corresponding to the main color, performing interpolation using both values, and outputting the correction data obtained by replacing the correction data for the main color with the interpolation result. The gist.

An image processing method corresponding to this image processing apparatus is as follows:
An image processing method for color-correcting and outputting a multicolor image represented by coordinate values in a two-dimensional or higher color space,
For each pixel of the image, input color image data expressing the coordinate value with a predetermined number of gradations,
Dividing the color space by the number of gradations smaller than the number of gradations representing the coordinate value, storing the coordinate values of grid points obtained by performing the division for each dimension for the color space;
Corresponding to each grid point, a color correction table storing correction data related to the color of the color image data is prepared,
The coordinate value in the color space of the input color image data is converted into the coordinate value of the stored grid point according to a method in which the distance from the grid point is equal to or less than a predetermined value on average,
Read out the correction data of the grid points corresponding to the converted coordinate values from the color correction table,
A value corresponding to the main color is selected from the read correction data of the grid point, and for each main color, the correction data of the adjacent grid point in which only the coordinate value corresponding to the main color is shifted from the grid point. The gist is to read out the value corresponding to the main color, perform interpolation using both values, and output the correction data in which the correction data for the main color is replaced with the interpolation result.

  According to this image processing method, the interpolation calculation is performed only for the main color, so that the calculation amount is small, and the quantization error for the main color is eliminated by the interpolation. Therefore, it is possible to minimize degradation of visually recognized image quality without reducing the processing speed.

  In addition, it is also possible to perform the lattice point conversion using the lattice points finely divided in the low density region, the interpolation processing for the main color, and the averaging processing between adjacent lattice points in combination. . In this case, by taking advantage of the respective advantages, the time required for calculation can be shortened, and deterioration in image quality due to grid point conversion can be minimized.

Next, a preferred embodiment of the image processing apparatus of the present invention will be described in detail with reference to the drawings. FIG. 1 shows an example of a color image processing system centering on an image processing apparatus 30 as a first embodiment of the present invention. In this image processing system, original color image data ORG output from an image input device 10 such as a scanner is input to the image processing device 30. The image data processed by the image processing device 30 is finally output to the image output device 20 such as a printer, where a final image is obtained. Since the description of the embodiment is diverse,
(1) Hardware of image processing apparatus (2) Outline of image processing-1
The first and second embodiments will now be described. After that, (3) Outline of Image Processing-Part 2
The third to sixth embodiments will be described as follows.

(1) Hardware of Image Processing Device The image processing device 30 performs color correction and tone number conversion as image processing that matches the input original color image data ORG with the color reproduction characteristics of the image output device 20. The color correction is a process for correcting the output characteristics of the image output apparatus such as gamma correction. The gradation number conversion is color correction when the number of gradations that can be output by the image output apparatus 20 is smaller than the number of gradations that can be output by the color image data ORG output from the image input apparatus 10. This is a process of converting the color image data that has been processed into a final number of gradations that matches the image output device 20. For example, the image data ORG read from a scanner or the like has 256 gradations (for 8 bits) for each color of R, G, and B, and the image output apparatus 20 performs expression by turning on / off ink. In addition, there can be a configuration in which the final number of gradations is two. In this case, the image processing apparatus 30 converts the image data of 256 gradations into 2 gradations, and outputs it to the image output apparatus 20 as final color image data FNL. In the above description, the tone number conversion is collectively called, but actually, the number of gradations is reduced by assigning the input original color image data ORG to a grid point having a small number of tone levels before color correction. Pre-gradation number conversion to be performed and gradation number conversion by so-called halftone processing that binarizes the data after color correction in accordance with the gradation number that can be expressed by the printer. In the following description, the former is called pre-gradation conversion and the latter is called post-gradation conversion. Each gradation number conversion will be described in detail later.

  FIG. 2 is a block diagram showing a specific configuration example of the color image processing system shown in FIG. Here, as the image input device 10, a scanner 12 that optically reads a color image from a document is used. The scanner 12 outputs the read color image data as original color image data ORG composed of three color components R, G, and B. In the embodiment, each color of R, G, and B is expressed by 8-bit digital data, and the number of gradations is 256. In this case, the scanner 12 expresses primary color image data with the three primary colors R, G, and B, and the color of each pixel is a three-dimensional color space having the R, G, and B colors as coordinate axes. The position of the pixel is expressed by a coordinate position value, that is, a coordinate value. However, even if any other color system such as L * a * b * is adopted, the color of a pixel is in its color space. It can be expressed as a coordinate value where it is located.

  In addition to the scanner 12, for example, a video camera, a host computer for creating computer graphics, and other means can be used as the image input device 10.

  In the image processing system of the embodiment, a color printer 22 that cannot perform gradation control in units of pixels is used as the image output device 20. The color printer 22 needs to perform binarization processing that reduces the number of gradations of each color component of the original color image data ORG output from the scanner 12 to two gradations corresponding to on / off of each pixel. Mentioned above.

  In addition to this, for example, a color display 21 or the like can also be used as the image output device 20. Many color displays 21 for computers, for example, have a smaller number of displayable gradations than ordinary home TVs. Even when such a color display 21 is used, it is necessary to convert the number of gradations of the original color image data ORG into the number of gradations corresponding to the display 21.

  Next, a specific configuration corresponding to the image processing apparatus 30 will be described. FIG. 2 shows a configuration using a computer 90 as the image processing apparatus 30. The computer 90 is provided with a CPU, RAM, ROM, etc. (not shown), and an application program 95 operates under a predetermined operating system. A video driver 91 and a printer driver 96 are incorporated in the operating system, and the final color image data FNL is output from the application program 95 via these drivers. An application program 95 that performs image retouching and the like reads an image from the scanner 12, performs predetermined processing on the image, and displays the image on the CRT display 93 via the video driver 91. When the application program 95 issues a print command, the printer driver 96 of the computer 90 receives image information from the application program 95 and receives a signal that can be printed by the printer 22 (here, a binarized signal for CMY). ). In the example shown in FIG. 2, the printer driver 96 includes a rasterizer 97 that converts color image data handled by the application program 95 into image data in dot units, and an image output device (for image data in dot units). Here, the color correction module 98 for performing color correction in accordance with the ink colors CMY and color development characteristics used by the printer 22), the color correction table CT referred to by the color correction module 98, and the image information after color correction, in dot units. There is provided a halftone module 99 for generating so-called halftone image information expressing the density in a certain area depending on the presence or absence of ink.

  The pre-gradation conversion described above is performed by the color correction module 98, and the post-gradation conversion is performed by the halftone module 99. The processing in the color correction module 98 corresponds to lattice point conversion processing.

  Next, the structure of the printer 22 as an output device will be briefly described. FIG. 3 is a schematic configuration diagram of the printer 22. As shown in the figure, the printer 22 includes a mechanism for transporting the paper P by the paper feed motor 23, a mechanism for reciprocating the carriage 31 in the axial direction of the platen 26 by the carriage motor 24, and a print head mounted on the carriage 31. And a control circuit 40 that controls the exchange of signals with the paper feed motor 23, the carriage motor 24, the print head 28, and the operation panel 32. Yes.

  The carriage 31 of the printer 22 can be mounted with a black ink cartridge 71 and a color ink cartridge 72 containing three colors of cyan, magenta, and yellow. A total of four ink ejection heads 61 to 64 are formed on the print head 28 below the carriage 31. An introduction pipe 65 (leading ink from the ink tank to the color heads is provided on the bottom of the carriage 31. (See FIG. 4). When the black ink cartridge 71 and the color ink cartridge 72 are mounted on the carriage 31 from above, an introduction tube is inserted into a connection hole provided in each cartridge, and ink from each ink cartridge to the ejection heads 61 to 64 is inserted. Supply becomes possible.

  A mechanism for ejecting ink will be briefly described. As shown in FIG. 4, when the ink cartridges 71 and 72 are mounted on the carriage 31, the ink in the ink cartridge is sucked out via the introduction tube 65 using the capillary phenomenon, and is provided at the lower portion of the carriage 31. The print heads 28 are guided to the respective color heads 61 to 64. When the ink cartridge is first installed, an operation of sucking ink to each of the color heads 61 to 64 is performed by a dedicated pump. In this embodiment, a pump for suction, a cap for covering the print head 28 at the time of suction, etc. The illustration and description of this configuration are omitted.

  As shown in FIG. 4, each of the color heads 61 to 64 is provided with 32 nozzles n for each color, and is one of the electrostrictive elements for each nozzle and has excellent responsiveness. PE is arranged. FIG. 5 shows the structure of the piezo element PE and the nozzle n in detail. As shown in the drawing, the piezo element PE is installed at a position in contact with the ink passage 80 that guides ink to the nozzle n. As is well known, the piezo element PE is an element that transforms electro-mechanical energy at a very high speed because the crystal structure is distorted by application of a voltage. In the present embodiment, by applying a voltage having a predetermined time width between the electrodes provided at both ends of the piezo element PE, the piezo element PE extends for the voltage application time as shown in the lower part of FIG. One side wall of 80 is deformed. As a result, the volume of the ink passage 80 contracts according to the expansion of the piezo element PE, and the ink corresponding to the contraction becomes particles Ip and is ejected from the tip of the nozzle n at high speed. Printing is performed by the ink particles Ip soaking into the paper P mounted on the platen 26.

  In the printer 22 having the hardware configuration described above, the carriage 31 is reciprocated by the carriage motor 24 while simultaneously rotating the platen 26 and other rollers by the paper feed motor 23 to convey the paper P, and at the same time, each color of the print head 28. The piezo elements PE of the heads 61 to 64 are driven to discharge each color ink and form a multicolor image on the paper P.

  The mechanism for transporting the paper P includes a gear train (not shown) that transmits the rotation of the paper feed motor 23 not only to the platen 26 but also to a paper transport roller (not shown). Further, the mechanism for reciprocating the carriage 31 has an endless drive belt 36 stretched between the carriage motor 24 and a slide shaft 34 that is mounted in parallel with the axis of the platen 26 and slidably holds the carriage 31. And a position detection sensor 39 for detecting the origin position of the carriage 31.

  The control circuit 40 includes a CPU (not shown). However, color correction according to the output characteristics of the printer 22 is processed in the computer 90, and the printer 22 converts the number of gradations. No processing related to color correction is performed. The processing executed inside the printer 22 is simply to receive the data output from the computer 90 and drive each color piezo element PE of the print head 28 in synchronization with the paper feed and the reciprocation of the carriage 31 described above. is there. Therefore, a detailed description of the control circuit 40 of the printer 22 and its processing are omitted.

(2) Outline of Image Processing Next, an outline of image processing will be described. The image processing employed in this embodiment includes a wide range of pre-gradation conversion, color correction, and host gradation conversion. There are also many variations of these processes, such as error diffusion method, average error minimum method, systematic dither method, software color correction, hardware color correction, etc. explain.
[A] Outline of image processing [B] Outline of image processing common to first to seventh embodiments [C] Details of image processing common to first to third embodiments [D] First to third embodiments Example [E] Details of image processing common to the fourth to seventh embodiments [F] Fourth to seventh embodiments

[A] Outline of Image Processing As shown in FIG. 6, the correction table CT referred to by the color correction module 98 in the computer 90 has a three-dimensional color space composed of three colors R, G, and B in a grid pattern. This is a divided color table. The table CT is read from, for example, a hard disk when the printer driver 96 is incorporated in the operating system, and is expanded in the RAM of the computer. For each grid point of this color table, for example, a color document for reading such as the scanner 12 and an output color image printed on a recording sheet using, for example, the color printer 22 have the same color. CMY color gradation correction data obtained by converting the gradation value data of R, G, and B is stored.

  The computer 90 performs correction processing on the original color image data ORG input from the scanner 12 using correction data stored in the correction table CT (pre-gradation number conversion), and further performs color correction on the color data. For example, as described above, the post-gradation number conversion process for converting to the final gradation number corresponding to the gradation number N of the image output device 20 such as the color printer 22 is performed. The final color image data FNL thus converted may be output to the color printer 22 as it is, and is stored in the memory of the computer 90 as final color image data FNL for one screen. Thereafter, the data may be output to the color printer 22.

[B] Overview of Image Processing Common to First to Eighth Embodiments FIG. 7 shows a functional block diagram of an image processing apparatus 30 realized by the color correction module 98 and the halftone module 99 of the computer 90. ing. In FIG. 7, the image processing apparatus 30 includes a pre-gradation number conversion unit 140, a color correction unit 142, a post-gradation number conversion unit 146, and a color correction table memory 134 that stores the color correction table CT described above. It is composed of First, correction data stored in the color correction table memory 134 will be described.

  As shown in FIG. 6, the input original color image data ORG is defined as coordinate values in a color space having coordinate axes corresponding to R, G, and B color components. Each coordinate axis sets the number of gradations of each color component as a coordinate value. Then, this color space is divided into grids, and correction data for each color component is stored for each grid point 300.

  There are various methods for determining correction data to be stored in the color correction table memory 134. The process of determining it and storing it in the table CT will be briefly described. Usually, first, various R, G, and B values are given to a target output system (for example, the color printer 22), and the actual output result is subjected to color measurement. Then, the correspondence relationship between the R, G, and B values given to the output system and the R, G, and B values obtained by measuring the output result is examined. The actual ink color in the printer 22 is CMY, but here it is handled in R, G, and B for the sake of simplicity. The conversion from R, G, B to C, M, Y will also be described in detail later.

  Next, the correspondence relationship is reversed, and the tone number correction data for each of the R, G, and B color components necessary for obtaining the corresponding color is checked for each grid point 300, and this is used as the correction data. Stored in the correction table memory 134.

  In this way, the color correction result corresponding to each grid point 300 is stored in the color correction table memory 134. In the color space, the image quality improves as the number of divisions increases. However, since the capacity of the color correction table memory 134 increases, an appropriate number of divisions is determined from the balance between the color correction table capacity and the image quality.

  The image processing apparatus 30 according to the embodiment uses the correction data stored in the color correction table memory 134 in this way, and performs image processing on the original color image data ORG as follows.

  First, the pre-gradation number conversion unit 140 uses a predetermined gradation number conversion method to set the number of gradations of the R, G, and B color components of the original color image data ORG to the optimum in the color space shown in FIG. Pre-gradation number conversion is performed to convert the gradation coordinate values of the grid points. The converted data is output to the color correction unit 142 as lattice point color image data GCD. Here, the number of pre-gradations is converted so that each color component of the original color image data ORG is Nr gradation, G is Ng gradation, and B is Nb gradation, and the result is expressed as Rk, Gk, Output as BK.

  Note that as the method of converting the number of gradations in the pre-gradation number conversion unit 140, various methods such as a multi-value error diffusion method, an average error minimum method, and a multi-value systematic dither method can be employed. Details thereof will be described later.

  The color correction unit 142 reads, from the color correction table memory 134, gradation correction data of lattice points corresponding to the input lattice point color image data GCD. Then, the number of gradations of the color components Rk, Gk, and Bk of the lattice point color image data GCD is corrected and output to the post gradation number conversion unit 146 as color correction data CCD. Here, the data of each color component of the color correction data CCD is represented as Rc, Gc, Bc.

  Then, the post gradation number conversion unit 146 converts the color correction data CCD obtained from the color correction unit 142 to the post gradation number to be finally converted by the error diffusion method or the average error minimum method. This is output as final color image data FNL in which each color data is represented by Rp, Gp, Bp.

  Therefore, by outputting the final color image data FNL to, for example, the color printer 22, the color printer 22 can print out a color image having good color reproducibility on the recording paper.

  Among the configurations described above, it has been conventionally known to divide a color space into a grid and prepare a color correction table memory 134 in which color correction data is recorded for each grid point (for example, Japanese Patent Laid-Open No. Hei 7). -30772). That is, the image processing apparatus 30 uses the pre-gradation number conversion unit 140 before referring to the color correction table memory 134, and the input original color image data ORG is a color on a grid point of the color correction table memory 134. The pre-gradation conversion is performed so that the data becomes data, color correction is performed with reference to the color correction table memory 134, and post-gradation conversion (for example, halftoning processing) is performed. As a result, a complicated interpolation calculation when referring to the color correction table CT can be omitted, and the calculation time required for the color correction process can be greatly shortened.

  The explanation will be further supplemented for each of the processes described above. First, the case where the error diffusion method or the average error minimum method is used for the pre-gradation number conversion processing in the pre-gradation number conversion unit 140 will be described as an example. Now, it is assumed that color regions having gradations of R = 12, G = 20, and B = 24 in the original color image data ORG input from the scanner 12 are continuous for a certain area. Assume that the color data of this color region is input to the image processing apparatus 30.

  FIG. 8 shows eight grid points near the coordinate position in the color space of the original color image data ORG. Each lattice point is located at each vertex of the cube shown in FIG. 8, and the coordinate position is expressed by the following equation (1).

  In the color correction table memory 134, color correction values of R, G, and B color components for the color data of each grid point 300 are prepared. In the example shown in FIG. 8, the original image data ORG is data represented by coordinate values of (R, G, B) = (12, 20, 24).

  When the color correction process is performed using the conventional technique for performing the interpolation calculation, the color data correction values at eight lattice points in the vicinity of the coordinate position in the color space of the original image data ORG are referred to. Then, using the eight color correction values referred to, an interpolation calculation process is performed to obtain a weighted average according to the distance between the coordinate position of the original color image data ORG and each grid point, thereby performing the color correction process. . In the example shown in FIG. 8, since the coordinate positions of the original color image data ORG and each grid point are set at equal distances, the interpolation calculation simply averages the color correction values of the eight grid points, and R, G , B color correction data is obtained.

  On the other hand, the color correction process of the present invention is performed in the following procedure. First, the pre-gradation number conversion unit 140 uses an error diffusion method or a minimum average error method so that the input original color image data ORG becomes any color data on eight grid points. Tone number conversion is performed. Next, the color correction unit 142 refers to the color correction table memory 134 and performs color correction processing.

  The error diffusion method and the average error minimum method used in the pre-gradation number conversion unit 140 according to the embodiment are arranged so that each piece of data is adjacent so that the average value of local colors in a certain region is as equal to the original image data as possible. It works to convert to the data value of the grid point. Therefore, in the example of FIG. 8 where the coordinate position of the original color image data ORG is equidistant with respect to each grid point, the result of the pre-gradation conversion is that each of the 8 grid point data is approximately 8 minutes. A color area mixed at an equal ratio of 1 is obtained. As a result, as a result of performing color correction with reference to the color correction table CT on the color data that has been subjected to the pre-gradation number conversion process, a color region in which the color correction values of the eight grid point colors are mixed with almost equal probability is obtained. Will be.

  As described above, the results obtained by the color correction processing performed by the conventional interpolation calculation and the color correction processing of the present embodiment are compared when the average value of the colors in an appropriate area is compared. Almost equal. That is, in this color correction unit 142, quantization noise generated by converting the number of pre-gradations in each pixel unit is added, but nevertheless, when taking a local average value, A color correction result almost equivalent to that obtained when the conventional interpolation calculation is performed can be obtained.

  When the number of gradations that can be expressed by the image output apparatus 20 is sufficiently large, there is a problem of image quality degradation due to quantization noise caused by the above-described pre-gradation number conversion process. In the case of the present embodiment, after the pre-gradation number conversion process, the post-gradation number conversion unit 146 is used to correspond the color-corrected image data to the number of gradations that can be expressed by the image output device 20. Since the post-gradation number conversion process for converting to the final gradation number is performed, the quantization noise associated with the gradation number conversion also occurs here. Therefore, if the quantization noise generated by the pre-gradation number conversion at the pre-processing stage can be sufficiently reduced as compared with the quantization noise generated by the post-gradation number conversion at the final stage, there will be no practical problem. . Therefore, in the apparatus of the embodiment, the number of gradations that can be converted by the pre-gradation number conversion unit 140 is formed to be sufficiently larger than the number of gradations that can be converted by the post-gradation number conversion unit 146. Therefore, the deterioration of the image quality accompanying the pre-gradation number conversion process is prevented.

  In particular, in this embodiment, since the error diffusion method or the average error minimum method is used for the tone number conversion in the post tone number conversion unit 146, the region is obtained through both the pre-tone number conversion and the post-tone number conversion. A mechanism for minimizing the error when the local average value of the value is taken will work. Therefore, the number of gradations that can be converted in the pre-gradation number conversion process is sufficiently larger than the number of gradations that can be converted in the post-gradation number conversion process as in the embodiment, thereby Compared with the conventional correction processing in which a grid point is referred to and an interpolation operation is performed, an image quality that is completely comparable can be obtained.

  Actually, when each color component of the original color image data ORG has 256 gradations, the number of gradations that can be converted by the pre-gradation number conversion unit 140 is 32 for each color component, and the conversion by the post-gradation number conversion unit 146. The number of possible gradations is set to 2 for each color component, and color correction and gradation number conversion are performed. As a result, after interpolation correction is performed and color correction is performed as in the conventional case, the result is completely distinguished from the result of two gradations. It was confirmed that an excellent reproduction image that cannot be obtained can be obtained. Further, it has been confirmed that even if the number of gradations that can be converted in the pre-gradation number conversion process is reduced to about 8 gradations, an image quality having no practical problem can be obtained.

  As described above, the memory capacity of the color correction table memory 134 can be reduced as the number of converted gradations in the pre-gradation number conversion process is reduced. For example, if the output data after color correction in the color correction unit 142 is m color components and n-bit data for each color component, the memory capacity required for the color correction table memory 134 is pre-gradation. When the number of gradations that can be converted in the number conversion process is each color L gradation (q bits), it is expressed by the following equation.

Assuming that the number m of color components is 3, 8 bits (1 byte) for each color component, and the number of gradations is 32 gradations (5 bits) for each color.
Q = 3 × 8 × 2 ^ (5 × 3)
= 98304 × 8 [bits]
= 96K [bytes]
Note that 1K [bytes] is 1024 [bytes].

  Similarly, when the number of gradations that can be converted to the number of pre-gradations is 16 gradations for each color, the memory capacity of the color correction table is calculated for each of the 8 gradations. In the case of bytes and 8 gradations, a memory capacity as small as 1.5 Kbytes is sufficient. On the other hand, when the original image data is 8 bits (256 gradations) for each color, if an attempt is made to prepare a color correction table for all the original color image data ORG that can be expressed, the data is about 16 Mbytes. Will be needed.

  In the above description, the case where the error diffusion method or the average error minimum method is used as the gradation number conversion method in the pre-gradation number conversion unit 140 has been described as an example. In addition to this, the number of gradations can be converted using other methods such as a systematic dither method.

  In this case, in the pre-gradation number conversion unit 140, the mechanism for minimizing the error when the local average value of the region is taken does not work, so that the image quality is higher than when the error diffusion method is used. There is a possibility of deterioration. This may cause a problem in the case of performing a large gradation number conversion that reduces the number of gradations of the original image data from 256 to 8 gradations. However, when the difference between the number of gradations used for representing the original image data and the number of gradations that can be converted by the pre-gradation number conversion unit 140 is not so large, for example, the number of gradations of the original color image data ORG. Is reduced to 32 or 16 gradations by pre-gradation number conversion processing, or when the spatial frequency of quantization noise is sufficiently high with sufficiently high resolution for the output device Therefore, the influence of the quantization noise generated here was not so much a problem when actually outputting various images. In such a case, a simpler tone number conversion method such as a systematic dither method may be employed instead of the error diffusion method. When a simple gradation number conversion method such as systematic dithering is employed, the calculation time required for gradation number conversion can be shortened.

[C] Details of Image Processing Common to First to Third Examples Next, a more specific example of the image processing apparatus as the above embodiment will be described. Here, the original color image data ORG having 8 bits for each of R, G, and B and 256 gradations is input to the computer 90 as the image processing apparatus 30 from the scanner 12 shown in FIG. Assume that the original color image data ORG is processed and the final color image data FNL is output to the color printer 22. The color printer 22 used here is one that prints with two gradations of ON (dotted) / OFF (no dot) of each color dot using three color inks of cyan C, magenta M, and yellow Y. . The specific configuration of printing in the printer 22 has already been described.

  FIG. 9 shows a specific block diagram of the image processing apparatus 30 used in this case. The pre-gradation number conversion unit 140 and the color correction unit 142 perform color correction on the input original color image data ORG, further convert the color system from R, G, B to C, M, Y, and color correction data Output as CCD. Then, the color correction data CCD is binarized corresponding to the number of gradations that can be displayed by the printer 22 using the post gradation number conversion unit 146, and is output as final color image data FNL.

  Here, correction data is set in the color correction table memory 134 as follows. In order to determine correction data, first, only the post-gradation number conversion unit 146 (halftone module 99 shown in FIG. 2) is extracted from the image processing apparatus shown in FIG. 9, and the target color printer 22 is combined. Configure the system. Then, various C, M, and Y values are given to the post-gradation number conversion unit 146 to be binarized, and then the result output to the target color printer 22 is subjected to color measurement. Then, the correspondence relationship between the C, M, and Y values given to the post tone number conversion unit 146 and the R, G, and B values obtained by measuring the output result of the color printer 22 is examined.

  Next, the correspondence relationship is reversed, and C, M, and Y values necessary for obtaining colors of R, G, and B values corresponding to the grid point color data in the color space are obtained, and the color correction is performed. Data is set in the color correction table memory 134 as data. Prior to actual image processing, the above processing is performed to prepare the color correction table CT.

  The pre-gradation number conversion unit 140 converts the color components of R0, G0, and B0 of the input original color image data ORG into 16 gradations and B is 8 gradations. Each color component of Pk, Gk, Bk is output as grid point color image data GCD. This is an example when Nr and Ng are 16 and Nb is 8 in the embodiment shown in FIG. For this purpose, in the present embodiment, multilevel conversion is performed by an error diffusion method or an average error minimum method, but this multilevel conversion process itself may use an existing method.

  Next, a specific example of multi-value processing will be described by taking a case where the B component is converted to 8 gradations as an example. Now, it is assumed that the B component of the original color image data ORG is expressed by binary 8 bits, and the number of gradations is 256 gradations from 0 to 255. Assume that the following eight types (i = 0, 1,...) Of lattice point color data pre_B are converted into eight values.

pre_B [0], pre_B [1], ..., pre_B [7]
Specifically, the B component of the original color image data ORG is converted to the number of pre-gradations as follows. In this example, the eight values are substantially equally spaced, but in the first and second embodiments to be described later, the lower concentration region is set more finely.

pre_B [0] = 0;
pre_B [1] = 36;
pre_B [2] = 73;
pre_B [3] = 109;
pre_B [4] = 146;
pre_B [5] = 182;
pre_B [6] = 219;
pre_B [7] = 255;
In the following description, lattice point color data may be abbreviated as pre [i].

Also, seven types of threshold values used for pre-gradation processing are defined as follows.
slsh_B [0], slsh_B [1],..., slsh_B [6]

Each threshold is set as follows.
pre_B [i] <slsh_B [i] <pre_B [i + 1]
(I = 0, 1, 2, ..., 6)

The threshold value is often set as follows.
slsh_B [i] = (pre_B [i] + pre_B [i + 1]) / 2
In this case, the threshold values are as follows.

slsh_B [0] = 18;
slsh_B [1] = 54;
slsh_B [2] = 91;
slsh_B [3] = 127;
slsh_B [4] = 164;
slsh_B [5] = 200;
slsh_B [6] = 236;

  Further, the original color image data ORG input as image data to the image processing apparatus 30 is normally input in order from the left end pixel to the right end pixel, starting from the pixel at the upper left corner of the image. Then, after the pixels for one row are input, the process moves to the left end of the row one pixel below, and the subsequent data is similarly input toward the right end. By repeating such an image input operation, image data for one screen is input.

  For this reason, the pre-gradation processing by the pre-gradation number conversion unit 140, that is, the order in which the original color image data ORG is octalized is performed in accordance with the input order of such image data. That is, octarization is performed in order from the left end pixel to the right end pixel starting from the pixel in the upper left corner of the image, and when the octarization of pixels for one row is completed, the process moves to the left end of the row below one pixel, and similarly The operation of octalizing toward the right end is repeated, and the octal of image data for one screen is performed.

  In this case, as shown in FIG. 10, all the pixels above the target pixel 400 and the pixels on the left side of the line are pixels that have already been subjected to multi-value quantization. Then, all the pixels below the target pixel 400 and the pixels on the right side of the same line are pixels that have not yet been subjected to multi-value conversion.

  As a technique for performing the above-described pre-gradation processing, that is, 8-level processing, consider a case where, for example, an error diffusion weight matrix shown in FIG. This weight matrix indicates that the error occurring in the pixel of interest 400 in FIG. 11A is distributed in a ratio of 2 to the right adjacent pixel, 1 to the lower pixel, and 1 to the lower right pixel. is there.

  FIG. 12 shows a flowchart of 8-level processing (pre-gradation conversion processing) performed by the pre-gradation number conversion unit 140 using the error diffusion method using this matrix.

Assume that the pixel of interest is expressed as p rows and q columns, the original color image data ORG of the B component of the pixel of interest is represented by dataB [p] [q], and this is converted into eight values. Note that the quantization error caused by the 8-level quantization of the pixels in the p-th row and the q-th column is represented as err [p] [q].
(1) 1st process; Octarization process (step S1-S6)
First, the pixel-of-interest data is converted into eight values by comparing with a threshold value to obtain grid point color data pre_B [p] [q]. Hereinafter, this is abbreviated as lattice point color data pre_B. This process will be briefly described. If the original color image data ORG is less than the threshold value slsh [0] (step S1), the lattice point color data pre_B of the target pixel is set to the smallest lattice point color data pre [0] (step S4). Similarly, it is determined which threshold value (slsh [i] to slsh [i + 1]) the original color image data dataB [p] [q] is in (step S2), and the lattice point of the target pixel The color data pre_B is processed as the corresponding grid point color data pre [i] (step S5). If the original color image data ORG is equal to or greater than the threshold value slsh [7] (step S3), the lattice point color data pre_B of the target pixel is set to the largest lattice point color data pre [7] (step S6). . Note that dataB [p] [q] shown in FIG. 12 is not the pixel-of-interest data itself but data that has already been corrected in response to error diffusion of pixel colors that have already been converted into eight values. The error diffusion method will be described in the third step.

(2) Second step; error calculation step (step S7)
An error err [p] [q] that has occurred in the pixel of interest due to the 8-value conversion is obtained. The original color image data dataB [p] [q] does not originally exist on the lattice point, but is assigned to the value of the lattice point by the above octalization process, so that a quantization error occurs. The magnitude of this quantization error err [p] [q] is obtained.

(3) Third step; error diffusion step (step S8)
The error is diffused to nearby pixels that have not yet been 8-valued. Here, according to the weight matrix of FIG. 11A, 1/2 of the error is the pixel on the right side of the pixel of interest ([p + 1] [q]), and 1/4 is the lower pixel ([p] [q + 1] ) Is diffused to adjacent pixels ([p + 1] [q + 1]) on the lower right and added to the original color image data data_B. The pixel-of-interest data dataB [p] [q] used in the first step described above is data after being subjected to error diffusion in this way.

  The above is an example of octalization by the error diffusion method. A specific example in which this is performed by the average error minimum method will be described below.

FIG. 13 is a flowchart of the octalization process using the average error minimum method.
(1) First step; error correction step (step S10)
In this step, the pixel-of-interest data is corrected with an eight-valued error that has occurred in pixels that have already been eight-valued in the vicinity. The error diffusion matrix is the same as that shown in FIG. That is, when viewed from the target pixel, 1/2 is added to the quantization error generated in the pixel on the left, and 1/4 is added to the quantization error generated in the pixel immediately above. ¼ is added to the quantization error generated in step.

(2) 2nd process; 8-value process (steps S11-S16)
Since this process is the same as steps S1 to S6 shown in FIG. 12, the description thereof is omitted.

(3) Third step; error calculation (step S17)
In this step, an error err [p] [q] generated in the pixel of interest due to octarization is obtained. The details are exactly the same as step S7 shown in FIG.

  The only difference between the error diffusion method shown in FIG. 12 and the average error minimum method shown in FIG. 13 is whether the error diffusion operation is performed immediately after the error calculation is completed or just before the target pixel is converted to N-values. Except for handling at the edge of the image, they are equivalent.

  As for the weight matrix of the error diffusion method, in addition to FIG. 11 (a), the example of FIG. 11 (b) with a larger matrix size, or the inverse of FIGS. 11 (c) and 11 (d). It is possible to adopt various types as necessary, such as examples. FIG. 11D shows the most simplified example, and the error diffusion target is only one pixel on the right.

  Examples of the multi-level error diffusion method include JP-A-3-18177, JP-A-3-34777, JP-A-3-80767, JP-A-3-147480, and the like. The method can be adopted.

[Embodiment of Pre-gradation Number Conversion Unit 140 Using Systematic Dither Method]
The pre-gradation number conversion unit 140 can employ a gradation number conversion method such as a systematic dither method in addition to a configuration using the error diffusion method or the average error minimum method. An example of processing when the systematic dither method is used as the pre-gradation number conversion unit 140 is shown in FIG.

  Here, when the B component of the original color image data ORG has a value of 64 gradations from 0 to 63, this is converted to 17 gradations by the pre-gradation number conversion unit 140 using the systematic dither method. Is assumed. A dither matrix having a size of 2 × 2 is used. After adding systematic dither noise that changes in a two-pixel cycle in both the vertical and horizontal directions to the data (step S20), by comparison with threshold values (steps S21 to S23), 17 gradations of 0, 1, 2,... The number of gradations is converted into a value (S24 to S26).

That is, the 16 threshold values slsh_B [i]
slsh_B [i] = (i + 1) × 4-2
(I = 0, 1, ..., 15)
Next, systematic dither noise (dither_noize [p% 2] [q% 2]) uniquely determined by the pixel positions p and q is added to the target pixel data data_B [p] [q] (step S20). ). % Is a surplus operator, and p% 2 means “remainder when p is divided by 2”. If q is an even number, it is 0; The value of dither_noize [p% 2] [q% 2] is, for example,
dither_noize [0] [0] = 1
dither_noize [0] [1] = − 1
dither_noize [1] [0] =-2
dither_noize [1] [1] = 0
Set as follows. Thereafter, the data is compared with the threshold value slsh_B [i] (steps S21 to S23), and the blue component preB of the lattice point color data is obtained according to the magnitude, and is converted into 17 gradations (S24 to S26).

  Compared with the example of 8 gradations using the average error minimum method shown in FIG. 13, the error average diffused from the surrounding binarized pixels is added to the target pixel data data_B [p] [q]. However, when the systematic dither method is adopted, a periodic noise dither_noise [p% 2] [q% 2] that is uniquely determined by the position of the target pixel is added. Accordingly, the method shown in FIG. 14 eliminates the process corresponding to the error calculation step (step S17) performed by the average error minimum method. When the systematic dither method as shown in FIG. 14 is used as compared with the minimum average error method or the error diffusion method, the error diffusion calculation is not required, so that the speed is further increased and the error memory is stored. Therefore, there is a great merit that a memory or the like is unnecessary and hardware resources are saved.

  On the other hand, the mechanism for minimizing the error when the “local average value of the region” in the error diffusion method is taken does not work. For this reason, if the number of gradations is reduced too much by the pre-gradation number conversion means, the image quality may be deteriorated. However, as shown in FIG. 14, the number of gradations of the original image data is reduced to about 1/4. In the case of a reduction to such a level, pseudo contours and the like are not generated even with a matrix size as small as 2 × 2, and a reduction in resolution is minimized, which is very advantageous.

In general, an optimal pre-gradation number conversion method may be employed in consideration of processing speed, necessary memory capacity, and image quality. For example, the pre-gradation number conversion method to be employed can be determined from the following viewpoints.
・ If the resolution of the final output device is sufficiently high and there is no problem with a slight decrease in resolution due to pre-gradation conversion, a systematic dither method with a larger matrix size. If there are not many and there is no need to reduce the number of gradations significantly by pre-gradation conversion, a smaller matrix size systematic dither method / priority of image quality, or by pre-gradation conversion If it is desired to reduce the number of gradations to reduce the capacity of the color correction table, an error diffusion method or a gradation number conversion method other than the error diffusion method or the systematic dither method may be used.

  Alternatively, a systematic dither method for only the B component, and a configuration in which pre-gradation number conversion is performed by the error diffusion method for the R and G components may be employed. In general, the resolution of the human eye with respect to the B component is lower than that of R and G, and such a configuration is also effective.

  Although the case where the B component included in the original color image data ORG is converted into 8 gradations using the pre-gradation number conversion unit 140 has been described above, other color components included in the original color image data ORG are described. That is, the R component and the G component are also converted to the pre-gradation number of 16 gradations by a similar method.

As a result, the R, G, and B components of the original color image data ORG input to the pre-gradation number conversion unit 140 are pre-gradation converted to lattice point color image data GCD as shown in the following equation. It will be.
R component is pre_R [0], pre_R [1] ... pre_R [15] 16-value G component is pre_G [0], pre_G [1] ... pre_G [15] 16-value B component is pre_B [0] , Pre_B [1]... Pre_G [7].

[Specific Example of Color Correction Unit 142]
The color correction unit 142 performs color correction processing on multi-valued (pre-gradation processing) grid point color data, and also converts the color system from R, G, B to C, M, Y. To do. In other words, in the example shown in FIG. 7, the color correction unit 142 only performs color correction between RGB for convenience of explanation, but in reality, the color correction unit 142 performs other than color correction. The color system conversion from R, G, B to C, M, Y is also performed at the same time.

  In the following, the case where the color correction unit 142 and the color correction table memory 134 are formed as software and the case where they are formed as hardware will be described as examples.

  FIG. 15A shows an embodiment in which the color correction table memory 134 is realized in software using C language notation. The color correction tables for the C, M, and Y color components are three-dimensional arrays C_table [Nr] [Ng] [Nb], M_table [Nr] [Ng] [Nb], Y_table [Nr] [Ng] [Nb], respectively. ]. Here, Nr, Ng, and Nb indicate the number of gradations of each color (see FIG. 7). In this example, the size of each table is 16 × 16 × 8. Since these tables are unsigned char type arrays that handle only positive integers in 8 bits, 8 bits of data in the range of 0 to 255 can be stored as color correction results. In advance, conversion data from RGB to CMY is stored in this three-dimensional array.

  FIG. 15B refers to the three-dimensional color correction table of FIG. 14A, and R, G, B grid point color data pre_R [i], pre_G [j], pre_B [k] This is an embodiment of the color correction unit 142 that converts the CMY value corresponding to the ink amount in the printer 22. In this method using the C language, the C, M, and Y values after color correction can be obtained simply by referring to the array declared in FIG. 15A based on the grid point color data.

  Next, as an example when the color correction table memory 134 is realized by hardware, an example in which the color correction table CT is stored in a semiconductor memory is shown in FIG. The C ROM 134C, the M ROM 134M, and the Y ROM 134Y are ROMs in which the color correction results of the C, M, and Y color components are stored, respectively, and given values determined according to RGB grid point color data as address data. For example, corrected cyan data, magenta data, and yellow data are output.

  FIG. 17 shows a more detailed embodiment of the C ROM 134C of FIG. 15, which uses an 11-bit ROM with an address bus A0 to A10 and an 8-bit ROM with a data bus D0 to D7. Corresponding to the grid point color data pre_R [i], pre_G [j], and pre_B [k], the values obtained by binarizing the i, j, and k values are the lower 4 bits (A0 to A3) of the address bus. , Middle 4 bits (A4 to A7) and upper 3 bits (A8 to A10). The number of bits corresponds to the number of gradations of RGB (R and G are 16 gradations and B is 8 gradations).

  That is, in this embodiment, 0 ≦ i ≦ 15, 0 ≦ j ≦ 15, and 0 ≦ k ≦ 7, so that 4 bits are allocated for i, 4 bits for j, and 3 bits for k. It's enough. From the data bus, the corresponding cyan color correction data is output to the data bus as an 8-bit value having a value between 0 and 255.

  In the configuration shown in FIG. 16, three separate ROMs are prepared for each CMY component. However, if the number of bits of the address bus is increased and a color selection signal is added thereto, one ROM having a larger capacity is provided. It is also possible to deceive. Further, when a writable RAM is used instead of the ROM, the contents of the table can be freely rewritten.

  In the configurations shown in FIGS. 15 and 16, the color correction unit 142 performs color correction by referring to the three-dimensional table memory 134 in software or hardware.

[Specific example of post-gradation conversion]
After the color correction, the post-gradation number conversion unit 146 (halftone module 99) finally binarizes each CMY data color-corrected by the color correction unit 142 by the error diffusion method or the average error minimum method. Do. For this part, the existing method may be applied as it is. The binarization process by the error diffusion method is almost the same as the multi-value conversion process in the pre-gradation number conversion unit 140 shown in FIG. 12, and only the 8-value conversion is changed to the binarization. The octalization in steps S1 to S6 in FIG. 9 is changed to binarization, data_B [p] [q] is replaced with data_C [p] [q], and binarization is performed as shown in FIG. . Here, when cyan data is binarized to a value of 255, cyan dots are present, and when binarized to a value of 0, no dots are present. Since the subsequent error calculation and error diffusion steps are the same as steps S7 and S8 in FIG. 12, the description thereof is omitted.

  However, if the size of the error diffusion weight matrix is made too small in this binarization process, a unique dot pattern that leads to image quality deterioration tends to occur. For this reason, it is better to use a slightly larger matrix (for example, the one shown in FIG. 11B) than that used for multi-value conversion in pre-gradation conversion. Further, instead of the error diffusion method, a method in which multi-value conversion by the average error minimum method of the embodiment of FIG. 13 is changed to binarization may be used.

  One of the major features of the configuration described above is that the size of the error diffusion weight matrix used in the multi-value conversion in the pre-gradation number conversion unit 140 is very small as shown in FIGS. It is in the place where sufficient image quality can be obtained even with. Normally, when binarization processing is performed, as shown in FIGS. 11C and 11D, when error diffusion is performed with a small matrix such that the error diffusion target is two or less adjacent pixels, the dots become linear. Patterns peculiar to error diffusion that appear in succession become conspicuous and cause deterioration in image quality. However, when the number of gradations to be converted is 8 or more as in the pre-gradation number conversion unit 140 of the present embodiment, the small error diffusion matrix shown in FIGS. 11C and 11D is used. Even if it is used, the image quality does not deteriorate so much. Therefore, the pre-gradation number conversion unit 140 can use a matrix having a smaller size than the error diffusion matrix used in the post-gradation number conversion unit 146 in the subsequent stage. The error diffusion process such as step S8 in FIG. 12 occupies most of the calculation amount in the error diffusion processing, and the calculation amount is substantially proportional to the size of the error diffusion weight matrix. For this reason, the multi-value processing in the pre-gradation number conversion unit 140 can be performed with an extremely small amount of calculation. For example, compared with the binarization processing in the post tone number conversion unit 146, the amount of calculation in the pre tone number conversion unit 140 is extremely small. Furthermore, since the color correction in the color correction unit 142 having the above configuration is merely referring to the contents of the color correction table memory 134, the total data processing in the pre-gradation number conversion unit 140 and the color correction unit 142 is as follows. Can be done very fast.

  In the above configuration, the color correction unit 142 performs conversion from RGB to CMY simultaneously with color correction using the color correction table memory 134 shown in FIGS. 15 and 16. For printers that also use black ink K, conversion to CMYK four-color components may be performed. For example, when the color correction table realized by the software shown in FIG. 15A is expanded to a four-color table of CMYK, a table for four colors of CMYK may be prepared as shown in FIG. . In this way, if the number of prepared color correction tables is increased, it is possible to easily cope with an increase in the number of necessary color components due to color correction.

  In the above description, the case where the original color image data ORG is composed of RGB three-color components has been described. However, the original color image data ORG can be any of CMY, CIE L * a * b *, XYZ, etc. The color system may be used, or the color image data may be converted into another color system expression by the color correction unit 142 as shown in FIG.

[D] First to Third Embodiments Each embodiment of the present invention will be described on the premise of the configuration of the image processing apparatus 30 described in detail above.
(1) First Example In the above description regarding the image processing apparatus 30, in order to simplify the description, the color space is divided at equal intervals as shown in FIG. It is assumed that a color correction table having data is prepared. In the first embodiment, the color space is finely divided in the low density region of the ink CMY used by the printer 22 which is the final output device. As already described, if the number of gradations after the conversion of the pre-gradation number conversion is sufficiently larger than the final gradation number of the post-gradation number conversion, the quantization noise of the pre-gradation number conversion is sufficiently small. If the error diffusion or the average error minimization method is employed, the local average value becomes 0, and the practical effect of the quantization noise on the entire output image is small. However, since the quantization noise of the pre-gradation number conversion exists even if it is very small, this appears as a disturbance of the appearance position of the on-dot in the output image. Therefore, strictly speaking, the influence of this quantization noise is not substantially a problem in a high density or medium density area where the on-dot density is large, but it is a cause of image quality deterioration in a low density area where the dot density is small. It can be said that there is a possibility. Therefore, in order to solve this problem, in this embodiment, as shown in FIG. 20, a color correction table CT having lattice points 500 whose intervals are narrowed in a low density region is used. Correspondingly, in the pre-gradation number conversion, the original image data is converted into the gradation value (grid point color data) of the lattice point 500.

As a specific example, when the original image data takes gradation values from 0 to 255, this is converted into pre-gradation value conversion for each color of R, G, and B.
0, 16, 32, 48, 80, 96, 112, 128, 144, 160,
176, 192, 208, 224, 240, 248, 255
Are quantized into 18 gradation levels. In this example, the quantization step, that is, the interval between adjacent lattice points is basically 16, but in the vicinity of the gradation value 255, that is, in the low density region, the quantization step is as shown by the value 8 or the value 7. It is getting smaller. In the first embodiment, pre-gradation number conversion was performed by using the lattice points by the error diffusion method shown in FIG.

  As described above, the feature of the present embodiment is that the quantization step is reduced in the low concentration region. In the conventional interpolation calculation method, when the printer 22 such as an ink jet is used as an output device, it is appropriate to conversely increase the lattice point spacing in the low density region and narrow the lattice point spacing in the medium and high density region. . This is because, in the low density region, since the probability that dots are in contact with each other or overlap each other is low, line formation between input and output data is high, and sufficient accuracy is obtained even with a primary interpolation operation. This is because the linearity is greatly disturbed in the medium-high density region where the dot density is high. In this embodiment, contrary to the conventional common sense, the color correction table CT in which the lattice point interval in the low density area is narrowed is used to set the change amount of the output data to be small in the low density area. Yes. As a result, since the quantization noise of the pre-gradation number conversion in the low density region is reduced, image quality deterioration due to this quantization noise in the low density region is reduced, and the image quality of the output image is further improved. In addition, since the lattice point spacing is reduced only in the low density region, the capacity of the color correction table memory 134 does not increase so much.

  By the way, the data after the pre-gradation number conversion is then color-corrected according to the color correction table CT. In general, since the conversion characteristics of the color correction table CT are not linear, the quantization step of pre-gradation number conversion takes a different form after color correction. Depending on the conversion characteristics of the color correction table CT, even if the quantization step of the pre-gradation number conversion is sufficiently small in the low density region, the quantization step of the data after color correction will be in the low density region for each ink color. It is possible that the quantization step is not reduced. In this case, as shown in FIG. 20, even though the color correction table CT is divided finely in the low density regions on each of the RGB axes, the final image is obtained in the low density region. The influence of quantization noise appears.

Therefore, not only is the pre-gradation number conversion quantization step sufficiently small in the low density area, but also the pre-gradation so that the quantization step of the data after color correction is sufficiently small in the low density area. It is desirable to modify the quantization step of the number transform. This correction can be performed by the following steps, for example, as shown in FIG.
(1) First, the quantization step of the pre-gradation number conversion is provisionally determined so as to be sufficiently small in the low density region as described above (step S31), and the color correction table CT is created in accordance with this step. Actually, color correction is tried (step S32).
(2) The quantization step of the color-corrected data obtained as a result of this trial is checked (step S33). If the quantization step is greater than or equal to a predetermined upper limit value for eliminating the influence of the quantization error in the low density region at any location in the low density region (step S34), In the quantization step of pre-gradation number conversion, the portion related to the large portion is set smaller again (step S35), and the lattice point spacing of the color correction table is also corrected accordingly (step S36).

  In this way, finally, the quantization step for pre-gradation conversion and the corresponding grid point spacing of the color correction table CT so that the quantization step after color correction becomes smaller than the upper limit value in the low density region. To correct. By performing the pre-gradation conversion and color correction using the thus corrected quantization step and color correction table CT, the influence of the quantization noise of the pre-gradation conversion in the low density region is reliably reduced. .

(2) Second Embodiment A second embodiment of the present invention will be described in which the pre-gradation number conversion unit 140 performs the gradation number conversion by the dither method in the same configuration as the first embodiment. In this embodiment, the pre-gradation number conversion is performed by the process shown in FIG. In this embodiment, a 4 × 4 dither matrix is used. An example of the dither matrix is shown in FIG. In this example, the original color image data Da (0 to 255) having N gradations (256 gradations in this embodiment) is converted into M gradations (6 gradations in this embodiment) for each RGB color. It has been converted. The grid point color image data obtained by the pre-gradation number conversion is represented by GC. In the following description, DiTh indicates a dither matrix number, and takes a value from 1 to 16, as shown in FIG. Also, the grid points prepared for pre-gradation number conversion are narrowly divided in the low density region, and RSLT [0] = 0, RSLT [1] = 70, RSLT [2] = 130, RSLT [3]. = 190, RSLT [4] = 235, and RSLT [5] = 255. In addition, the distance Dist [i] between the lattice points is set to Dist [i] = RSLT [i + 1] −RSLT [i].
i = 0, 1,... 4
It is defined as It should be noted that Dist [5] = 1 is defined in order to guarantee the result of the calculation for obtaining the distance offst described later when the original color image data Da takes the value 255.

  When the pre-gradation conversion shown in FIG. 22 is started, first, the process of inputting the original color image data Da of the pixel of interest is performed (step S40). Thereafter, a process of obtaining a value DDH obtained by normalizing the dither matrix number DiTh corresponding to this pixel in the range of values 0 to 1 is performed (step S41). The value DDH is obtained by the following equation (3).

  The dither matrix number DiTh is obtained by obtaining the position value of [p% 4, q% 4] from the matrix shown in FIG. 23, where p is the position in the scanning direction of the pixel of interest and q is the position in the sub-scanning direction. Done. Here,% is a remainder operator. In order to obtain the dither matrix number DiTh from the matrix of FIG. 23, a function having elements [0, 0] to [3, 3] (for example, GetMatrix [x, y]) may be defined in advance.

  After obtaining the value DDH obtained by normalizing the dither matrix number DiTh in this way, a value 0 is set to the variable X indicating the converted gradation number number (step S42), and then the lattice point determined by the gradation number number X is set. A process of comparing the value RSLT [X] with the original color image data Da is performed (step S43). In this description, the original color image data Da to be compared does not particularly specify a color component, but actually, a comparison is made for each color component. If the original color image data Da is not less than or equal to the grid point value RSLT [X], the variable X is incremented by 1 (step S44), and the two are compared again. That is, the value corresponding to the lattice point is sequentially increased until the original color image data Da becomes equal to or less than the lattice point value RSLT [X].

  As a result, the original color image data Da will eventually become less than or equal to the lattice point value RSLT [X] (step S43), and then a process of calculating the offset offst between the original color image data Da and the lattice point compared in step S43. Is performed (step S45). The distance offst is calculated by the following equation (4) as a value normalized with respect to the distance Dist [X−1] between the grid points between which the original color image data Da is sandwiched.

  Then, the process of comparing this distance offst with the value DDH obtained by normalizing the dither matrix number DiTh is performed (step S46). If the difference between the two is larger, the original color image data Da is assigned to the grid point having the larger value of the grid points sandwiching the original color image data Da. X] is set to the result value RSL (step S47), and if the distance offst is smaller, the variable X A process of setting the value RSLT [X−1] of the grid point corresponding to −1 to the result value RSL is performed (step S48). Thereafter, a process of moving the target pixel to the next pixel is performed (step S49), and the above-described process is repeated until the end of the original color image data.

  According to the above processing, the number of gradations of the original color image data Da can be converted from 256 gradations to 6 gradations, and a lattice that is moderately dispersed using the distributed dither matrix (FIG. 23). It can be converted into point color image data. In the example using FIG. 14, such variation is created by adding data by dither matrix as noise to the original color image data. However, in this embodiment, the original color image data Da is assigned to either of the grid points sandwiching the data. In determining whether or not, variation using a dither matrix is generated. In other words, in this embodiment, the value DDH normalized using the dither matrix number DiTh prepared as the dither matrix is used to determine the distance from the adjacent grid points. Even if the original color image data Da has the same value, it may be assigned to different grid points. Compared to the average error minimum method and error diffusion method when using the systematic dither method, error diffusion calculation is not required, so that the time required for image processing can be further shortened and error storage is also possible. There is a great merit that the memory and the like are unnecessary and hardware resources are saved.

(3) Third Embodiment As shown in FIG. 24, the image processing apparatus 30 as the third embodiment includes main data in the pre-gradation number conversion unit 140 (see FIG. 9) described in the first embodiment. It has an output unit 140a and a sub data output unit 140b, and further has an interpolation calculation unit 148. That is, the image processing apparatus 30 according to the second embodiment performs color correction based on the color correction table CT from the pre-gradation conversion in addition to the configuration of the first embodiment in order to reduce quantization noise in the pre-gradation conversion. In this process, an interpolation calculation is performed.

  The conventional method of interpolation calculation performed in a three-dimensional space of R, G, B is an eight-point interpolation method using coordinate values of eight vertices of a rectangular parallelepiped, and the number of vertices to be taken into consideration in order to simplify the calculation is 6 A point interpolation method and a four-point interpolation method (for example, Japanese Examined Patent Publication No. 58-16180) are known, and these are all three-dimensional interpolation methods. In contrast, this embodiment uses a one-dimensional interpolation method. That is, in this embodiment, only the input color component having the greatest influence on the output color to be obtained (hereinafter referred to as the main color) among the three colors of R, G, and B is referred to Interpolation is performed using the quantized value. For example, when converting RGB input data to CMY output data, the main color of C is R, the main color of M is G, the main color of Y is B, and the respective complementary color components are the main colors. Interpolation is performed using front and rear quantized values only for color components. This is because the main color component has a great influence on the output result of each output color component, and the output result does not change much even if the values of color components other than the main color change slightly. Is used to obtain sufficient accuracy by interpolation only for the main color components.

  This will be specifically described. Here, the original image data is represented by (R, G, B), and the data after the pre-gradation conversion is represented by (R ′, G ′, B ′). Further, the C, M, and Y color components of the data obtained by applying the color correction table CT to the data (R ′, G ′, B ′) are converted into LUT_C (R ′, G ′, B ′), LUT_M (R ', G', B ') and LUT_Y (R', G ', B'). The final data after color correction conversion is represented by (C ″, M ″, Y ″).

Now, as shown in FIG. 25, the original image data (R, G, B) of a certain pixel is shown in FIG.
R'1 <R <R'2, G'1 <G <G'2, B'1 <B <B'2
The data (R′1, G′2, B′1) is obtained as a result of pre-gradation conversion of the original data (R, G, B) (hereinafter, this data is referred to as main data). Called data).

In this case, in the pre-gradation conversion, in addition to the main data (R′1, G′2, B′1), only the main data and the main color component are on the grid points on the opposite side across the original data. The data at is output as sub-data. The main data is common to each color component, but the sub data is different for each color component. Next, a color correction table is applied to both main data and sub data for each color component, and two values necessary for one-dimensional interpolation are acquired as follows.
(1) With R as the main color of C,
LUT_C (R′1, G′2, B′1) and LUT_C (R′2, G′2, B′1)
(2) G as the main color of M
LUT_M (R′1, G′2, B′1) and LUT_M (R′1, G′1, B′1)
(3) B as the main color of Y
LUT_Y (R′1, G′2, B′1) and LUT_Y (R′1, G′2, B′2)
In this way, in addition to applying the color correction table to the main data (R′1, G′2, B′1), for each color, only the main color component is changed to that of the sub-data. By applying the color correction table CT, two values necessary for one-dimensional interpolation are acquired.

  Next, as the final stage of color correction, one-dimensional interpolation is performed as follows to obtain final data (C ″, M ″, Y ″).

C ″ is an interpolation calculation result of LUT_C (R′1, G′2, B′1) and LUT_C (R′2, G′2, B′1),
M ″ is an interpolation calculation result of LUT_M (R′1, G′2, B′1) and LUT_M (R′1, G′1, B′1),
Y ″ is an interpolation calculation result of LUT_Y (R′1, G′2, B′1) and LUT_Y (R′1, G′2, B′2).
Here, the interpolation calculation of each color component is performed using a weighting coefficient corresponding to the distance between the original data, main data, and sub data for each color component. For example, regarding the R (C) component, if the original data and the main data and the sub data are separated by distances d1 and d2, as shown in FIG. 26, C ″ is obtained by the following equation (5). .

  The one-dimensional interpolation for only the main colors as described above is easy to calculate and can effectively reduce the quantization noise of the pre-gradation number conversion. Note that this interpolation calculation may be performed only in the low density region.

[E] Details of Image Processing Common to Fourth to Eighth Embodiments In the first to third embodiments described above, when the color correction table CT is created, a grid obtained by finely dividing the color space in the low density region Use points. On the other hand, in the third and subsequent embodiments, the color correction table CT is created by dividing the color space at equal intervals as shown in FIG. In the following embodiment, instead of the configuration in which the color space is finely divided in the low density region, a configuration including a smoothing processing unit 150 is employed as shown in FIG. Hereinafter, image processing common to the embodiments will be described focusing on the configuration of the smoothing processing unit 150.

  In the image processing apparatus 30A described below, smoothing processing is performed after pre-gradation conversion and color correction in order to reduce quantization noise in pre-gradation conversion. In this process, each pixel in the color-corrected image is focused sequentially in the raster scan order, and a smoothing filter having a weighting coefficient is applied to the data of the pixel of interest and the data of several neighboring pixels. A weighted average of data is obtained, and the weighted average value is used as new data for the pixel of interest. This is repeated for all pixels.

FIG. 28 shows some examples of smoothing filters. FIG. 1A shows an example of a one-dimensional three-pixel smoothing filter, and the weighting coefficient of the pixel is shown in each pixel. As shown in the figure, a smoothing filter having an equal weight coefficient of 1/3 is applied to the pixel of interest Pi and two pixels Pi-1 and Pi + 1 adjacent to the pixel of interest Pi on the same line. That is, the data of these three pixels Pi-1, Pi, Pi + 1 are respectively (Ci-1, Mi-1, Yi-1), (Ci, Mi, Yi), (Ci + 1, Mi + 1, Yi). +1), the final data (C, M, Y) of the pixel of interest Pi is
C = (Ci-1 + Ci + Ci + 1) / 3
M = (Mi-1 + Mi + Ci + 1) / 3
Y = (Yi-1 + Yi + Yi + 1) / 3
It becomes. Note that the weighting factor may be non-uniform such that it is heavy on the target pixel and light on adjacent pixels. 28B to 28D show examples of a two-dimensional smoothing filter that refers to pixel data of adjacent lines. FIG. 4B refers to an example having an equal weighting factor by referring to six pixels including a line immediately before the line to which the target pixel Pi belongs. FIG. 9C refers to nine pixels including the upper and lower lines, and shows an example of a filter having an equal weight coefficient. Further, FIG. 4D shows an example of a filter in which only the weight coefficient is made unequal from FIG.

  By performing such smoothing processing, it is possible to solve the problem of degradation in image quality of the final output image due to the influence of the quantization error caused by the pre-gradation number conversion. In addition, when using the conventional distributed dither method in post-gradation conversion, the effect of quantization error in pre-gradation conversion appears and image quality deteriorates. In addition, by performing smoothing, even if a distributed dither is used for post-gradation number conversion, no deterioration in image quality is observed. Therefore, in addition to the error diffusion method, the average error minimum method, and the centralized dither, the distributed dither can be employed as the post-gradation number conversion.

  By the way, the smoothing process reduces the quantization noise of the pre-gradation number conversion in an area where gradations are evenly or continuously graded, but there is a problem that the clarity of edges of characters and figures is lost. Therefore, it is desirable to use edge detection together and correct the detected edge pixel smoothing process so as not to impair edge clarity. For example, the difference between the value of the target pixel and the value of the adjacent pixel is obtained, and if this difference is larger than a predetermined value, it is determined as an edge, and the smoothing process for the value of the target pixel does not refer to the pixel adjacent to the target pixel. To do.

  Such processing will be described in detail in an embodiment described later. The outline of the processing is as follows. In the case of the one-dimensional three-pixel smoothing shown in FIG. 28A, assuming that a predetermined value as a criterion for determining whether or not an edge is A, the processing of the following equation (6) can be performed on the value of the C color component. It ’s fine. Here, Ci is the value of the C color component of the pixel of interest i, and Ci−1 and Ci + 1 are the values of the C color component of pixels adjacent to the pixel of interest before and after.

  Similar processing is performed for the M and Y color component values. As a result, for the pixels in the edge portion where the absolute value of the gradation difference from the adjacent pixel is equal to or greater than the predetermined value A, the adjacent pixel is not referred to in the smoothing process, so the image quality of the image in the region other than the edge Increases smoothing without losing edge clarity.

[F] Fourth to Seventh Embodiments (1) Fourth Embodiment In the image processing apparatus 30A of the fourth embodiment, the pre-gradation number conversion unit 140, the color correction unit 142, and the smoothing processing unit 150 are shown in FIG. This is realized by the processing shown. The smoothing processing unit 150 is included in the color correction module 98 in the configuration shown in FIG. In this embodiment, the position of the target pixel in the main scanning direction is indicated by h, and the position of the sub scanning direction is indicated by v. Further, in the following description, each color component of the original color image data ORG before the gradation number conversion by the pre-gradation number conversion unit 140 is expressed as Rs [h, v], Gs [h, v], Bs [h. , V], and each color component of the grid point color image data GCD after the pre-gradation conversion is represented as Rn [h, v], Gn [h, v], Bn [h, v]. . By the pre-gradation number conversion, the number of gradations is reduced to 16 gradations or 8 gradations in each color in this embodiment, and this data Rn [h, v] etc. is assigned to which grid point. It can also be seen as indicating whether it is a thing. In the fourth embodiment, the color correction table CT is a table including conversion from RGB to four colors of CMYK. The color components of the color correction data GCD are represented by Cc [h, v], Mc [h, v]. , Yc [h, v], Kc [h, v]. In FIG. 29, [h, v] may be omitted for convenience of illustration.

  When the processing shown in FIG. 29 is started, first, the pre-gradation number conversion unit 140 performs pre-gradation number conversion (step S50). In the pre-gradation conversion, each color component Rs [h, v], Gs [h, v], Bs [h, v] corresponding to the original color image data ORG is assigned to a grid point prepared in advance, and its gradation This is a process of reducing the number. This method has been described in detail in the first to third embodiments. In FIG. 29, the pre-gradation number conversion is shown as a function PreConv (). Each color component of the grid point color image data GCD obtained by this pre-gradation number conversion is Rn [h, v], Gn [h, v], Bn [h, v].

  Next, color correction is performed by referring to the color correction table CT stored in advance in the color correction table memory 134 based on each color component of the grid point color image data obtained by the pre-gradation number conversion (step S51). . This process corresponds to the process by the color correction unit 142. In accordance with the color correction, conversion from RGB to four colors of CMYK which are ink colors output from the final color printer 22 is also performed. Each color component of the color correction data CCD obtained by the color correction process is Cc [h, v], Mc [h, v], Yc [h, v], Kc [h, v]. In FIG. 29, the color correction processing is shown as a function RefLUT () because it refers to (looks up) the color correction table CT.

  After obtaining each color component of the color correction data CCD in this way, smoothing processing is performed by the smoothing processing unit 150 of the present embodiment. In this smoothing process, first, it is determined whether or not to perform the smoothing process between the pixel of interest [h, v] and the adjacent pixels (step S52). As described above, various methods can be considered as to when to perform the smoothing process. In this embodiment, as shown in FIG. 30, after the pre-gradation conversion of the pixel of interest. The difference between the data Rn [h, v], Gn [h, v], Bn [h, v] and the data of the immediately preceding pixel [h−1, v] adjacent to the data Rn [h, v], Gn [h, v] When the value is 1 or less, it is determined that the smoothing process is performed. That is, when each color component of the pixel of interest and each color component of the previous pixel in the main scanning direction are on the same or adjacent grid points by the conversion by the pre-gradation number conversion unit 140 It is determined that the smoothing process is performed. Based on such determination, it is determined that smoothing processing is not performed on an edge or the like inherent in the image.

  If it is determined that the smoothing process is not performed, the color components Cc [h, v], Mc [h, v], Yc [h, v], and Kc [h, v] of the pixel of interest are corrected. The output data Cs, Ms, Ys, and Ks are output as they are (step S53) and output to the post-gradation number conversion unit 146. On the other hand, if it is determined that the smoothing process is to be performed, each color component Cc [h−1, v], Mc [h−1, v], Yc [h−1, after color correction of the previous pixel is performed. v], Kc [h−1, v] and the respective color components of the pixel of interest are calculated and averaged (step S54). Using this as output data, the post-gradation number conversion unit 146 performs the post-gradation number conversion. Migrate to In the present embodiment, as the post-gradation number conversion process, since the output of the color printer 22 is a binary process of forming / not forming ink dots, a halftoning process such as error diffusion is performed. . A description of this process is omitted.

  In the image processing apparatus 30A of the fourth embodiment configured as described above, since the smoothing processing unit 150 performs smoothing, the influence of the quantization error caused by the pre-gradation number conversion can be reduced, resulting in the quantization error. Degradation of image quality can be prevented. In addition, based on the data after the pre-gradation conversion, when one of the respective color components is assigned to a grid point that is separated from the adjacent grid point by comparison with the pixel immediately before the target pixel The smoothing process is not performed. As a result, sharpness such as edges that are originally present in the image is not lost by the smoothing process. In addition, since whether or not to perform smoothing is determined by comparison with the data of the pixel processed immediately before, the data stored for comparison is not wasted and the storage capacity can be reduced. be able to. Furthermore, since only the comparison with one adjacent pixel is sufficient, the amount of calculation can be reduced, and the overall processing can be speeded up.

  In the fourth embodiment described above, whether or not smoothing is performed is determined based on the separation of each color component of the data after the pre-gradation conversion, but various variations can be considered for this determination. . For example, as shown in FIG. 31, instead of the determination based on the data Rn [h, v], Gn [h, v], Bn [h, v] after the pre-gradation conversion, this is a source for obtaining this. The determination may be made using data Rn0 [h, v], Gn0 [h, v], Bn0 [h, v] indicating the range to which the original color image data belongs. At this time, whether the data Rn0 [h, v], Gn0 [h, v], Bn0 [h, v] belong to any of the rectangular parallelepipeds in which the pixel of interest is surrounded by eight adjacent lattice points 31, the condition shown in FIG. 31 is that the pixel of interest and the pixel adjacent to it (the pixel immediately before in the main scanning direction) are the same rectangular solid surrounded by eight lattice points or the same Is equal to the determination of whether or not it belongs to a rectangular parallelepiped adjacent to. Under the conditions shown in FIG. 30, if the grid points assigned by the pre-gradation conversion for the respective color components are the same or adjacent to each other, the smoothing process is performed regardless of the relationship as the original color image data. However, in the example of FIG. 31, it is a condition that the original color image data is in the same space surrounded by adjacent grid points or a space adjacent thereto.

  Further, as shown in FIG. 32, smoothing is performed only when the original color image data of the pixel of interest and the adjacent pixels belong to the same rectangular space surrounded by adjacent grid points without considering the adjacent range. It is also possible to determine that

  In these examples, the distance between adjacent pixels is set to a value of 1 or less. However, the smoothing condition may be relaxed so that smoothing is performed even if the distance is 2 or more. Since the number of gradations after the conversion of the number of pre-gradations differs for each color component, it is possible to make the value 2 or more only for the determination of a specific color. Alternatively, as shown in the following formula (7), the determination may be made based on the sum of the distances between the color components.

  In this case, even if a specific color is assigned to a position separated from adjacent grid points, smoothing is performed if other colors are assigned to the same grid point. Or it is also possible to consider the condition that the sum of the squares of the distances between the color components is equal to or less than a predetermined value, as in the following equation (8).

  It is not determined based on the data after the pre-gradation conversion, but the original color image data Rs [h, v], Gs [h, v], Bs [h, v] are represented as shown in FIG. It is also possible to judge by using. In the embodiment, since the original color image data is data of 256 gradations from 0 to 255, in FIG. 33, smoothing is performed when the separation of each color component is 16 gradations or less between adjacent pixels. Judgment was made. Note that the separation for each color component can be freely increased or decreased from 16 gradations. It is desirable that this gap as a criterion for determination is determined by actually processing a predetermined image and verifying it. It is also desirable to use it together with the judgment on the data after the pre-gradation number conversion described above.

  In particular, in the case of the color printer 22, the influence of the quantization error due to the pre-gradation number conversion is felt most strongly in the low density area of the finally obtained image, the so-called highlight area. Therefore, it is also practical to perform smoothing only near the highlight. FIG. 34 shows an example in which it is determined that smoothing is performed only in the highlight area. In this case, since the original color image data is determined to be a low density region, the original color image data Rs, Gs, Bs of the pixel of interest [h, v] and its adjacent pixels [h-1, v] are used. However, smoothing processing is performed when both of 224 gradations of 0 to 255 gradations are used. In this example, the value of 224 gradations adopted as the determination threshold value can be variously changed. Appropriate values can be determined by processing various images.

(2) Fifth Example Next, a fifth example will be described. In the image processing apparatus 30A of the fifth embodiment, the pre-gradation number conversion unit 140, the color correction unit 142, and the smoothing processing unit 150 are realized by the process shown in FIG. In this embodiment, the smoothing process is performed not only between pixels adjacent in the main scanning direction but also between pixels adjacent in the sub-scanning direction. When the processing shown in FIG. 35 is started, first, the pre-gradation number conversion by the pre-gradation number conversion unit 140 (step S60) and the color correction (step S head 61) by the color correction unit 142 described in FIG. 29 are performed. Next, it is determined whether or not smoothing is performed in the main scanning direction (step S62). The contents of this determination are as described in the fourth embodiment and its modifications, and various standards can be adopted.

  It is determined whether or not smoothing is performed (step S62), and when it is determined that smoothing is not performed between adjacent pixels in the main scanning direction, color correction data Cc [h, v], Mc of the target pixel. [H, v], Yc [h, v], Kc [h, v] are doubled as they are and temporarily stored as data Cs, Ms, Ys, Ks (step S63). On the other hand, when it is determined that smoothing is performed between pixels adjacent in the main scanning direction, as described in the fourth embodiment, in order to average the color correction data between adjacent pixels, A process of adding the color correction data of the pixel of interest and the adjacent pixel and temporarily storing it as data Cs, Ms, Ys, Ks is performed (step S64).

  Next, it is determined whether or not smoothing is performed between adjacent pixels in the sub-scanning direction (step S65). As shown in FIG. 36, whether or not smoothing is performed depends on the difference between the grid point color image data for each color component with respect to the pixel [h, v−1] immediately before in the sub-scanning direction. This can be done by determining whether the value is 1 or less. Also for this determination, various determinations shown in the above-described fourth embodiment and its modifications can be applied. Both determinations (steps S62 and S65) do not necessarily have the same contents. For example, when the resolution of the vertical direction and the horizontal direction of the color printer 22 as the final image output device is different, it is also practical that the judgment as to whether or not to perform smoothing differs between the main scanning direction and the sub-scanning direction. is there.

  If it is determined that smoothing is not performed between adjacent pixels in the sub-scanning direction, the color correction data Cc [h, v], Mc [h, v], Yc [h, v], Kc of the target pixel [H, v] is directly added to the previously obtained data Cs, Ms, Ys, Ks (step S66). On the other hand, when it is determined that smoothing is performed between adjacent pixels in the sub-scanning direction, the previously obtained data Cs, Ms, and Ys are used to average the color correction data with the adjacent pixels. , Ks, the color correction data Cc [h, v−1], Mc [h, v−1], Yc [h, v−] of the pixels in the line immediately before the target pixel (pixels adjacent in the sub-scanning direction). 1], Kc [h, v-1] are added (step S67).

  Thereafter, the data Cs, Ms, Ys, Ks obtained in step S66 or S67 is divided by the value 3, an average value is obtained, and this is used as the smoothed data Cs, Ms, Ys, Ks (step S68). The process shifts to a post tone number conversion process by the post tone number conversion unit 146.

  According to this configuration, the smoothing process can be performed not only on the pixels adjacent in the main scanning direction but also on the pixels adjacent in the sub-scanning direction. Can be prevented in both the main scanning direction and the sub-scanning direction, and the image quality can be further improved.

(3) Sixth Embodiment Next, as a sixth embodiment of the present invention, a color printer which is the image output device 20 uses a cyan ink in addition to the cyan ink C, magenta ink M, yellow ink Y and black ink K. A description will be given of smoothing when light cyan ink C2 having a lower density than C2 and light magenta ink M2 having a lower density than magenta ink M can be ejected. In the hardware configuration shown in FIG. 2, the color printer includes six rows of color heads 61 to 66 corresponding to the six colors of ink on the carriage 31, and the color ink cartridge 72 contains black ink K. Except for the five shades of ink. FIG. 37 shows the composition of the five shades of ink. Here, the light cyan ink C2 and the light magenta ink M2 having low density for cyan and magenta are provided when these inks are ejected at a low density in order to express an area where the density of the original color image data is low. This is because ink dots are visually recognized in that region, and a deterioration in image quality due to a so-called granular feeling is felt. In such an area where the density of the original color image data is low, the quality of the image is greatly improved by printing using the low density ink. The reason why the yellow ink Y does not have a low-density ink is that the yellow ink Y has high brightness and originally hardly produces a grainy feeling.

  FIG. 38 shows dot recording rates of cyan and magenta for low density ink (hereinafter referred to as light ink) and normal density ink (hereinafter referred to as dark ink). In this embodiment, printing is performed mainly for light ink in an area where the density of the original color image data is low, and the amount of dark ink used is gradually increased as the density of the original color image data increases. Reduce the amount of ink used. Above a predetermined density, printing is performed using only dark ink.

  In the image processing apparatus 30A of the sixth embodiment having such a configuration, the pre-gradation number conversion unit 140, the color correction unit 142, and the smoothing processing unit 150 are realized by the process shown in FIG. When this process is started, first, each color component Rs, Gs, Bs of the original color image data is subjected to a process corresponding to the pre-gradation conversion in the pre-gradation conversion unit 140 (step S70), and then converted. The color correction table CT is referred to based on the color components Rn, Gn, Bn of the grid point color image data, and a process corresponding to color correction in the color correction unit 142 is performed (step S71). At this time, the color correction table CT is prepared as a conversion table from six colors of R, G, B to C, M, Y, K, C2, M2.

  Next, it is determined whether or not to perform the smoothing process on the target pixel (step S72). This processing can basically use the judgment described in the fourth embodiment and its modification. If it is determined that smoothing is not performed, the color components Cc [h, v], Mc [h, v], Yc [h, v], Kc [h, v], C2c [ h, v], M2c [h, v] are used as output data Cs, Ms, Ys, Ks as they are (step S73) and output to the post-gradation number conversion unit 146. On the other hand, if it is determined that the smoothing process is to be performed, each color component Cc [h−1, v], Mc [h−1, v], Yc [h−1, after color correction of the previous pixel is performed. v], Kc [h-1, v], C2c [h-1, v], M2c [h-1, v] and each color component Cc [h, v], Mc [h, v], Yc of the pixel of interest. The arithmetic mean of [h, v], Kc [h, v], C2c [h, v], M2c [h, v] is calculated (step S74), and this is used as output data for the next post-gradation number. The process shifts to post-gradation number conversion by the conversion unit 146.

  According to this embodiment, in addition to the effects of the fourth embodiment, the color printer 22 includes the light cyan ink C2 and the light magenta ink M2, and appropriately prints light ink and dark ink to perform printing. Smoothing can also be performed with ink to prevent degradation in image quality due to quantization error in pre-gradation number conversion.

(4) Seventh Embodiment Next, a seventh embodiment of the present invention will be described. In the above embodiment, whether or not to perform smoothing depends on whether the data before the pre-gradation number conversion is used or the data after the pre-gradation number conversion is used, and a pixel adjacent to the target pixel (for example, Judgment was made based on the distance between the adjacent pixels in the main scanning direction or the sub-scanning direction. In contrast, in the seventh embodiment, the smoothing process is performed for each color component, and whether or not smoothing is to be performed is also determined for each color.

  In the image processing apparatus 30A of the seventh embodiment, the pre-gradation number conversion unit 140, the color correction unit 142, and the smoothing processing unit 150 are realized by the processing shown in FIG. When this process is started, first, each color component Rs, Gs, Bs of the original color image data is processed corresponding to the pre-gradation number conversion in the pre-gradation number conversion unit 140 (step S80), and then converted. The color correction table CT is referred to based on the color components Rn, Gn, and Bn of the grid point color image data, and processing corresponding to color correction in the color correction unit 142 is performed (step S81).

  Next, processing is performed to determine the distance between adjacent pixels with respect to the R component of the grid point color image data after the pre-gradation conversion of the target pixel (step S82). If the smoothing is not performed, the data Cc [h, v] color-corrected is set as the data Cs to be output as it is (step S83). On the other hand, if the separation of the R component of both pixels on the grid point is 1 or less, smoothing is performed and the cyan component Cc [h, v] of the color-corrected data and the previous one in the main scanning direction are used. A process of calculating an average with the cyan component Cc [h-1, v] of the data after the color correction of the pixel is performed (step S84).

  In the above processing, the smoothing determination for cyan C is performed by determining the R component of the original color image data. In the conversion from RGB to CMY, R and C, G and M, and B and Y are used. This is because it has a strong correlation.

  Therefore, after the above processing for cyan, the smoothing for magenta ink M is determined by the distance on the grid point for the G component (steps S85 to 87), and the smoothing for yellow ink Y is performed for the grid for the B component. Judgment is made based on the distance between points (steps S88 to S90). After the above processing, the process shifts to post-gradation number conversion.

  In the image processing apparatus 30A of the seventh embodiment having the above-described configuration, the same effects as those of the fourth embodiment can be obtained, and whether to perform smoothing for each color can be determined. Even in the case of conversion, it is possible to achieve both the smoothing effect and the sharpness of the boundary by not performing this.

  The preferred embodiment of the present invention has been described above, but the present invention is not limited to this embodiment, and various modifications, improvements, or modifications can be made without departing from the scope of the present invention. it can.

  For example, in the fourth to seventh embodiments, the smoothing determination is performed by paying attention to the distance between the target pixel and the pixel on the front side in the main scanning direction or the front side in the sub-scanning direction of the pixel. The determination may be made between the rear pixel [h + 1, v] or [h, v + 1] in the direction or the sub-scanning direction. Further, the smoothing range may be a range other than the range shown in FIG. In the above embodiment, the smoothing is determined or performed for all the color components. However, the smoothing process may be performed only for a specific color. For example, since the yellow ink Y has high brightness and is difficult to be visually recognized, the yellow ink can be excluded from smoothing from the beginning. In this case, the processing speed can be further increased. In addition, in a printer provided with both dark and light inks, it is also suitable that light inks that are less likely to produce graininess are excluded from smoothing.

  In the fourth to seventh embodiments, the details of the pre-gradation number conversion which is the premise of smoothing has not been described in particular, but not only the pre-gradation number conversion by the error diffusion method, the average error minimum method, etc. It is obvious that this can be combined with the pre-gradation number conversion by the dither method described in the second embodiment. By performing the smoothing, it is possible to sufficiently compensate for the deterioration of the image quality due to the dither method, and to obtain a good image output.

1 is a block diagram showing a schematic configuration of an image processing system of the present invention. It is a block diagram which shows an example of the aspect which implement | achieves the image processing system shown in FIG. 1 is a schematic configuration diagram illustrating a configuration of a color printer 22 as an example of an image output apparatus 20. 3 is an explanatory diagram illustrating the structure of a print head 28. FIG. It is explanatory drawing explaining the principle of the discharge of an ink. It is explanatory drawing which shows an example of the color space divided | segmented into the grid | lattice form. It is a block diagram which shows the function of the image processing apparatus 30 shown in FIG. 1 with a block. It is explanatory drawing which shows the position in color space of color image data and the lattice point color data of the vicinity. It is a block diagram which shows the function of the image processing apparatus in the case where a color printer is assumed as an image output apparatus. It is a schematic explanatory drawing of the pre-gradation number conversion process of image data. It is explanatory drawing of the weight matrix in the case of performing a pre-gradation number conversion process using an error diffusion method. It is a flowchart which shows the outline | summary of a process at the time of using an error diffusion method with a pre-gradation number conversion means. It is a general | schematic flowchart of a process at the time of using the average error minimum method in a pre gradation number conversion means. It is a flowchart which shows the outline | summary of a process at the time of using a systematic dither method with a pre-gradation number conversion means. It is explanatory drawing which shows the outline | summary of the process in the case of comprising the color correction part 142 like software. It is a schematic block diagram in case the color correction part 142 is comprised hard. FIG. 17 is an explanatory diagram showing a more detailed embodiment of the C ROM shown in FIG. 16. It is explanatory drawing which shows the outline | summary of the process which implement | achieves a post-gradation number conversion means. It is explanatory drawing which shows the outline | summary of a process in the case of calculating | requiring 4 color components of CMYK from the data of 3 colors of RGB, and performing color correction. It is explanatory drawing which shows an example of the grid point used in the image processing apparatus 30 of 1st Example, and was finely divided | segmented in the low density area | region of color space. It is a flowchart which shows the outline | summary of the process in the case of correcting a color correction table. It is a flowchart which shows the pre-gradation number conversion process in the image processing apparatus 30 of 2nd Example. It is explanatory drawing which shows an example of the dither matrix referred with the dither method employ | adopted by the pre gradation number conversion process of 2nd Example. It is a block diagram which shows the function of the image processing apparatus 30 of 3rd Example with a block. It is explanatory drawing for demonstrating the one-dimensional interpolation process performed about a main color. It is explanatory drawing explaining the principle of a primary interpolation process similarly. It is the block diagram which showed the function of 30 A of image processing apparatuses common to 4th thru | or 7th Example with the block. It is explanatory drawing which shows the specific example of a smoothing filter. It is a flowchart which shows the image processing routine in 4th Example. It is explanatory drawing which shows the detail of judgment of whether smoothing is performed. It is explanatory drawing which shows the other example of judgment of whether smoothing is performed. It is explanatory drawing which shows the other example of judgment of whether smoothing is performed. It is explanatory drawing which shows the other example of judgment of whether smoothing is performed. It is explanatory drawing which shows the other example of judgment of whether smoothing is performed. It is a flowchart which shows the image processing routine in 5th Example. It is explanatory drawing which shows the detail of judgment whether smoothing is performed in 5th Example. It is explanatory drawing which shows the component of the ink in the printer using 6 color ink of 6th Example. It is a graph which shows the recording rate of the light ink and dark ink in 6th Example. It is a flowchart which shows the image processing routine in 6th Example. It is a flowchart which shows the image processing routine in 7th Example.

Explanation of symbols

DESCRIPTION OF SYMBOLS 10 ... Image input device 12 ... Scanner 20 ... Image output device 21 ... Color display 22 ... Color printer 23 ... Paper feed motor 24 ... Carriage motor 26 ... Platen 28 ... Print head 30 ... Image processing device 30A ... Image processing device 31 ... Carriage DESCRIPTION OF SYMBOLS 32 ... Operation panel 34 ... Sliding shaft 36 ... Drive belt 38 ... Pulley 39 ... Position detection sensor 40 ... Control circuit 61 ... Ink ejection head 65 ... Introducing pipe 71 ... Black ink cartridge 72 ... Color ink cartridge 80 ... Ink Path 90 ... Computer 91 ... Video driver 93 ... CRT display 95 ... Application program 96 ... Printer driver 97 ... Rasterizer 98 ... Color correction module 99 ... Halftone module 134 ... Color correction table memory 134 ROM for ... C
134M ... ROM for M
134Y ... ROM for Y
DESCRIPTION OF SYMBOLS 140 ... Pre gradation number conversion part 140a ... Main data output part 140b ... Sub data output part 142 ... Color correction part 146 ... Post gradation number conversion part 148 ... Interpolation calculation part 150 ... Smoothing processing part 300 ... Grid point 400 ... Attention Pixel 500 ... Lattice points

Claims (38)

An image processing apparatus for color-correcting and outputting a multicolor image expressed by coordinate values in a two-dimensional or higher color space,
For each pixel of the image, input means for inputting color image data expressing the coordinate value using a predetermined number of gradations;
The color space is divided by the number of gradations smaller than the number of gradations representing the coordinate value, and the color space is divided more finely than the other areas in a predetermined low density area of the color space, and the division is performed. Grid point information storage means for storing the coordinate values of grid points obtained by performing for each dimension for the color space;
A color correction table storing correction data related to the color of the color image data corresponding to each grid point;
The coordinate values of the input color image data in the color space are determined according to a method in which a distance from the grid point is an average value or less on the average of grid points stored in the grid point information storage unit. Grid point conversion means for converting to coordinate values;
Color correction means for reading out correction data of lattice points corresponding to the converted coordinate values from the color correction table and outputting the corrected color image data;
An image processing apparatus.   2. The image processing apparatus according to claim 1, wherein the lattice point conversion means is means for converting the coordinates of the lattice points using an error diffusion technique.   The image processing apparatus according to claim 1, wherein the lattice point conversion means is means for converting the coordinates of the lattice points using a threshold matrix of distributed dither. The image processing apparatus according to claim 1,
The color image data is expressed in a format in which the number of gradations representing the coordinate value corresponds to the density of each color component of the color image data,
The color correction table stores information on the gradation of each color component of the color image data;
The grid point information storage unit is configured to store the color space such that the gradation interval of each color component of the color image data stored in the color correction unit is narrower than the other density regions in the predetermined low density region. An image processing apparatus characterized in that the predetermined low density region is divided. The image processing apparatus according to claim 4,
Furthermore, gradation number conversion means for converting the gradation number of each color component of the corrected color image data output by the color correction means and converting it to a final gradation number suitable for the image output device,
An image processing apparatus, wherein the number of gradations obtained by the coordinate value conversion performed by the grid point conversion unit is greater than the number of gradations converted by the gradation number conversion unit. The image processing apparatus according to claim 1,
The color correction table is an image processing apparatus that stores, as the correction data, data that matches color reproduction characteristics of an image output apparatus that finally processes the color image data output by the color conversion unit. An image processing apparatus for color-correcting and outputting a multicolor image expressed by coordinate values in a two-dimensional or higher color space,
For each pixel of the image, input means for inputting color image data expressing the coordinate value using a predetermined number of gradations;
The color space is divided by the number of gradations smaller than the number of gradations representing the coordinate value, and the color space is divided more finely than the other areas in a predetermined low density area of the color space, and the division is performed. Grid point information storage means for storing the coordinate values of grid points obtained by performing for each dimension for the color space;
A color correction table storing correction data related to the color of the color image data corresponding to each grid point;
The coordinate values of the input color image data in the color space are determined according to a method in which a distance from the grid point is an average value or less on the average of grid points stored in the grid point information storage unit. Grid point conversion means for converting to coordinate values;
Color correction data reading means for reading correction data of lattice points corresponding to the converted coordinate values from the color correction table;
A value corresponding to the main color is selected from the read correction data of the grid point, and for each main color, the correction data of the adjacent grid point in which only the coordinate value corresponding to the main color is shifted from the grid point. A value corresponding to the main color is read out, interpolation is performed using both values, and the correction data obtained by replacing the correction data for the main color with the interpolation result is output as final corrected color image data. An image processing apparatus comprising:   The image processing apparatus according to claim 7, wherein the interpolation in the color correction unit is linear interpolation. The image processing apparatus according to claim 7,
The coordinate value of the color image data is defined as a value corresponding to the density of each color component of the color image data, and the grid point conversion means performs first conversion of grid point coordinates as tone number conversion. Gradation number conversion means,
Furthermore, a second gradation number conversion means for converting the gradation number for each color component of the corrected color image data output from the color correction means into a gradation number suitable for the image output apparatus. Prepared,
An image processing apparatus characterized in that the number of gradations obtained by converting the coordinate value by the first gradation number conversion means is larger than the gradation number after being converted by the second gradation number conversion means. . An image processing apparatus for color-correcting and outputting a multicolor image expressed by coordinate values in a two-dimensional or higher color space,
For each pixel of the image, input means for inputting color image data expressing the coordinate value with a predetermined number of gradations;
A grid in which the color space is divided by the number of gradations smaller than the number of gradations expressing the coordinate value, and the coordinate values of the grid points obtained by performing the division for each dimension are stored for the color space. Point information storage means;
A color correction table storing correction data related to the color of the color image data corresponding to each grid point;
The coordinate values of the input color image data in the color space are determined according to a method in which a distance from the grid point is an average value or less on the average of grid points stored in the grid point information storage unit. Grid point conversion means for converting to coordinate values;
Color correction data reading means for reading correction data of lattice points corresponding to the converted coordinate values from the color correction table;
An image processing apparatus comprising: averaging processing means for performing processing for averaging correction data of each pixel read by the color correction data reading means based on correction data of pixels in the vicinity of each pixel.   The image processing apparatus according to claim 10, wherein the lattice point conversion means is means for converting coordinates of lattice points using a threshold matrix of distributed dither.   The image processing apparatus according to claim 10, wherein the lattice point conversion means is means for converting the coordinates of the lattice points using an error diffusion technique. The image processing apparatus according to claim 10,
The lattice point conversion means includes a first coordinate conversion means for converting the coordinates of the lattice points using an error diffusion technique, and a second coordinate conversion means for performing coordinate conversion of the lattice points using a threshold matrix of a distributed dither. Conversion means determining means for determining which of the first coordinate conversion means and the second coordinate conversion means is used,
Further, the averaging means performs the averaging process in a region where the density of the input color image data is low when the conversion means determination means determines the use of the first conversion means. An image processing apparatus.   The image processing apparatus according to claim 10, wherein neighboring pixels on which the averaging processing unit performs the averaging process are adjacent pixels along a direction in which the input unit inputs the color image data.   The image processing apparatus according to claim 14, wherein the adjacent pixel is a pixel on the near side in the input direction.   The image processing apparatus according to claim 14, wherein the adjacent pixel is a rear pixel in the input direction.   The image processing apparatus according to claim 10, wherein neighboring pixels on which the averaging processing unit performs the averaging process are both adjacent pixels before and after the input unit inputs the color image data.   11. The neighboring pixels on which the averaging processing unit performs the averaging process are adjacent pixels along a direction in which the input unit inputs the color image data and pixels in a direction intersecting the direction. Image processing apparatus. The image processing apparatus according to claim 10,
Further, the image forming apparatus includes density detecting means for detecting the density of the pixels based on the color image data of the pixels input by the input means,
The averaging processing means is an image processing apparatus which is means for performing averaging processing by the averaging processing means when the detected density of the pixel is not more than a predetermined value.   11. The image processing apparatus according to claim 10, wherein the averaging processing means is means for performing the averaging processing on the predetermined color component only in a region where the density of the predetermined color component of the input image data is not more than a predetermined value. . The image processing apparatus according to claim 10,
The coordinate value of the color image data is defined as a value corresponding to the density of each color component of the color image data, and the grid point conversion means performs first conversion of grid point coordinates as tone number conversion. Gradation number conversion means,
Furthermore, a second gradation number conversion means for converting the gradation number for each color component of the corrected color image data output from the color correction means into a gradation number suitable for the image output apparatus. Prepared,
An image processing apparatus characterized in that the number of gradations obtained by converting the coordinate value by the first gradation number conversion means is larger than the gradation number after being converted by the second gradation number conversion means. . The image processing apparatus according to claim 21, wherein
The second gradation number converting means is a means for performing binarization and expressing gradation by the distribution density of the binarized dots.
The averaging processing means is an image processing apparatus which is means for performing the averaging process when the density of the binarized dots is a predetermined value or less. The image processing apparatus according to claim 10,
The averaging processing means includes
Means for comparing the color correction data read by the color correction data reading means with the color correction data of adjacent pixels;
An image processing apparatus comprising: means for performing the averaging process when the difference between the two color correction data compared is equal to or less than a predetermined value. The image processing apparatus according to any one of claims 10 to 23, wherein:
The averaging processing unit determines whether or not to perform the averaging process for at least one of the colors constituting the color image data, and performs the averaging process for each color based on the determination. An image processing apparatus as means.   The averaging processing unit is a unit that performs the averaging process when it is estimated that a gap in the color space of the input color image data with respect to adjacent pixels is equal to or less than a predetermined distance. Item 15. The image processing apparatus according to Item 10.   The averaging processing unit is a unit that estimates that a distance between adjacent pixels is equal to or less than the predetermined distance when adjacent pixel points converted by the lattice point conversion unit are adjacent to each other. The image processing apparatus according to claim 25.   The averaging processing unit is configured such that when both the color image data of adjacent pixels are included in a unit space formed by lattice points stored in the lattice point information storage unit, 26. The image processing apparatus according to claim 25, which is means for estimating that the distance is equal to or less than the predetermined distance. The image processing apparatus according to claim 10,
The averaging processing means includes
Means for estimating whether or not a gap in the color space of the input color image data is less than or equal to a predetermined distance for adjacent pixels;
Means for determining the establishment of other execution conditions for the averaging process;
An image processing apparatus comprising: means for performing the averaging process only when it is determined that the distance between the two adjacent pixels is equal to or less than a predetermined distance and the other execution condition is satisfied. The image processing apparatus according to claim 10,
The grid point information storage means stores the coordinates of the grid points by dividing the color space more finely in a predetermined low density area of the color space than in other areas. An image processing apparatus for color-correcting and outputting a multicolor image expressed by coordinate values in a two-dimensional or higher color space,
For each pixel of the image, input means for inputting color image data expressing the coordinate value with a predetermined number of gradations;
A grid in which the color space is divided by the number of gradations smaller than the number of gradations expressing the coordinate value, and the coordinate values of the grid points obtained by performing the division for each dimension are stored for the color space. Point information storage means;
A color correction table storing correction data related to the color of the color image data corresponding to each grid point;
The coordinate values of the input color image data in the color space are determined according to a method in which a distance from the grid point is an average value or less on the average of grid points stored in the grid point information storage unit. Grid point conversion means for converting to coordinate values;
Color correction data reading means for reading correction data of lattice points corresponding to the converted coordinate values from the color correction table;
A value corresponding to the main color is selected from the read correction data of the grid point, and for each main color, the correction data of the adjacent grid point in which only the coordinate value corresponding to the main color is shifted from the grid point. An image processing unit comprising: a color correction unit that reads out a value corresponding to the main color, performs interpolation using the two values, and outputs the correction data obtained by replacing the correction data for the main color with the interpolation result. apparatus. The image processing apparatus according to claim 30, wherein
The coordinate value of the color image data is defined as a value corresponding to the density of each color component of the color image data, and the grid point conversion means performs first conversion of grid point coordinates as tone number conversion. Gradation number conversion means,
Furthermore, a second gradation number conversion means for converting the gradation number for each color component of the corrected color image data output from the color correction means into a gradation number suitable for the image output apparatus. Prepared,
An image processing apparatus characterized in that the number of gradations obtained by converting the coordinate value by the first gradation number conversion means is larger than the gradation number after being converted by the second gradation number conversion means. .   31. An averaging processing means for performing an averaging process on the correction data output by the color correction means based on correction data of pixels in the vicinity of each pixel and outputting as final color image data. The image processing apparatus according to claim 31. The image processing apparatus according to claim 10,
Furthermore, an image enlarging unit that assumes a new pixel between the pixel adjacent to the pixel by the color image data of each pixel input by the input unit and sets the input color image data to the new pixel. With
The image processing apparatus, wherein the averaging processing means is means for performing the averaging process with a pixel assumed by the image enlarging means. An image processing method for color-correcting and outputting a multicolor image represented by coordinate values in a two-dimensional or higher color space,
For each pixel of the image, input color image data expressing the coordinate value using a predetermined number of gradations,
The color space is divided by the number of gradations smaller than the number of gradations representing the coordinate value, and the color space is divided more finely than the other areas in a predetermined low density area of the color space, and the division is performed. Storing the coordinate value of the grid point obtained by performing for each dimension for the color space,
Corresponding to each grid point, a color correction table storing correction data related to the color of the color image data is prepared,
The coordinate value in the color space of the input color image data is converted into the coordinate value of the stored grid point according to a method in which the distance from the grid point is equal to or less than a predetermined value on average,
An image processing method for reading correction data of lattice points corresponding to the converted coordinate values from the color correction table and outputting the corrected color image data. An image processing method for color-correcting and outputting a multicolor image represented by coordinate values in a two-dimensional or higher color space,
For each pixel of the image, input color image data expressing the coordinate value using a predetermined number of gradations,
The color space is divided by the number of gradations smaller than the number of gradations representing the coordinate value, and the color space is divided more finely than the other areas in a predetermined low density area of the color space, and the division is performed. Storing the coordinate value of the grid point obtained by performing for each dimension for the color space,
Corresponding to each grid point, a color correction table storing correction data related to the color of the color image data is prepared,
The coordinate value in the color space of the input color image data is converted into the coordinate value of the stored grid point according to a method in which the distance from the grid point is equal to or less than a predetermined value on average,
Read out the correction data of the grid points corresponding to the converted coordinate values from the color correction table,
A value corresponding to the main color is selected from the read correction data of the grid point, and for each main color, the correction data of the adjacent grid point in which only the coordinate value corresponding to the main color is shifted from the grid point. A value corresponding to the main color is read out, interpolation is performed using both values, and the correction data obtained by replacing the correction data for the main color with the interpolation result is output as final corrected color image data. Image processing method. An image processing method for color-correcting and outputting a multicolor image represented by coordinate values in a two-dimensional or higher color space,
For each pixel of the image, input color image data expressing the coordinate value with a predetermined number of gradations,
Dividing the color space by the number of gradations smaller than the number of gradations representing the coordinate value, storing the coordinate values of grid points obtained by performing the division for each dimension for the color space;
Corresponding to each grid point, a color correction table storing correction data related to the color of the color image data is prepared,
The coordinate value in the color space of the input color image data is converted into the coordinate value of the stored grid point according to a method in which the distance from the grid point is equal to or less than a predetermined value on average,
Read out the correction data of the grid points corresponding to the converted coordinate values from the color correction table,
An image processing method for performing a process of averaging the read correction data of each pixel based on correction data of pixels in the vicinity of each pixel.   37. The image processing method according to claim 36, wherein the averaging process is performed when it is estimated that a gap in the color space of the input color image data for adjacent pixels is equal to or less than a predetermined distance. An image processing method for color-correcting and outputting a multicolor image represented by coordinate values in a two-dimensional or higher color space,
For each pixel of the image, input color image data expressing the coordinate value with a predetermined number of gradations,
Dividing the color space by the number of gradations smaller than the number of gradations representing the coordinate value, storing the coordinate values of grid points obtained by performing the division for each dimension for the color space;
Corresponding to each grid point, a color correction table storing correction data related to the color of the color image data is prepared,
The coordinate value in the color space of the input color image data is converted into the coordinate value of the stored grid point according to a method in which the distance from the grid point is equal to or less than a predetermined value on average,
Read out the correction data of the grid points corresponding to the converted coordinate values from the color correction table,
A value corresponding to the main color is selected from the read correction data of the grid point, and for each main color, the correction data of the adjacent grid point in which only the coordinate value corresponding to the main color is shifted from the grid point. An image processing method for reading out a value corresponding to the main color, performing interpolation using the two values, and outputting the correction data obtained by replacing the correction data for the main color with the interpolation result.

Images