JP4479665B2 - Image processing apparatus, image processing method, and image processing program for halftone processing - Google Patents

Description

  The present invention relates to an image processing apparatus, an image processing method, and an image processing program that perform halftone processing, and in particular, performs halftone processing by a fixed cell and uses an input gradation value even for a low-bit output gradation value. The present invention relates to an image processing apparatus capable of generating an output gradation value that can form an image without impairing density characteristics and gradation characteristics.

  An image forming apparatus such as a printer has an image processing apparatus that generates output gradation data by performing halftone processing on input gradation data composed of multi-gradation values of pixels. The print engine forms an image according to the output gradation data generated by the image processing apparatus.

  As halftone processing, processing by a dot concentration type dither method called multilevel dither method is known. In the multi-value dither method, dots (halftone dots) having an area corresponding to an input gradation value are formed in a cell composed of a plurality of pixels. For this purpose, the image processing apparatus uses a gamma table (multi-value dither table) in which input gradation values of a plurality of pixels in a cell are used as input values and an output value for forming a halftone dot is used as a table. A corresponding dither matrix arranged in matrix is provided in advance, and an output gradation value is generated by referring to a gamma table corresponding to the dither matrix for the input gradation value.

  A desired screen can be provided by appropriately designing the dither matrix and the gamma table. Image processing for performing such halftone processing is described in, for example, Patent Documents 1 and 2.

In a general multi-value dither method, a dither matrix is formed so that a halftone dot grows from the center of a cell (the halftone dot size increases) as the input gradation value increases in a cell composed of a plurality of pixels. It is configured. As described above, when the halftone dot is grown at the fixed center position in the cell, the fine line of the input image is cut off due to missing dots, or the halftone dot cannot be grown at a position corresponding to the edge portion of the input image. Jaggy (jaggy shape) may occur in the shape of the edge portion.
JP 2000-85187 A JP 2000-228728 A

  Therefore, in order to solve the above problem, the present applicant obtains the centroid position of the cell from the gradation value of each pixel in the cell and matches the center of the halftone dot to the centroid position. A dither method has been proposed in an earlier application. For example: Such a multi-value dither method is called an AAM method (Advanced Amplitude Modulation method).

Japanese Patent Application No. 2004-137326 (not disclosed at the time of confirmation before filing)
Japanese Patent Application No. 2005-196256 (unpublished)
According to this AAM method, the ideal output total value to be output is set in advance corresponding to the input gradation value of the pixel in the cell, and the input gradation value is sequentially corresponded from the pixel at the center of gravity position. The output values are output by referring to the gamma table, and the output values are output to pixels near the center of gravity within a range where the sum of the output values in the cell does not exceed the ideal output total value. By doing so, even when non-uniform input gradation values occur in the cell, a high output value exceeding the ideal output total value is not generated.

  On the other hand, in recent years, due to a cost reduction request, the number of bits of an output value generated by halftone processing is made smaller than the usual 8 bits, for example, 4 bits, and driving for beam generation in the print engine from the output value There is a need to simplify the configuration of pulse width modulators that generate pulse signals. The pulse width modulator has a simpler circuit configuration as the number of input bits is smaller, and can meet the demand for cost reduction.

  However, if the number of bits of the output value of halftone processing is reduced, the number of gradations of the dot size that can be represented by the output value will be reduced, and the dots in the reproduced pixel will have high gradation characteristics and accurate density characteristics. Can not give. For example, if the output value is 8 bits, 256 gradations can be expressed, but if the output value is 4 bits, only 16 gradations can be expressed, and the gradation characteristics become low and a desired density cannot be reproduced. Become.

  Accordingly, an object of the present invention is to provide an image processing apparatus and an image processing method having high gradation characteristics and accurate density characteristics even when the number of gradations of an output value for image reproduction generated by halftone processing by the AAM method is small. And providing an image processing program.

  In order to achieve the above object, according to a first aspect of the present invention, there is provided an image processing apparatus for performing halftone processing on an input image in units of cells composed of a plurality of adjacent pixels. For each cell with a fixed number of pixels, the unit refers to the dither matrix configured corresponding to the pixels of the plurality of cells, and converts the input gradation values of the pixels of the input image to form halftone dots in the cells. Convert to output tone value. In this halftone process, the input tone value is converted into the output tone value by referring to the dither matrix so that the pixel at the center of gravity in the cell corresponds to the halftone dot center pixel of the dither matrix. The dither matrix has a plurality of gamma tables having a correspondence between the input gradation value and the output gradation value corresponding to each pixel of the plurality of cells, and a predetermined displacement with respect to the halftone dot center pixel in each cell. To the same index pixel group arranged on the vector, gamma tables having different discrete output tone values for the same input tone value are distributed and allocated within the same index pixel group.

  As described above, in the AAM method for generating the halftone dot output gradation value corresponding to the input gradation value for the fixed cell so that the barycentric pixel and the dot center pixel in the fixed cell correspond to each other, the gradation value conversion is performed. By applying a diffusion screen in which the output tone values of the gamma table for use are dispersed or diffused, the cost can be reduced or the image quality can be improved.

  In that case, in the AAM method, an ideal output total value consisting of the sum of the gradation values to be output corresponding to the average value or the total value of the input gradation values of the pixels in the cell is obtained, and this is calculated from the halftone dot center pixel. The generation of output gradation values in the order of surrounding pixels is repeated until the accumulated value reaches the ideal output total value. And by using the diffusion screen, the ideal output total value is higher than the average of the output gradation total values by the plurality of dither matrices for the cells of uniform input gradation values, or near its maximum value, An appropriate output gradation value can be generated in a cell having a uniform input gradation value.

In order to achieve the above object, according to a second aspect of the present invention, in an image processing apparatus for performing halftone processing in units of cells composed of a plurality of adjacent pixels with respect to an input image,
Refers to the dither matrix configured corresponding to the pixels of each cell having a fixed number of pixels, and converts the input tone values of the pixels of the input image to output tone values that form halftone dots in the cells. A halftone processing unit that
The halftoning unit is
A center-of-gravity position generation unit for determining a center-of-gravity position in the cell according to an input gradation value and a position of a pixel constituting the cell;
A gradation value conversion unit that generates an output gradation value corresponding to the input gradation value by referring to the dither matrix so that the pixel at the center of gravity in the cell corresponds to the halftone dot center pixel of the dither matrix And
The dither matrix has a plurality of gamma tables having a correspondence between input gradation values and output gradation values corresponding to the pixels, and is on a predetermined displacement vector with respect to the halftone dot center pixel in the plurality of cells. A representative gamma table is assigned to each of the arranged identical index pixel groups, and each representative gamma table is composed of a plurality of diffusion gamma tables distributed in a plane in the same corresponding index pixel group. The gamma table has any one of discrete gradation values in the vicinity of the output gradation value of the representative gamma table corresponding to the input gradation value.

Furthermore, in order to achieve the above object, a third aspect of the present invention provides an image processing method for performing halftone processing in units of cells composed of a plurality of adjacent pixels with respect to an input image.
Refers to the dither matrix configured corresponding to the pixels of each cell having a fixed number of pixels, and converts the input tone values of the pixels of the input image to output tone values that form halftone dots in the cells. A halftone processing step to
The halftone processing step includes
A center-of-gravity position generation step of obtaining a center-of-gravity position in the cell according to an input gradation value and a position of a pixel constituting the cell;
A gradation value converting step of generating an output gradation value corresponding to the input gradation value by referring to the dither matrix so that the pixel at the center of gravity in the cell corresponds to the halftone dot center pixel of the dither matrix And
The dither matrix has a plurality of gamma tables having a correspondence between input gradation values and output gradation values corresponding to the pixels, and is on a predetermined displacement vector with respect to the halftone dot center pixel in the plurality of cells. A representative gamma table is assigned to each of the arranged identical index pixel groups, and each representative gamma table is composed of a plurality of diffusion gamma tables distributed in a plane in the same corresponding index pixel group. The gamma table has any one of discrete gradation values in the vicinity of the output gradation value of the representative gamma table corresponding to the input gradation value.

Furthermore, in order to achieve the above object, a fourth aspect of the present invention provides an image processing program that performs halftone processing in units of cells composed of a plurality of adjacent pixels with respect to an input image.
Refers to the dither matrix configured corresponding to the pixels of each cell having a fixed number of pixels, and converts the input tone values of the pixels of the input image to output tone values that form halftone dots in the cells. A halftone processing unit to build
The halftoning unit is
A center-of-gravity position generation unit for determining a center-of-gravity position in the cell according to an input gradation value and a position of a pixel constituting the cell;
A gradation value conversion unit that generates an output gradation value corresponding to the input gradation value by referring to the dither matrix so that the pixel at the center of gravity in the cell corresponds to the halftone dot center pixel of the dither matrix And
The dither matrix has a plurality of gamma tables having a correspondence between input gradation values and output gradation values corresponding to the pixels, and is on a predetermined displacement vector with respect to the halftone dot center pixel in the plurality of cells. A representative gamma table is assigned to each of the arranged identical index pixel groups, and each representative gamma table is composed of a plurality of diffusion gamma tables distributed in a plane in the same corresponding index pixel group. The gamma table has any one of discrete gradation values in the vicinity of the output gradation value of the representative gamma table corresponding to the input gradation value.

  According to the above aspect of the invention, since the plurality of pixels constituting the cell are fixed and the halftone dot center is generated at the center of gravity of the cell, the reproducibility of the fine line is good and the edge jaggy can be suppressed. In addition, since a diffusion screen is used, image quality does not deteriorate even if the bit number of the output gradation value is lowered, or the image quality is improved if the bit number of the output gradation value is kept high.

  Hereinafter, embodiments of the present invention will be described with reference to the drawings. However, the technical scope of the present invention is not limited to these embodiments, but extends to the matters described in the claims and equivalents thereof.

  FIG. 1 is a system configuration diagram having an image processing apparatus according to the present embodiment. This system includes a host computer 10 that generates image data, and an image forming apparatus 20 that forms and outputs an image from the image data.

  The host computer 10 has an application unit 11 and a rasterizing unit 12. The application unit 11 is configured by installing an application program such as a word processor or a graphic generation tool, and generates image data to be formed such as character data, graphic data, and image data. The rasterizing unit 12 includes, for example, a specific function of the operation system and a printer driver, and converts the image data generated by the application unit 11 into input image data composed of a plurality of bit gradation values corresponding to pixels to form an image. Output to the device 20. If the input image data is monochrome data, it has, for example, 8 bits and 256 gradation values, and if the input image data is color data, RGB or CMYK has 8 bits and 256 gradation values, respectively. The rasterizing unit 12 may be provided in the image processing apparatus 21 in the image forming apparatus 20 in some cases.

  The image forming apparatus 20 includes an image processing device 21 and a print engine 22. The image processing apparatus 21 includes a halftone processing unit 211 and a pulse width modulation unit 212. The halftone processing unit 211 receives input image data from the host computer 10 and converts it into output image data having quantized data forming halftone dots. The pulse width modulation unit 212 converts the output image data into a laser drive pulse corresponding to the pixel, and outputs the laser drive pulse to the print engine 22.

  The print engine 22 includes a laser driver 221 and a laser diode or a line-shaped light emitting element 222. The laser driver drives the laser diode or the line-shaped light emitting element 222 according to the drive pulse, and outputs image data to the photosensitive drum. And an image is formed on a print medium. The print engine 22 may be an ink jet method or a thermal transfer method in addition to the electrophotographic method.

  The image processing apparatus 21 in the present embodiment can be configured by a dedicated hardware circuit. Alternatively, it may be configured by a combination of software and a processor that executes the software.

  FIG. 2 is a configuration diagram of the image processing apparatus 21 in the present embodiment. In this example, a ROM 25 for storing a program for the image processing apparatus 21 to perform halftone processing, a processor 24 for executing the program, a primary storage memory 26, and an input interface 23 for inputting input image data from a host computer. Composed. The pulse width modulation section can also be configured by this program and processor, or can be configured by a dedicated hardware circuit.

[AAM method]
FIG. 3 is a diagram showing a screen structure in the present embodiment. FIG. 3A shows an example of a fixed cell, in which a cell 200 composed of 8 pixels is assigned to a matrix pixel PX. When the gradation value in the cell is uniform, the pixel near the center of each cell 200 (pixel indicated by a dot) corresponds to the central pixel of the halftone dot formed in the cell. However, the shape of each cell 200 is not point-symmetric because only the pixel 210 is distorted to the right.

  In contrast, FIG. 3B is an example of a fixed cell and an example of a point-symmetric cell. That is, the cells 200 adjacent to the left and right share the pixel 210 at the boundary, thereby making the shape of the cell 200 substantially point-symmetric. If a point-symmetric cell is used, the center of gravity of the cell when the input gradation is uniform coincides completely with the center of the cell (dot in the figure). Therefore, if the distribution of the input gradation is slightly changed, the center of gravity of the cell is located within the center pixel of the cell, and thus the shift of the center pixel of the halftone dot is suppressed. Therefore, it is possible to suppress the generation of unpleasant noise due to the shift of the halftone dot position when the asymmetric cell is used.

  In the present embodiment, the screen of the halftone processing unit is configured based on the point-symmetric cell structure shown in FIG. Then, the handling of the gradation value of the shared pixel required along with the point symmetry is performed based on the sharing rate corresponding to the number of adjacent cells sharing the shared pixel.

  FIG. 4 is a diagram showing a screen structure and a sharing rate in the present embodiment. As described above, in the point-symmetric cell 200 shown in FIG. 4A, the left and right adjacent cells share the shared pixel 210. FIG. 4B shows an example of the sharing rate. In this screen structure, pixels 210 that are equidistant from the center pixel of each cell are shared by adjacent cells in order to make the cells symmetric. The number of shared pixels 210 is not necessarily one, and a plurality of shared pixels may be used. Further, the upper and lower adjacent cells may share the shared pixel.

  As described above, since the shared pixel 210 is shared by the left and right cells 200, the pixel is divided into pseudo equal parts in the drawing. Then, in order to equally divide the shared pixel, a ratio (sharing rate or contribution rate) of the pixel belonging to the cell is given to a plurality of pixels constituting the cell 200. FIG. 4B shows an example of the sharing rate. Since the sharing pixel 210 is shared by the two cells on the left and right, the sharing rate is 0.5 (the sharing rate is 1 because the number of cells to be shared is 2). / 2), and the other pixels are given a sharing rate of 1.0 (the sharing rate is 1/1 since the number of cells to be shared is 1). Although the sharing rate may be given only to the shared pixel 210, by giving a sharing rate of 1.0 to the pixels that are not shared as described above, it is possible to perform processing with the sharing rate without distinguishing all the pixels. By utilizing the sharing rate in this way, it is possible to design a screen having cells in which the shared pixels are equally divided.

[AAM halftone processing]
The outline of the halftone process of the present embodiment is as follows. A dither matrix that forms a halftone dot of a size corresponding to a gradation value in a point-symmetrical cell shown in FIG. 4 is designed in advance, and in halftone processing, the input gradation of a pixel is referred to by referring to the dither matrix. The value is converted into an output gradation value forming a halftone dot. However, the centroid position of the cell is obtained based on the input gradation value and coordinates of the pixels constituting the cell and the total value of the input gradation values, and the dither matrix of the dither matrix is set so that the pixel at the centroid position corresponds to the halftone dot center pixel. Make a reference. Thereby, since the halftone dot position in the cell matches the barycentric position, the image quality of the fine line and the edge portion of the image can be improved.

  Further, in the halftone process, the shared output gradation value is output as output gradation data by multiplying the converted output gradation value by the sharing rate. That is, by outputting an output gradation value corresponding to the sharing rate to a shared pixel shared by a plurality of adjacent cells, the shared pixel is prevented from having an unnecessarily high gradation value. The halftone processing unit prevents the sum of the shared output tone values (output tone values multiplied by the sharing rate) in the cell from exceeding the ideal output total value to be output in the cell. Note that the ideal output total value is a value obtained in consideration of the sharing rate, for example, an output value that corrects the device characteristics of the print engine in accordance with the average value of the input gradation values, and has a table in advance. It is calculated and obtained from the average input value by referring to it.

  In the present embodiment, the average input tone value and the gravity center position obtained from the input tone value are obtained by a calculation independent of the sharing rate. Thereby, an average value and a gravity center position faithful to the input gradation value can be obtained. Also, the output gradation value converted with reference to the dither matrix is multiplied by the sharing rate and output as a shared output gradation value. This prevents an excessive output gradation value from being given to the shared pixel.

  5 and 6 are flowcharts of the halftone process in the present embodiment. The halftone process will be described in detail according to this flowchart. In that case, the processing will be described by taking the first input gradation of FIGS. 7 to 10 as an example.

[Example of first input gradation value]
FIG. 7 is a diagram showing a first example of input gradation data. FIG. 7 shows an example of a uniform gradation image in which the input gradation values of all the pixels are all “40”, and a cell 200 composed of nine pixels is associated with the pixel matrix 100. Therefore, as shown in FIG. 7B, the halftone processing target cell 200 is composed of nine pixels each having an input gradation value of “40”.

  As shown in FIG. 5, the processor 24 constituting the image processing apparatus 21 of FIG. 2 reads the halftone processing program from the ROM 25 and starts processing (S10).

  Next, the image processing apparatus 21 calculates the sum of the input gradation values of the nine pixels in the cell 200 and divides it by the number of pixels “9” in the cell to obtain the average input gradation value (S11). ). In the case of the input tone values shown in FIG. 7, the sum of the input tone values is “360”, and the average input tone value is 360 ÷ 9 = 40.

  Next, the image processing device 21 calculates the barycentric position depending on the input tone value of the cell 200 from the input tone value of the pixel in the cell, the pixel position coordinate, and the total input tone value. (S12). The position of the center of gravity is obtained by the following arithmetic expression.

  FIG. 8 is a diagram showing the position of the center of gravity in the cell and the processing order in the cell. As shown in FIG. 8A, in the case of a cell 200 with uniform input gradation values, the center of gravity 110 is the center position of the cell 200. Therefore, the pixel to which the centroid 110 belongs is also the central pixel of the cell 200.

  As shown in FIG. 5, the image processing apparatus 21 determines the ideal output total value to be output in the cell from the average input gradation value “40” (S13). The ideal output total value is a total value of output gradation values to be output in a point-symmetric cell, and is a value depending on the number of shared pixels and the sharing rate. That is, in the example of FIG. 7, the average input gradation value “40” is multiplied by the number of pixels of the cell to obtain an ideal output total value “320” (= 40 × 8 pixels) in consideration of the number of shared pixels and the sharing rate. Become.

  FIG. 34B is a table example showing the correspondence between the average input gradation value and the ideal output total value. In this table, ideal output total values 0 to 2040 (= 255 × 8) are set as table elements corresponding to the average input gradation values 0 to 255 (8 bits, 256 gradations) on the horizontal axis. Yes. The maximum value of the ideal output total value is 2040 (= 255 × 8). Seven pixels in the cell have a sharing ratio of 1.0 and two pixels have a sharing ratio of 0.5. Therefore, the maximum gradation value 255 of the pixel should be 8 times (7 + 2 × 0.5 = 8).

  This table is set so that the ideal output total value corresponding to the average input tone value has a non-linear relationship, thereby correcting the device characteristics of the print engine (the characteristics of the output density with respect to the tone data). Yes. In the example of FIG. 34B, the ideal output total value is set higher in the low average input tone value region, and the ideal output total value is set lower in the high average input tone value region. In addition to this, a conversion table having different characteristics can be used depending on the device characteristics.

  The image processing device 21 determines an ideal output total value corresponding to the average input gradation value with reference to a preset ideal output total value table (S13).

  Next, the image processing device 21 refers to the dither matrix and converts the input tone value of each pixel in the cell into an output tone value corresponding to the input tone value (S17). Therefore, the processing order of the pixels is determined so that processing can be performed from the order close to the center-of-gravity pixel of the cell (S14). That is, as shown in FIG. 8B, the center-of-gravity pixel is in the processing order “1”, the left and right neighboring pixels are in the processing order “2, 3, 4, 5”, and the surrounding pixels are in the processing order “6”. , 7, 8, 9 ". In this embodiment, when referring to the dither matrix, the center of the dither matrix is shifted to the center pixel of the cell so that the center pixel of the halftone dot is positioned at the center of gravity of the cell (S15).

  FIG. 9 is a diagram showing a dither matrix in the present embodiment. This dither matrix is composed of an index matrix shown in FIG. 9A and a multi-value dither table shown in FIG. 9B. The index matrix has a table number indicating which multi-value dither table (gamma table) each pixel in the cell should refer to. The multi-value dither table associates an output tone value with an input tone value. There are four types of tables 0, 1, 2, and 3. Since table 0 has the highest output gradation value in the region where the input gradation value is low, it is associated with the halftone dot center pixel. Since the tables 1, 2, and 3 have the highest output gradation value as the input gradation value becomes higher in order, they are associated with pixels around the halftone dot.

  In the process of process step S15, the index matrix of the dither matrix is shifted so that the center pixel (table 1) of the halftone dot is located at the center of gravity of the cell. In the example of FIG. 7, since the center of gravity of the cell is at the center of the cell, the shift amount is 0 both in the upper and lower directions.

  5 and 6, the image processing apparatus 21 converts the input gradation value of each pixel of the cell into an output gradation value that forms a halftone dot, and multiplies the sharing rate to thereby share the output gradation value. Generate a value. For this purpose, each pixel is processed in the processing order of FIG. First, the processing target pixel number n = 1 is set (S16). Then, the output gradation value corresponding to the input gradation value “40” of the nth (first) pixel is determined with reference to the multi-value dither table 0 of FIG. 9 (S17). That is, the output gradation value is the maximum gradation value “255”. Then, the image processing apparatus multiplies the output gradation value converted by the multi-value dither table 1 by the sharing rate “1.0” to obtain a shared output gradation value (S18).

  The above processes S17 and S18 are repeated according to the process order until the process is completed for all the pixels in the cell (S21 and S24). While repeating this processing, the image output apparatus stores the shared output gradation value in the output buffer in the memory 26 (S20). However, the processes S17, S18, and S20 are repeated as long as the total of the shared output gradation values in the cell does not exceed the ideal output total value (NO in S19). If so (YES in S19), the output tone value of the pixel being processed is adjusted so as not to exceed the ideal output total value (S25).

  FIG. 10 is a diagram illustrating an example of the shared output gradation value in the output buffer in the example of FIG. The steps S17 to S25 in the case of the input gradation value shown in FIG. 7 will be described with reference to FIG. For the pixel with the processing order “1”, the input tone value “40” is converted into the output tone value “255” by referring to the multi-value dither table 0 according to the index matrix, and the sharing rate “1.0” is multiplied. The shared output gradation value “255” is written in the output buffer 120 as shown in FIG. The sum of the shared output gradation values at this time is “255”, which does not exceed the ideal output total value “320”, and therefore, step S19 is NO.

  Next, n = n + 1 is set (S24), and the input tone value “40” is converted into the output tone value “16” with reference to the multi-value dither table 1 for the pixel in the processing order “2”. , The shared output gradation value “16” multiplied by the sharing rate “1.0” is written in the output buffer 120 as shown in FIG. The sum of the shared output gradation values at this time is “271 (= 255 + 16)” and does not exceed the ideal output total value “320”, and therefore, step S19 is NO.

  Similarly, with respect to the pixels of the processing order “3”, “4”, and “5”, the input tone value “40” is converted into the output tone value “16” with reference to the multi-value dither table 1, and the sharing rate “ The shared output gradation value “16” multiplied by “1.0” is written into the output buffer 120 as shown in FIG. At the end of this process, the total of the shared output gradation values is “319” and does not exceed the ideal output total value “320”. Therefore, for the pixels of the processing order “6”, “7”, “8”, and “9”, referring to the multi-value dither tables 2 and 3, the input gradation value “40” is set to the output gradation value “0”. Become. Therefore, the gradation value conversion process that refers to the dither table ends.

  Finally, if an output value exists in the shared pixel in the processing in another cell, the value is added to the output buffer (S22). Since the output value is a shared output gradation value multiplied by the sharing rate, the output value of the added shared pixel does not exceed the maximum gradation value.

  If the total of the shared output gradation values exceeds the ideal output total value “320” (YES in S19), the shared output gradation value of the pixel is set so that the total does not exceed the ideal output total value. Adjust and output (S25).

  By forming a halftone dot according to the shared output tone value in the output buffer shown in FIG. 10C, a halftone dot centered on the pixel at the center of gravity is formed.

[Example of second input gradation value]
FIG. 11 is a diagram illustrating a second example of input gradation data. FIG. 11 shows an example of a non-uniform gradation image in which the input gradation value of the leftmost three pixels of the cell 200 is “40” and the other input gradation value “0”. A cell 200 composed of nine pixels is associated. Therefore, as shown in FIG. 11B, the cell 200 to be halftone processed includes three pixels having an input gradation value “40” and six pixels having an input gradation value “0”. It is configured.

FIG. 12 is a diagram showing the centroid position, processing order, and index matrix with respect to FIG. FIG. 13 is a diagram showing the shared output gradation value written in the output buffer. First, in steps S11 and S12 of FIGS. 5 and 6, the total input tone value “120”, the average input tone value “13 (= 120 ÷ 9)” in the cell, and the barycentric position 110 are obtained. The center of gravity position 110 is located at a pixel shifted by one to the left from the center pixel as shown in FIG. Also, the ideal output total value “100” to be output in step S13 is determined with reference to a table (FIG. 35) showing the correspondence between the average input gradation value and the ideal output total value. That is, in the example of FIG. 11, the ideal output total value “100” obtained in consideration of the average input gradation value and the sharing rate (= 40 × 1.0 × 2 pixels + 40 × 0.5 × 1 pixel). It becomes.
This ideal output total value is smaller than the total input gradation value “120” because the sharing rate is taken into consideration.

  Based on the position of the center of gravity, the processing order in the cell is determined as shown in FIG. The processing order “1” is a pixel shifted by one to the left from the center-of-gravity pixel. Accordingly, the halftone dot center of the index table of the dither matrix is shifted so as to coincide with the barycentric pixel by processing S15. The shift amount is (−1, 0) as shown in FIG.

  Therefore, the processes S17 to S25 are repeated. As shown in FIG. 13, the input tone value “40” of the pixel in the processing order “1” is converted to the output tone value “255” with reference to the multi-value dither table 0 (S17), and the sharing rate “1” .0 "to determine the shared output gradation value" 255 "(S18). However, since the shared output gradation value “255” has already exceeded the ideal output total value “100” (YES in S19), the shared output gradation value is “0” so as not to exceed the ideal output total value. 100 "(S25). As a result, the processes S17 to S25 are completed.

[Diffusion screen]
FIG. 14 is a configuration diagram of the image processing apparatus according to the present embodiment. The image processing device 21 corresponds to the halftone processing unit 211 that converts the input gradation value IN of the pixel into the image reproduction output data OUT as the output gradation value corresponding to the dot size of the pixel, and the input gradation value IN. And a lookup table 122 having a plurality of gamma tables 124-1 to 124-N each having an output gradation value OUT corresponding to the pixel position (X, Y) of the input image data. The lookup table 122 is composed of, for example, a high-speed static RAM, and includes an index table 120 having an index Index that specifies a gamma table to be referred to corresponding to the pixel position (X, Y), and a plurality of gamma tables 124- 1-124-N. The halftone processing unit 211 obtains the pixel position (x, y) of the index table 120 to be referred to according to the pixel position (X, Y), and refers to the index table 120 based on the pixel position (x, y). To do. Then, the output gradation value OUT is acquired with reference to the gamma tables 124-1 to 124-N corresponding to the index number Index of the referenced index table 120. Then, the halftone processing unit 211 supplies the acquired output gradation value OUT to the pulse width modulation unit 212, and the pulse width modulation unit 212 generates a drive pulse width signal PW and supplies it to a print engine (not shown). In the print engine, the photosensitive drum is irradiated with a beam based on the drive pulse width signal PW to form an electrostatic latent image on the photosensitive drum. Then, toner is attached to the electrostatic latent image, developed into dots, and transferred onto a recording medium such as recording paper. Therefore, the output gradation value OUT is data for reproducing a print image related to the pixel dot size. As will be described later, the output gradation value OUT includes information L / R indicating whether the position of the dot generated in the pixel is right-justified or left-justified. In the embodiment described below, the information L / R indicating whether the data is right-justified or left-justified is output from the index table 120.

  FIG. 15 is a diagram illustrating a relationship between a general index table and a gamma table. The index table 120 shown here stores gamma table numbers 1 to 6 to be referred to in a 12 × 12 pixel matrix. The index table 120 is a repetition of a 4 × 6 minimum matrix unit surrounded by a thick frame, and within this minimum matrix unit, there are two gamma tables 1, four gamma tables 2, two gamma tables 3, There are 8 gamma tables 4, 4 gamma tables 5, and 4 gamma tables 6. By repeating this minimum matrix unit, the gamma table 1 is repeatedly arranged at the positions of the displacement vectors (+2, +3), (+2, −3), (−2, +3), and (−2, −3). The However, the upper right position of the index table 120 in FIG. 2 is the origin of coordinates (X, Y). The gamma table 2 is adjacent in the X direction, and is repeatedly arranged at the positions of the displacement vectors (+2, +3), (+2, −3), (−2, +3), and (−2, −3). Other gamma tables are similarly arranged on unique displacement vectors.

  Thus, the same index number is given to the pixel groups arranged on the common displacement vector, and the gamma table corresponding to the index number is referred to. Therefore, the pixel group arranged on the same displacement vector in this way is referred to as the same index pixel group.

  FIG. 15 shows six types of gamma tables 124. The gamma table stores an output pulse width value that is an output gradation value corresponding to the input gradation value. In the figure, the gamma table 1 is a minimum input value to a maximum value in a low input gradation value region. The output pulse width value grows to As shown in the figure, the gamma tables 2, 3, and 4 have output pulse width values that grow from the minimum value to the maximum value in the intermediate input gradation value region, and the gamma tables 5 and 6 are relatively high. It has an output pulse width value that grows from the minimum value to the maximum value in the input gradation value region.

  FIG. 16 is a diagram illustrating the growth of the dot of the pixel when the input gradation value is increased in the example of FIG. According to the gamma table and the index table 120 as described above, in the region where the input gradation value is low (input gradation values 10, 40, 90) as the input gradation value increases, the pixel of index 1 is assigned. A dot is formed (input gradation value 10), a dot is subsequently formed at the pixel at index 2 (input gradation value 40), and a dot is formed at the pixel at index 3 in parallel with the pixel at index 2 (Input gradation value 90). That is, in an area where the input gradation value is low, dots grow sequentially on the pixels of indexes 1, 2, and 3 in the index table 20. Then, in a region where the input tone value is high (input tone values 150, 220, 250), a dot grows on the pixel with index 4 (input tone value 150), and subsequently, the pixels with index 5, 6 Dots grow (input gradation values 220 and 250).

  Therefore, a screen line is formed on the line connecting the pixels with the index 1 as indicated by the growth dots having the input gradation value of 90 or 150.

  The gamma table 124 of FIG. 15 has an 8-bit output pulse width for an 8-bit input gradation value. Therefore, it has an output of 256 gradations (0 to 255) with respect to an input of 256 gradations. Therefore, the output gradation value of the gamma table can be changed as necessary in accordance with the change in the gradation of the input gradation value. However, when the output pulse width data is 6 bits, the output can take only 64 gradations (0 to 63), and the output gradation value of the gamma table is set in accordance with the change in gradation of the input gradation value. Sometimes it cannot be changed. In addition, when the output pulse width data is 4 bits, the output is only 16 gradations (0 to 15), and the output gradation value cannot be changed corresponding to the gradation change of the input gradation value. Will further increase. Furthermore, when the number of bits of output data decreases, the gradation becomes more discrete, and the gradation resolution of the dot size that can be reproduced decreases.

  FIG. 17 is a diagram illustrating the change amount of the output tone value with respect to the change of the input tone value of the gamma table. In the gamma table shown in FIG. 15, when the output data is 8 bits, (B) the output data is 6 bits, and (C) the output data is 4 bits, the input gradation value is the first floor. The difference (change amount) of the total gradation values of the output data of the six gamma tables when the tone changes is shown. 17A, 17B, and 17C, the upper row shows the gamma table, and the lower row shows the difference between the total output gradation values.

  In the case of 8-bit output data in FIG. 17A, the gamma table 124 is the same as the gamma table 124 in FIG. In this case, the difference in the output pulse width when the input gradation value changes by one gradation is all other than 0 gradation, and the total output pulse width is also 255 for the 255 gradation changes of the input gradation value. The gradation changes over time. Therefore, an output pulse width value that changes faithfully with respect to a change in input gradation value can be generated as output data.

  In the case of 6 (M = 6) -bit output data in FIG. 17B, the minimum output width of each gamma table 1-6 is 255 / (2M-1) = 4, and the upper gamma table 124 is shown. According to the above, it can be seen that the interval between the discrete values of the output pulse width is rough compared to the case of FIG. Therefore, with respect to 255 gradation changes in the input gradation value, the total output pulse width becomes zero gradation change over 44 times, and the gradation change only 211 times.

Further, in the case of 4 (M = 4) bit output data in FIG. 17C, the minimum output width of each gamma table 1 to 6 is 255 / (2 ^M −1) = 17. According to the gamma table 124, the interval between the discrete values of the output pulse width is further roughened compared to the cases of FIGS. 17 (A) and 17 (B). For this reason, the total output pulse width has changed only 82 times with respect to 255 gradation changes of the input gradation value, and the gradation change becomes zero over 173 times. Moreover, the difference between the output pulse widths is mostly 0 or 17, and the difference between the output pulse widths may be 34 in the high input gradation value region.

  That is, as a result of reducing the number of output data bits and the number of output data gradations, the number of times the difference in output pulse width becomes zero increases, and the difference amount increases even if it changes. This means that the number of gradations of the output pulse width is greatly reduced and the difference is greatly increased with respect to the number of gradations of 255 of the input gradation value, and high gradation characteristics and accurate density characteristics are obtained. It means being damaged.

  FIG. 18 is an enlarged view of the gamma table when the output gradation is 4 bits. In this figure, the solid line represents the input / output characteristics of the gamma table G, and the vertical axis represents the gradation value when the output pulse width is 8 bits (256 gradations). As described above, since the minimum width of the 4-bit output gradation is 17, the possible discrete points are 0, 17, 34, 51, 68, 85, 102, 8 bits (256 gradations). 119, 136, 153, 170, 187, 204, 221, 238, 255. A part of the output discrete points 153, 170, and 187 is shown on the vertical axis in FIG. If the gamma table G indicated by the solid line is 8 bits, the gradation values of the points indicated by the squares W, X, and Y are output as output data, but if it is 4 bits, 153, 170, and 187 discrete values are output. Since only points can be taken, the gradation values indicated by black circles are output as output data. For example, in the case of 8-bit output, the square X has an output gradation value 161, whereas in the case of 4 bits, the output gradation value 153 of a black circle in the vicinity thereof (1001 = 9 in the case of 4 bits). That is, the 4-bit output data becomes more discrete and cannot be matched with the gradation value of the 8-bit output data. Further, on the other hand, the output gradation values 145 and 161 of the squares W and X in the 8-bit output data become the same output gradation value 153 in the 4-bit output data, and the difference between the output gradation values is zero. However, on the other hand, the output gradation values 161 and 167 of the squares X and Y in the 8-bit output data become 153 and 170 in the 4-bit output data, and the difference between the output gradation values is 17. Become. Therefore, with a 4-bit output tone value, the difference may be zero with respect to the change in the input tone value, or different from the expected difference in the output tone value with respect to the change in the input tone value. It can be a big difference.

  FIG. 19 is a diagram for explaining the principle of the diffusion gamma table in the present embodiment. According to the principle of the diffusion gamma table, for a square X in the case of 8-bit output, 4-bit output discrete points X1 (output gradation value 153), X2 in the vicinity of the output gradation value 161 of the square X (Output tone value 170) is assigned to a plurality of diffusion gamma tables, the number of diffusion gamma tables having discrete points X1 and X2 is optimized, and their average output tone value is closest to 161 of square X Like that. For example, if the number of discrete points X1 is 5 and the number of discrete points X2 is 4, the average value of the output gradation values of these 9 discrete points is (153 × 5 + 170 × 4) /9=160.55, which is a square. It becomes a value close to the output gradation value 161 of X. In this way, even when only an output gradation value having a large discrete width can be taken as in the case of 4-bit output, by assigning different discrete values X1 and X2 to a plurality of pixels, output of the square X of the gamma table G A gradation value very close to the gradation value can be expressed. Here, the gamma table G is referred to as a representative gamma table, and a table having different discrete values is referred to as a diffusion gamma table.

  FIG. 20 is a diagram illustrating an example in which a plurality of diffusion gamma tables G1, G2, and G3 are assigned to the representative gamma table G. In this example, in order to simplify the display, three diffusion gamma tables G1, G2, and G3 are assigned to the representative gamma table G. The diffusion gamma table G1 takes white round discrete points (for example, X2) among 4-bit discrete points (for example, X1 and X2) located in the vicinity of the square (for example, X) of the representative gamma table G. It has an output value that changes up and down in the vicinity. Similarly, the diffusion gamma table G2 takes triangular discrete points among the neighboring 4-bit discrete points, and the diffusion gamma table G3 takes diamond-shaped discrete points among the neighboring 4-bit discrete points. As an example, the square X is diffused or dispersed into two discrete points X1 and one discrete point X2. In this case, the average value is (153 × 2 + 170 × 1) /3=158.7. Thus, by increasing the number of diffusion gamma tables to be diffused or dispersed, the average value of the output tone values can be made closer to the output tone value of the representative gamma table G.

  As shown in the index table of FIG. 15, since there are a plurality of pixels having the same representative gamma table G in a predetermined area of the pixel matrix, the representative gamma table G of the plurality of pixels is changed to a different diffusion gamma table. By assigning to, it is possible to express the output tone value of the representative ideal gamma table in a plane within the predetermined area. The arrangement of the plurality of diffusion gamma tables may be distributed, but is preferably distributed irregularly or randomly rather than regularly. Also, the discrete value taken by each diffusion gamma table may be any of the 4-bit discrete values in the vicinity of the output tone value of the representative gamma table G, but preferably any of the nearby 4-bit discrete values. Try randomly.

  The method for generating the diffusion gamma table from the representative gamma table is summarized as follows. For a certain representative gamma table G, an output pulse width X corresponding to the input gradation value IN is acquired. In the example of FIG. 18, X = 161. Then, the adjacent discrete points xl = 170 and xs = 153 with respect to the output pulse width X = 161 as the target value are distributed to the diffusion number N = 3, and the distribution number L = 1 and NL = 2 of xl and xs. Ask for. The distribution number L of xl can be obtained by, for example, L = INT [{(X−xs) / (xl−xs)} × N + 0.5]. Here, INT [] indicates a truncation process. In the above example, L = INT [{(161−153) / (170−153)} × 3 + 0.5] = 1. For N = 3 diffusion gamma tables, a discrete value xl = 170 is randomly given to L = 1 and a discrete value xs = 153 is given randomly to NL = 2. The arrangement of N = 3 diffusion gamma tables is also determined randomly.

  The average value x of the L = 1 output value xl = 170 and the NL = 2 output value xs = 153 of the diffused N = 3 diffusion gamma tables is x = {xl × L + xs. × (N−L)} / N = 158.7. By increasing the diffusion number N, the accuracy of the average value x can be increased.

  FIG. 21 is a diagram showing the same index pixel group having a representative gamma table in the present embodiment. The index table 120 shown here is an example in which the index numbers 1 to 6 in the index table 120 in FIG. 15 are converted into new index numbers 0 to 11 as shown in the correspondence table 121A in FIG. The new index numbers of even numbers 0, 2, 4, 6, 8, and 10 left-justify the dots in the pixels, and the odd numbers 1, 3, 5, 7, 9, and 11 the pixels that right-justify the dots in the pixels. And the even / odd information L / R of the index number is added to the output gradation value OUT to the pulse width modulation unit. Therefore, index numbers 0, 2, 3, 4, 6, 7, 8, 9, 10, and 11 are assigned in the upper left 4 × 6 matrix of the index table of FIG. 15 is the gamma table 1 of FIG. 15, the index numbers 2 and 3 are the gamma table 2, the index numbers 4 and 5 are the gamma table 3, and the gamma tables 4 and 6 are the gamma tables 4 and 8, respectively. Reference numeral 9 refers to the gamma table 5, and reference numerals 10 and 11 refer to the gamma table 6.

  In FIG. 21, the same index pixel group with the index number 0 arranged at the position of the displacement vector (± 2, ± 3) is 2 in the upper left 4 × 6 small pixel matrix as shown in the figure. 24 are arranged in the 16 × 18 pixel matrix of FIG. Therefore, in this embodiment, a plurality (9 types) of diffusion gamma tables are assigned to the representative gamma table 0 of the same index pixel group having the index number 0. In addition, a plurality of diffusion gamma tables are distributed in a plane in the same index pixel group, and are preferably distributed at random.

  FIG. 22 is a diagram showing an index table for the diffusion gamma table according to the present embodiment. In the present embodiment, 80 identical index pixel groups with index 0 arranged in a wide pixel matrix (32 × 30 pixel matrix) as will be described later with reference to FIG. 25 are assigned to nine types of diffusion matrices. Only a part of the pixel matrix is shown in the index table 120A of FIG. The correspondence table 121B shown at the bottom in FIG. 22 shows the correspondence between the index numbers in FIGS. 21 are assigned index numbers 0, 12, 24, 36, 48, 60, 72, 84, and 96 of the diffusion gamma table. Moreover, the allocation method is a random arrangement in a plane. Similarly, nine or eight types of diffusion gamma tables are assigned to the same index pixel group having other index numbers in FIG. The index number of the assigned diffusion gamma table is shown in the correspondence table 121B.

  FIG. 23 is a diagram showing an example of output gradation values that can be taken by the diffusion gamma table shown in FIG. For the output gradation value 161 of the square X shown in FIG. 18, the 4-bit discrete point X1 (output gradation value 153) and the discrete point X2 (output gradation value 170) are the diffusion gamma tables 0, 12, 24, 36, 48, 60, 72, 84, 96 are randomly assigned. Of the nine types of diffusion gamma tables associated with the representative gamma table 0 of 24 pixels shown here, the 13 diffusion gamma tables have 10 discrete points X1 (output gradation value 153). Discrete points X2 (output gradation value 170) are randomly assigned to the diffusion gamma table. The allocation ratio is roughly 13: 10 = 5: 4. In this way, the arrangement of the plurality of diffusion gamma tables assigned to the representative gamma table is randomized, and discrete points near the output tone value of the representative gamma table are also randomly assigned to the diffusion gamma table. A gradation value that is very close to the output gradation value 161 of the gamma table can be spatially expressed. Also for the other diffusion gamma tables shown in FIG. 22, 4-bit discrete points in the vicinity of the output gradation values of the corresponding representative gamma tables are randomly assigned.

  FIG. 24 is a diagram illustrating an example of an 8-bit output representative gamma table index table and its representative gamma table. The index table 120 is the same as the index table 20 in FIG. 21, and the pixel size is enlarged to 32 × 30. As described with reference to FIG. 21, the same index pixel group of index 0 (representative gamma table 0) in the index table 120 is arranged at the position of the same displacement vector (± 2, ± 3). In the pixel matrix 32 × 30, there are a total of 80 identical index pixels with index 0. The representative gamma table 0 has an 8-bit output pulse width as the input gradation increases, as indicated by the black square in FIG. If the representative gamma table 0 is simply changed to a 4-bit output pulse width, only discrete values can be taken as shown by white squares in FIG.

  24 includes indexes of representative gamma tables 2, 3, 4, 6, 7, 8, 9, 10, 11 in addition to index 0 of representative gamma table 0. These pixel groups having the same index are also arranged at positions on the same displacement vector.

  FIG. 25 is a diagram showing an index table and a gamma table for the diffusion gamma table according to the present embodiment. The index table 120A is the same as the index table 120A in FIG. 24, and the pixel size is enlarged to 32 × 30. In the figure, the pixel group of index 0 corresponding to the representative gamma table 0 is displayed in halftone. Nine types of diffusion gamma tables are assigned to 80 identical index pixel groups (index 0) in FIG. That is, the representative gamma table 0 is dispersed or diffused into nine diffusion gamma tables 0 and 12, eight diffusion gamma tables 124, and nine diffusion gamma tables 36, 48, 60, 72, 84, and 96. In the index table 120A, index numbers of a plurality of diffusion gamma tables are assigned to other representative gamma tables.

  In the diffusion gamma table 124 shown in FIG. 25, 4-bit output pulse widths of nine types of diffusion gamma tables are plotted. That is, for the representative gamma table G indicated by a broken line, 4-bit discrete values X1 and X2 adjacent to the output value are assigned as output values of the diffusion gamma table. That is, one of the discrete values X1 and X2 is assigned to the output values of nine types of diffusion gamma tables. Therefore, a total of nine types of diffusion gamma tables are assigned to the discrete values X1 and X2. This discrete value is a single or a plurality of 4-bit discrete values that are close to the output value of the representative gamma table G, and a plurality of diffusion values such that these discrete values approximate the output value of the representative gamma table on average. Assigned to gamma table.

  26, 27, and 28 are diagrams showing the arrangement and specific table values of the nine types of diffusion gamma tables in FIG. 26 shows the arrangement and table values in the index table of the diffusion gamma tables 0, 12, and 24, and FIG. 27 shows the arrangement and table values in the index table of the tables 36, 48, and 60 in FIG. Indicates the arrangement of the tables 72, 84, and 96 in the index table and the table values, respectively. The index numbers in the index table are the same as in FIG. 25, although the font size is small and difficult to determine. However, according to the index tables shown in FIGS. 26 to 28, the positions and numbers of the diffusion gamma tables arranged in a random manner can be confirmed. The table value of the diffusion gamma table has a 4-bit discrete value. In this example, one of the upper and lower discrete values close to the output value of the representative gamma table indicated by the broken line is assigned to each diffusion gamma table. . Accordingly, the diffusion gamma table in FIGS. 26 to 28 can confirm whether each diffusion gamma table has an upper or lower discrete value in each gradation value from 0 to 25. That is, it is the same as the diffusion gamma table shown in FIG. 10, and the nine types of diffusion gamma tables have discrete values that approximate the output values of the broken line representative gamma table as output values, and the ratio of allocation of the discrete values. By controlling this, the average value of the output values of the diffusion gamma table becomes very close to the output value of the representative gamma table.

  From the above description, the lookup table in the present embodiment no longer has a representative gamma table, and has a plurality of diffusion gamma tables instead of the representative gamma table. In that sense, a virtual representative gamma table is assigned to the same index pixel group, and the virtual representative gamma table has a plurality of diffusion gamma tables. Then, the average of the output values of the plurality of diffusion gamma tables approximates the output value of the representative gamma table.

  FIG. 29 is a diagram showing the gamma tables of FIGS. 14 and 15 and the difference in output pulse width with respect to changes in input gradation values. It is a figure of the same kind as FIG. FIG. 29A shows the difference between the representative gamma tables 0, 2, 3, 4, 6, 7, 8, 9, 10, and 11 shown in FIG. Show. That is, as in FIG. 17A, the total output pulse value changes with respect to all 255 changes in the input gradation value.

  On the other hand, FIG. 29B shows the difference between the diffusion gamma table shown in FIG. 25 and the total output pulse value with respect to the change of the input gradation value. That is, as shown in the correspondence table 121B of FIG. 22, the representative gamma table 0 is diffused to the diffusion gamma tables 0 to 96, and the representative gamma tables 2 and 3 are diffused to the diffusion gamma tables 2 to 98 and 3 to 87. The representative gamma table 4 is diffused to the diffusion gamma tables 4 to 100, the representative gamma tables 6 and 7 are diffused to the diffusion gamma tables 6 to 102 and 7 to 103, and the representative gamma tables 8 and 9 are diffused gamma. The representative gamma tables 10 and 11 are diffused to the diffusion gamma tables 10 to 106 and 11 to 107, respectively.

  As shown in FIG. 29B, the average output value can be made equal to that of the representative gamma table even if the output value is a 4-bit discrete value by diffusing or dispersing in a plurality of diffusion gamma tables. . As a result, the total output pulse width difference changes in all 255 stages with respect to the change in the input gradation value, and the tendency of the difference is the same as in FIG. However, the scale of the vertical axis is 360 in FIG. 29B, which is nine times that of 40 in FIG. 29A.

  FIG. 29C shows a table converted into an average value of the nine types of diffusion gamma tables shown in FIG. 29B, and the difference in output pulse width with respect to the change in input gradation value in that case. In terms of the average value, it can be understood that the gamma table is almost the same as the representative gamma table of FIG. Also, the difference in output pulse width is different from that in FIG. 29B only in the scale of the vertical axis, and is the same scale and the same difference as the vertical axis of FIG. In FIG. 29C, it can be confirmed that each representative gamma table is diffused or dispersed into a plurality of diffusion gamma tables. First, even if the number of output gradations is reduced and the number of output gradations is small, Very close output values can be reproduced (refer to the gamma table). Second, even if the number of output gradations is reduced and the number of output gradations is small, a high gradation characteristic of 255 can be obtained (refer to the output pulse width difference). ). Therefore, even if the halftone processing unit converts to image reproduction output data with a reduced bit, it can have high-accuracy density characteristics and high gradation characteristics equivalent to high-bit output data.

  In the above embodiment, when converting 8-bit input gradation data into 4-bit output pulse width data, the representative gamma table is diffused or dispersed into a plurality of diffusion gamma tables. That is, the output of the representative gamma table is 8 bits and 256 gradations, whereas the output of the diffusion gamma table is 4 bits and 16 gradations, which is smaller than that. Even in such a case, high gradation characteristics and high-precision density characteristics are realized by diffusing or dispersing in a plurality of diffusion gamma tables. The present embodiment is not limited to this. Even when the 8-bit input gradation data is converted into 8-bit output pulse width data using the diffusion gamma table, each representative gamma table is the same as that. The image quality can be improved by diffusing or dispersing the plurality of diffusion gamma tables having 8-bit output data. In this case, it is possible to express a high gradation output density that cannot be represented by the 8-bit representative gamma table. In other words, the 8-bit representative gamma table can express only 256 gradations, but by diffusing or dispersing in a plurality of diffusion gamma tables, gradation characteristics of 256 gradations or more can be given to the output data. However, it is necessary to replace one representative gamma table with a plurality of diffusion gamma tables, which increases the memory capacity of the gamma table. In that case, for example, in addition to a common representative gamma table, by providing a plurality of diffusion gamma tables having only a difference value with a small number of bits, an increase in the overall memory capacity can be suppressed. In addition, the halftone processing unit may randomly output a difference value to the output data of the representative gamma table and output it as output data of the diffusion gamma table.

[AAM method using diffusion screen]
In this embodiment, the diffusion screens shown in FIGS. 14 to 29 are applied to the AAM method shown in FIGS. As a result, the AAM method can be realized with a low-bit pulse width modulator.

  FIG. 30 is a diagram showing an array of fixed cells in the present embodiment. This fixed cell is a point-symmetrical cell having the shared pixel shown in FIG. Fixed cells 200 composed of adjacent nine pixels are spread on the pixel matrix. The pixels at the left and right ends of each fixed cell 200 are shared with the adjacent fixed cells, whereby the center point (black point) of each fixed cell is located at the center of the central pixel.

  As shown in FIG. 9, four types of gamma tables 0 to 3 are associated with 9 pixels of the fixed cell. As shown in the index table, the center pixel has a gamma table 0, and the center pixel has displacement vectors (+1, 0), (-1, 0), (0, +1), (0, -1). Is the gamma table 1, the displacement vectors (+1, +1), (+1, -1), (-1, +1), (-1, -1) with respect to the center pixel are the gamma table 2, and the center A gamma table 3 is provided for each of the displacement vectors (+2, 0) and (−2, 0) for the pixel.

  Therefore, in order to apply a diffusion screen to this AAM method, a plurality of diffusion gamma tables in which gamma tables 0 to 3 for the same index pixel group on each displacement vector randomly have discrete values near the output value of the representative gamma table. Consists of. In this embodiment, each of the gamma tables 0 to 3 is composed of 12 types of diffusion gamma tables, and 12 types of diffusion gamma tables are associated with 9 pixels of each cell in a random combination.

  FIG. 31 is a diagram showing an index matrix in the present embodiment. As described above, each of the gamma tables 0 to 3 is composed of 12 types of diffusion gamma tables. Therefore, there are 4 × 12 = 48 diffusion gamma tables. These diffusion gamma tables are associated in a random combination in 255 types of index tables. For example, the type 1 index matrix includes a diffusion gamma table 0-7 for the representative gamma table 0, a diffusion gamma table 1-1, 1-2, 1-3 for the representative gamma table 1, and a diffusion for the representative gamma table 2. Gamma tables 2-3 and 2-6 are associated with diffusion gamma tables 3-1 and 3-3 for the representative gamma table 3. Similarly, the type 255 index matrix includes diffusion gamma table 0-5 for representative gamma table 0, diffusion gamma tables 1-2, 1-4, and 1-5 for representative gamma table 1 for representative gamma table 2. The diffusion gamma tables 2-2 and 2-8 are associated with the diffusion gamma table 3-5 corresponding to the representative gamma table 3.

  Then, the 255 types of index matrices shown in FIG. 31 are associated with fixed cells as shown in FIG. This is associated with 255 types of fixed cells that are twice the size of 128 types of fixed cells of 8 rows and 16 columns shown in FIG. That is, the index matrix of the present embodiment has a scale corresponding to at least 8 pixels × 255 types = 2040 pixels. However, there are 48 diffusion gamma tables as described above.

  As shown in FIG. 31, the diffusion gamma table for the same index pixel group is different within the cell and of course between the cells. As a result, the halftone dot size of each fixed cell differs randomly for input images with uniform input tone values within the cell, but the average of those halftone dot sizes corresponds to the input tone value within a wide area. It is the size to do. In other words, even if a low-bit pulse width modulator is used, it is possible to suppress deterioration of gradation characteristics.

  FIG. 32 is a diagram showing an example of the diffusion gamma table in the present embodiment. Each of the four types of representative gamma tables 0 to 3 is composed of 12 types of diffusion gamma tables, and the average of the output tone values of the diffusion gamma table is equal to the output value of the representative gamma table. FIG. 32 is the same as the upper table in FIG. 29C, although the number of representative gamma tables is different.

  FIG. 33 is a diagram showing a diffusion gamma table example and an ideal output tone value table example. FIG. 34 is a diagram showing a specific numerical table thereof. Each of these is an example of a table indicating output gradation values corresponding to low input gradation values 0 to 30 and ideal output total values. As shown in FIGS. 33A and 34A, the 12 types of diffusion gamma tables γ1 to γ12 for the representative gamma table 0 have different discrete output values for the same input tone value. In this example, since the output gradation value is 4 bits, discrete outputs of 0, 15, 31, 47, 63, 79, 95, 111, 127, 143, 159, 175, 191, 207, 223, and 255 are used. Has a value. The 12 kinds of diffusion gamma tables randomly have discrete output values in the vicinity of the 8-bit output gradation value of the representative gamma table for each input gradation value, and the average value thereof is the representative gamma table. It becomes almost equal to the output gradation value.

  When halftone processing by the AAM method described above is performed with reference to such a diffusion gamma table, the sum of output gradation values of each cell differs even for an input image having uniform input gradation values. Become. That is, the total output gradation value of each cell varies due to a discrete value due to a low bit output value. For this reason, the ideal total output value used as the upper limit of the output gradation value generated by the AAM method needs to be set slightly higher in accordance with the variation in the output gradation value.

  The ideal output total values shown in FIGS. 33B and 34B are set to the maximum values of 12 types of diffusion gamma tables γ1 to γ12. That is, the ideal output total value is always set higher or equal than any of the output gradation values of the diffusion gamma tables γ1 to γ12 corresponding to the input gradation values 0 to 20. This prevents the output tone value generated by the diffusion gamma table from being limited by the low ideal output total value in the multi-value dither processing in which tone conversion processing is performed. be able to.

  When the input gradation value is about 0 to 20, the total output gradation value in the cell is determined only by the representative gamma table 0. Therefore, the ideal total output value may be set to the maximum value among the output values of the 12 types of diffusion gamma tables γ1 to 12 of the representative gamma table 0. However, when the input tone value increases, the sum of the output tone values in the cell is determined by the output tone values of a plurality of representative gamma tables. Therefore, the ideal output total value for such input gradation values may be the maximum value of 255 types of output gradation total values obtained by referring to 255 types of index matrices.

  FIG. 35 is a diagram showing an ideal output total value table when a diffusion screen is adopted in the AAM method. FIG. 35A shows the range of the ideal output total value table, and FIG. 35B is an example of the ideal output total value table when the maximum value is adopted. The horizontal axis represents the average input gradation value 0 to 255, and the vertical axis represents the ideal output total value 0 to 2040 (= 255 × 8 pixels).

  In the present embodiment, the output tone value of each cell is generated by referring to the diffusion gamma table by multi-value dither processing. Therefore, 255 different output gradation total values are generated for 255 types of index matrices. That is, as shown in FIG. 35A, the output tone total value of each cell is an area between the maximum value 301 and the minimum value 302 for an input image having an arbitrary input tone value IN uniformly. Distributed within 310. Therefore, it is desirable that the ideal output total value is larger than the average value 300 but does not exceed the maximum value 301. In particular, the ideal total output value is desirably a value in the vicinity of the maximum value 301. By setting the maximum value to 301, it is possible to obtain an output gradation total value appropriately dispersed for each cell by the diffusion screen for an image having a uniform input gradation value, and thereby output a plurality of cells. It is possible to make the average of the gradation total values coincide with the value designed by the diffusion gamma table. Alternatively, it is desirable that the value is at least sufficiently higher than the average value 300 even if it is lower than the maximum value 301.

  In order to create a table of ideal output total values, it is assumed that uniform input gradation values are given in a cell, and gradation value conversion is performed for each of input gradation values 0 to 255 using 255 types of index matrices. To generate 255 kinds of output gradation total values. Then, the maximum value of 255 types of output gradation total values is set as the ideal output total value for each input gradation value. This ideal output total value is higher than the output total value obtained by referring to the representative gamma table (which is approximately equal to the average value), and is higher than any of the multiple output tone total values that vary as a result of the diffusion gamma table reference. high. Therefore, in multi-value dither processing, for image data with uniform input tone values, the total output tone value in the cell obtained by referring to the diffusion gamma table does not exceed the ideal output total value. Is done.

  36 and 37 are flowcharts of halftone processing in the present embodiment. FIG. 37 is a flowchart of the multi-value dither process in FIG. The halftone process will be described along these flowcharts.

  As shown in FIG. 36, the input image is first read (S30). Then, an index matrix corresponding to a plurality of fixed cells as shown in FIG. 30 is set for the upper left of the pixel matrix of the input image (S31). Then, in the fixed cell of each index matrix, the total input gradation value of the pixels in the cell and its average value are calculated (S32). If the total value is not 0 (NO in S33), the center-of-gravity position in the cell is calculated (S34). The calculation of the center of gravity position is performed by the above-described calculation formula. Then, a multi-value dither process S35 for generating an output gradation value for generating a halftone dot for the cell is performed. The above processing is repeated until the number Nmax of fixed cells in the index matrix is reached (S36, S37).

  When the fixed cell number Nmax is reached, as long as there are unprocessed pixels in the input image on the right side of the right end of the index matrix (YES in S38), the process is repeated while moving the index matrix one block to the right (S39). If there are no unprocessed pixels in the input image on the right side of the right end of the index matrix (NO in S38), the index matrix is moved to the leftmost position one step lower and the process is repeated (S41).

  As shown in FIG. 37, in the multi-value dither processing S35, the ideal output total value is obtained by referring to the lookup table (FIG. 35) based on the average value or the total value of the input gradation values in the cell (S50). . Then, the index matrix is shifted so that the dot center coincides with the center of gravity of the cell (S51). After that, for the pixels in the cell, the dot non-output pixel closest to the center of gravity of the cell is selected (S52), and the output tone value is extracted by referring to the index matrix and the diffusion gamma table associated therewith. (S53). In a range where the accumulated value of the output gradation value in the cell does not exceed the ideal output total value (NO in S54), the extracted output gradation value is output to the search pixel output buffer (S56). The output gradation value is output to the output buffer by multiplying the sharing rate in the case of the shared pixel. Further, information on the dot growth direction of the pixel is also recorded in the output buffer (S57). This information is stored in a diffusion gamma table, for example.

  The above processes S52 to S57 are performed for all the pixels in the cell (NO in S58, NO in S59) while the sum of the output gradation values does not reach the ideal output total value. When the accumulation of output gradation values extracted from the diffusion gamma table exceeds the ideal output total value (YES in S54), the extracted output gradation value is adjusted so that the accumulation becomes equal to the ideal output total value ( (Subtracted) and output to the output buffer (S55, S56). The above multi-value dither processing is equivalent to the processing described in FIGS.

  FIG. 38 is a diagram illustrating an example of output values generated by the multi-value dither processing according to the present embodiment. Here, an example in which the input image data has an input gradation value “40” that is uniform within the cell will be described. As shown in FIG. 37A, the input gradation value is “40” for all nine pixels in the cell. In this case, the total input gradation value is 320 (= 40 × 8), but in this embodiment, “349” larger than that is the ideal output total value. Referring to the diffusion gamma table for this input, three types of outputs shown in FIGS. 38B, 38C, and 38D are extracted.

  In FIG. 37B, the output gradation value “255” is extracted from the center pixel, the output gradation value “47” is extracted from the right side pixel, and the output gradation value “0” is extracted from the left side pixel. Is extracted. In this case, the total output value is “302”. In FIG. 37C, the output gradation value “255” is extracted from the center pixel, the output gradation value “47” is extracted from the right side pixel, and the output gradation value “47” is extracted from the left side pixel. Is extracted. The total output value in this case is “349”. In FIG. 37D, the output gradation value “255” is extracted from the center pixel, the output gradation value “63” is extracted from the right side pixel, and the output gradation value “63” is extracted from the left side pixel. It is assumed that “0” is extracted. In this case, the total output value is “318”. The combination of these three types of output gradation values is set to a desired ratio, and the average output total value becomes “320”.

  As described above, the ideal output total value is set to the maximum value, so that the output gradation value is extracted from all index tables without exceeding the ideal output total value. Thereby, the total output gradation in the cell becomes equal to the average value designed by the diffusion gamma table.

  In the case of non-uniform input image data, the output tone value extracted from the diffusion gamma table is adjusted so as not to exceed the ideal output total value as described with reference to FIGS.

[Other system configuration examples]
FIG. 39 is a diagram illustrating another system configuration example. In this example, a halftone processing unit 211 is provided in the host computer 10, and a pulse width modulation unit 212 is provided in the image processing device 21 in the image forming apparatus 20. That is, the above halftone process is performed by the computer executing the printer driver installed in the host computer 10. Therefore, in this case, the host computer 10 outputs the shared output gradation value data to the image forming apparatus 20 as image data. Therefore, the printer driver has a halftone processing program.

  FIG. 40 is a diagram illustrating another system configuration example. In this example, the input image data output from the host computer 10 is RGB color data. Accordingly, the image processing apparatus 21 in the image forming apparatus 20 includes a color conversion processing unit 213 that converts RGB into CMYK. Then, the CMYK four-plane input gradation values are supplied to the halftone processing unit 211. The halftone processing unit 211 generates the shared output gradation value data by performing the halftone process using the above-described point symmetric cell for each of the input gradation values of the four planes.

[Other Embodiments]
In the above embodiment, the dither matrix is composed of an index matrix and a plurality of multi-value dither tables. However, a dither matrix in which the index matrix and the multi-value dither table are combined may be used. That is, the dither matrix has a multi-value dither table corresponding to each of the nine pixels of the cell.

1 is a system configuration diagram having an image processing apparatus in the present embodiment. It is a block diagram of the image processing apparatus 21 in this Embodiment. It is a figure which shows the screen structure in this Embodiment. It is a figure which shows the screen structure and sharing rate in this Embodiment. It is a flowchart figure of the halftone process in this Embodiment. It is a flowchart figure of the halftone process in this Embodiment. It is a figure which shows the 1st example of input gradation data. It is a figure which shows the gravity center position in a cell, and the process order in a cell. It is a figure which shows the dither matrix in this Embodiment. It is a figure which shows the example of the shared output gradation value in the output buffer in the example of FIG. It is a figure which shows the 2nd example of input gradation data. It is a figure which shows the gravity center position, processing order, and index matrix with respect to FIG. It is a figure which shows the shared output gradation value written in an output buffer. It is a block diagram of the image processing apparatus in this Embodiment. It is a figure which shows the relationship between a general index table and a gamma table. FIG. 3 is a diagram illustrating dot growth of a pixel when an input gradation value is increased in the example of FIG. 2. It is a figure which shows the variation | change_quantity of the output gradation value with respect to the change of the input gradation value of a gamma table. It is an enlarged view of a gamma table when the output gradation is 4 bits. It is a figure explaining the principle of the diffusion gamma table in this Embodiment. It is a figure showing an example at the time of assigning a plurality of diffusion gamma tables G1, G2, G3 to representative gamma table G. It is a figure which shows the same index pixel group which has the representative gamma table in this Embodiment. It is a figure which shows the index table with respect to the diffusion gamma table by this Embodiment. It is a figure which shows an example of the output gradation value which the diffusion gamma table shown in FIG. 22 can take. It is a figure which shows an example of the index table of an 8-bit output representative gamma table, and its representative gamma table. It is a figure which shows the index table with respect to the diffusion gamma table by this Embodiment, and a gamma table. It is a figure which shows arrangement | positioning and table value of nine types of diffusion gamma tables of FIG. It is a figure which shows arrangement | positioning and table value of nine types of diffusion gamma tables of FIG. It is a figure which shows arrangement | positioning and table value of nine types of diffusion gamma tables of FIG. It is a figure which shows the difference of the output pulse width with respect to the change of an input gradation value and the gamma table of FIG. 24 and FIG. It is a figure which shows the arrangement | sequence of the fixed cell in this Embodiment. It is a figure which shows the index matrix in this Embodiment. It is a figure which shows an example of the diffusion gamma table in this Embodiment. It is a figure which shows the example of a diffusion gamma table, and the example of an ideal output gradation value table. It is a figure which shows the specific numerical table of FIG. It is a figure which shows the ideal output total value table when a diffusion screen is employ | adopted for the AAM method. It is a flowchart figure of the halftone process in this Embodiment. It is a flowchart figure of the halftone process in this Embodiment. It is a figure which shows the example of the output value produced | generated by the multi-value dither process by this Embodiment. It is a figure which shows the other system configuration example. It is a figure which shows the other system configuration example.

Explanation of symbols

  200: Cell 21: Image processing apparatus

Claims (13)

In an image processing apparatus that performs halftone processing in units of cells composed of a plurality of adjacent pixels on an input image,
Refers to the dither matrix configured corresponding to the pixels of each cell having a fixed number of pixels, and converts the input tone values of the pixels of the input image to output tone values that form halftone dots in the cells. A halftone processing unit that
The halftoning unit is
A center-of-gravity position generation unit for determining a center-of-gravity position in the cell according to an input gradation value and a position of a pixel constituting the cell;
A gradation value conversion unit that generates an output gradation value corresponding to the input gradation value by referring to the dither matrix so that the pixel at the center of gravity in the cell corresponds to the halftone dot center pixel of the dither matrix And
The dither matrix has a plurality of gamma tables having a correspondence between input gradation values and output gradation values corresponding to the pixels, and is on a predetermined displacement vector with respect to the halftone dot center pixel in the plurality of cells. A representative gamma table is assigned to each of the arranged identical index pixel groups, and each representative gamma table is composed of a plurality of diffusion gamma tables distributed in a plane in the same corresponding index pixel group. The gamma table has any one of discrete gradation values in the vicinity of the output gradation value of the representative gamma table corresponding to the input gradation value. In claim 1,
The halftone processing unit further comprises:
An ideal output total for generating an ideal output total value corresponding to the total value or average value of the input gradation values of a plurality of pixels constituting the cell, and consisting of the sum of gradation values to be output from the plurality of pixels in the cell. A value generator,
The gradation value conversion unit repeatedly generates the output gradation value in the order from the halftone dot central pixel in the cell to the surrounding pixels, and the sum of the output gradation values in the cell is the ideal value. The generation of the output gradation value is repeated until the output total value is reached,
The ideal output total value is set in the vicinity of the maximum value of the output gradation total value corresponding to the uniform input gradation value in the cell obtained by referring to the dither matrix assigned to the plurality of cells. A featured image processing apparatus. In claim 1,
The halftone processing unit further comprises:
An ideal output total for generating an ideal output total value corresponding to the total value or average value of the input gradation values of a plurality of pixels constituting the cell, and consisting of the sum of gradation values to be output from the plurality of pixels in the cell. A value generator,
The gradation value conversion unit repeatedly generates the output gradation value in the order from the halftone dot central pixel in the cell to the surrounding pixels, and the sum of the output gradation values in the cell is the ideal value. The generation of the output gradation value is repeated until the output total value is reached,
The ideal output total value is set to a value larger than the average of the output gradation total values corresponding to the uniform input gradation values in the cells obtained by referring to the dither matrix assigned to the plurality of cells. An image processing apparatus. In claim 3,
The image processing apparatus according to claim 1, wherein the ideal output total value is set so as not to exceed a maximum value of the output gradation total value corresponding to the uniform input gradation value in the cell. In claim 2 or 4,
The gradation value conversion unit is configured such that when the sum of the output gradation values in the cell exceeds the ideal output total value, the generated output gradation value is a gradation that does not exceed the ideal output total value. An image processing apparatus characterized by changing to a value.   2. The image processing apparatus according to claim 1, wherein the plurality of diffusion gamma tables are randomly distributed in the same index pixel group corresponding thereto.   2. The plurality of diffusion gamma tables according to claim 1, wherein the plurality of diffusion gamma tables randomly include any one of discrete gradation values near the output gradation value of the representative gamma table corresponding to the input gradation value. An image processing apparatus.   The average value of output gradation values of a plurality of diffusion gamma tables of the same index pixel group according to claim 1 is substantially equal to the output gradation value of a virtual representative gamma table of the same index pixel group. An image processing apparatus. 2. The output gradation value of the representative gamma table has N bits and 2 ^N gradations, and the output gradation value of the diffusion gamma table has M bits (M <N) and 2 ^M gradations. An image processing apparatus comprising: In an image processing method for performing halftone processing in units of cells composed of a plurality of adjacent pixels on an input image,
Refers to the dither matrix configured corresponding to the pixels of each cell having a fixed number of pixels, and converts the input tone values of the pixels of the input image to output tone values that form halftone dots in the cells. A halftone processing step to
The halftone processing step includes
A center-of-gravity position generation step of obtaining a center-of-gravity position in the cell according to an input gradation value and a position of a pixel constituting the cell;
A gradation value converting step of generating an output gradation value corresponding to the input gradation value by referring to the dither matrix so that the pixel at the center of gravity in the cell corresponds to the halftone dot center pixel of the dither matrix And
The dither matrix has a plurality of gamma tables having a correspondence between input gradation values and output gradation values corresponding to the pixels, and is on a predetermined displacement vector with respect to the halftone dot center pixel in the plurality of cells. A representative gamma table is assigned to each of the arranged identical index pixel groups, and each representative gamma table is composed of a plurality of diffusion gamma tables distributed in a plane in the same corresponding index pixel group. The gamma table has any one of discrete gradation values near the output gradation value of the representative gamma table corresponding to the input gradation value. In claim 10,
The halftone processing step further includes:
An ideal output total for generating an ideal output total value corresponding to the total value or average value of the input gradation values of a plurality of pixels constituting the cell, and consisting of the sum of gradation values to be output from the plurality of pixels in the cell. A value generation process,
The gradation value converting step repeatedly generates the output gradation value in the order from the halftone dot center pixel in the cell to the surrounding pixels, and the sum of the output gradation values in the cell is the ideal value. The generation of the output gradation value is repeated until the output total value is reached,
The ideal output total value is set in the vicinity of the maximum value of the output gradation total value corresponding to the uniform input gradation value in the cell obtained by referring to the dither matrix assigned to the plurality of cells. A featured image processing method. In an image processing program that performs halftone processing in units of cells composed of a plurality of adjacent pixels for an input image,
Refers to the dither matrix configured corresponding to the pixels of each cell having a fixed number of pixels, and converts the input tone values of the pixels of the input image to output tone values that form halftone dots in the cells. A halftone processing unit to build
The halftoning unit is
A center-of-gravity position generation unit for determining a center-of-gravity position in the cell according to an input gradation value and a position of a pixel constituting the cell;
A gradation value conversion unit that generates an output gradation value corresponding to the input gradation value by referring to the dither matrix so that the pixel at the center of gravity in the cell corresponds to the halftone dot center pixel of the dither matrix And
The dither matrix has a plurality of gamma tables having a correspondence between input gradation values and output gradation values corresponding to the pixels, and is on a predetermined displacement vector with respect to the halftone dot center pixel in the plurality of cells. A representative gamma table is assigned to each of the arranged identical index pixel groups, and each representative gamma table is composed of a plurality of diffusion gamma tables distributed in a plane in the same corresponding index pixel group. An image processing program characterized in that the gamma table has any one of discrete gradation values in the vicinity of the output gradation value of the representative gamma table corresponding to the input gradation value. In claim 12,
The halftone processing unit further comprises:
An ideal output total for generating an ideal output total value corresponding to the total value or average value of the input gradation values of a plurality of pixels constituting the cell, and consisting of the sum of gradation values to be output from the plurality of pixels in the cell. A value generator,
The gradation value conversion unit repeatedly generates the output gradation value in the order from the halftone dot central pixel in the cell to the surrounding pixels, and the sum of the output gradation values in the cell is the ideal value. The generation of the output gradation value is repeated until the output total value is reached,
The ideal output total value is set in the vicinity of the maximum value of the output gradation total value corresponding to the uniform input gradation value in the cell obtained by referring to the dither matrix assigned to the plurality of cells. A characteristic image processing program.

Images