JP2008544667A - Multi-level frequency modulation halftone screen and its manufacturing method - Google Patents

Abstract

  A method for calculating a set of threshold masks for the purpose of digital halftoning printing using multiple colorants is disclosed. The method is computationally efficient and produces a dot distribution whose power spectrum has blue noise characteristics. The method uses a monovalent function with a limited two-dimensional domain. The monovalent function has a maximum value at or near its origin and decreases monotonically toward the bounds of the limited domain. The dot distribution corresponding to the various colorants is optimized relative to each other to minimize the visibility of graininess. The method makes it possible to drive a printer that can draw more than 3 density levels per pixel.

Description

  The present invention relates to a digital image processing method. In particular, the present invention relates to a method for digital halftoning continuous tone images for printing purposes. More particularly, the present invention relates to frequency modulation halftoning techniques using threshold masks. In particular, the present invention relates to a frequency modulation halftoning technique that supports printing devices that can print two or more density levels per pixel.

  Ink jet printing works by jetting ink droplets through nozzles onto a substrate. Pressure wave that propels micro ink droplets at high speed through the nozzle toward the substrate by applying mechanical or thermal energy to the ink in a small chamber. ) Is created.

  The pressure wave is controlled by shaping the length or profile of an electrical waveform applied to a thermal or mechanical transducer in the ink chamber.

  The operating conditions that create an effective pressure wave for the purpose of ink ejection are the physical dimensions of the ink chamber and the nozzle, the viscosity of the ink, the compressibility, the specific mass and the surface tension. They depend on parameters such as (surface tension), so they tend to be very narrow.

  As a result, many ink jet processes are optimized for printing with substantially fixed size ink droplets and corresponding density per pixel.

  The effect of droplets with varying volume by ejecting a varying number of microdroplets that coagulate before landing on a printable substrate at a high frequency. Is created. These droplets with varying volumes result in pixels with varying densities on the printable substrate.

  However, the tone resolution of such a printing system is usually too limited to render high quality graphics and continuous tone images.

  Accordingly, digital halftoning techniques have been developed to draw continuous tone images and graphics on devices with limited tone resolution.

  Digital halftoning is generally divided into two classes: amplitude modulated halftoning (also called clustered dot halftoning) and frequency modulation halftoning {disparse }, Which is also referred to as dedotted halftoning. An overview of the characteristics of both techniques can be found in Non-Patent Document 1.

  Clustered dot halftoning relies on a halftone dot geometry characterized by a fixed set of frequencies and angles. The tone values are modulated by changing the size of the halftone dots on the grid.

  Frequency modulation halftoning relies on changing the average distance between halftone dots of a fixed size in order to modulate the tone value.

  Hybrid digital halftoning also exists, where both halftone dot size or density and average distance are modulated.

  The term binary halftoning is halftoning adapted for devices that can only draw two tone levels per pixel, corresponding to printing the presence or absence of ink. Commonly used to call halftoning techniques.

  The term multilevel halftoning can draw more than two tone levels per pixel corresponding to printing at different densities of ink or different sizes of ink spots. Commonly used to call halftoning technology adapted for devices. These devices are called multi-level printing devices.

  Inkjet printing inherently reconstructs images using dispersed dots, and frequency modulated halftoning techniques-at a given resolution-can produce better image quality than amplitude modulation, so both For this reason, frequency modulation has been the preferred digital halftoning method for inkjet printing.

  The class of frequency modulation halftoning techniques themselves further includes a first subclass that uses an algorithmic approach to transform continuous tone images into frequency modulated halftones—called error diffusion—for this purpose. It can be subdivided into a second subclass that uses pre-calculated threshold masks.

  A desirable characteristic of a frequency modulation halftoning technique is that it has a “blue noise” spectrum, that is, its power spectrum is less energy at lower frequencies than at higher frequencies. To create dot distributions with spectra. It is known that error diffusion algorithms have this desirable property and are therefore capable of producing excellent image quality halftone dot distributions.

  An improvement on the standard error diffusion process related to the present invention is described in US Pat. This patent describes a technique for optimizing the relative positions of dots printed with various colorants. The method achieves several technically advantageous effects. It produces a more uniform dot distribution not only in tints printed with a single colorant, but also in tints printed with multiple colorants. A second advantage is that the uniform dot distribution improves the drying characteristics of the colorant and allows better color saturation at highlights.

  Multilevel error diffusion techniques have been described in the art, for example, in US Pat.

  The disadvantage of the error diffusion algorithm is that it requires a relatively large amount of computation due to the complexity of the algorithm. For high speed image processing, this translates to the need to deploy specialized computing hardware that is expensive and time consuming to develop. Another problem is that in the case of multiple levels of halftoning, the error diffusion algorithm controls the modulation trade-off between the number of halftone dots and the density level of the halftone dots. It offers only a few options to do.

  Several other methods have been described that rely on pre-computed dot patterns or threshold masks and require lesser ad hoc computations for this purpose.

  U.S. Patent No. 6,057,037 describes the use of a set of, for example, 256 pre-computed bitmaps, each corresponding to a tone level ranging from 0 to 255 in a continuous tone image. Each bitmap represents a distribution of halftone dots that depict the corresponding tone level and is computed—using an iterative optimization process—to produce a power spectrum with minimal visible noise. In addition, since there is a correlation between the limited bitmaps, no phase shift interference is introduced into the tone gradient of the image. The bitmap is typically a square having a size of, for example, 256 * 256 and is replicated (ie, “tiled”) horizontally and vertically across a two-dimensional addressable printer space like a tile on the floor. Tiled "). The main drawback of this approach is that it requires a lot of memory. If the size of the bitmap is 256 * 256 pixels and a 256 bitmap is needed for as many tone levels, more than 2 Mbytes of memory must be provided. Another disadvantage of this method is that the bitmap is a standard raster image processor {standard Raster Image Processors {used for rendering PostScript R {PostScript (R)} of PDF {PDF (R)} material. RIP's cannot be used because these devices perform halftoning using a threshold mask.

  Patent Document 3 discloses a technique for creating a square threshold mask that is replicated horizontally and vertically so that a threshold is generated at each pixel position of a page for the purpose of frequency halftoning. . Such a square threshold mask is compatible with a standard raster image processor and requires only 1 byte per element. By comparing each pixel in the image with a corresponding threshold in the threshold mask, it is determined whether a halftone dot is printed at the corresponding position.

  The way the square threshold mask is calculated in the previous method depends on accumulating successive levels of the stochastic dot pattern on top of each other. Each upper level (Every Higher Level) uses a dot distribution obtained by adding an additional dot to the lower level (Lower Level). A blue noise filter is applied to the dot distribution using a Fourier transform to determine which dots are added. By comparing the original dot distribution with the filtered dot distribution, it is determined where the dots should be added in order to optimally preserve the blue noise characteristics.

  The threshold mask obtained in this way creates a halftone dot distribution with blue noise characteristics.

  Once the threshold mask is calculated using the previous method, frequency modulation halftoning of the image requires less computation than the error diffusion method for that purpose, and the invention of Patent Document 2 Requires less memory. However, the threshold mask calculation of this method is complex and long because it involves many transformations that go back and forth between the spatial domain and the frequency domain.

  U.S. Patent No. 6,057,051 describes an alternative technique for calculating a threshold mask. By applying void and cluster filters (eg, low pass filters) to the dot distribution, there is a local excess of the void (where there is a local shortage of dots in the distribution) and clusters (a local excess of dots in the distribution) Location) is identified. This information forms the basis for determining where to add or remove halftone dots for higher or lower density drawing. Just like the previous method, a continuous dot profile is accumulated to obtain a threshold mask. The calculation of the threshold mask for this method is also complex and long because the filtering operation in the spatial domain using void and cluster filters requires a lot of calculations.

  The paper of Non-Patent Document 2 discloses yet another technique for generating a frequency modulated halftone mask. Non-Patent Document 2 authors prefer that dispersed order dither matrix be separated as far apart as possible as “continuous thresholds are separated as far as possible”. It has said. The method presented by the author achieves this goal through the use of a “repulsive force field” generated by all thresholds already contained in the mask in the previous process.

  In Patent Document 5, the concept of repulsive force field is for short distance-it results in local clustering of dots-it generates a suction field force and for long distance- Resulting in the dispersion of the clusters-refined by using a function that generates a repulsive force field.

  U.S. Patent No. 6,057,051 describes the use of a distance function that evaluates various candidate positions to add or remove halftone dots. Just as in the previous method, a continuous dot profile is accumulated to obtain the threshold mask. The calculation of the threshold mask for this method is complex and long because the evaluation of many candidate positions requires many calculations.

  In Patent Document 7, a technique for frequency modulation of a color image using a threshold mask is described for printing with a large number of colorants, in which 2 is used in order to minimize a visual cost function. Two or more threshold masks are combined and designed. The visual cost function uses different weights for different colorants. The method suggests the use of combinatorial optimization techniques combined with the viewing cost function to determine the optimal dot distribution from which the threshold mask is derived.

  U.S. Patent No. 6,057,052 describes a technique for reducing internal moire effects found in image amplitude modulation halftoning. According to the technique, the gray scale order of the original tile is modified by the addition of a bump function at each point in the original cell within the gray scale order. The The new gray scale order is stored in a new tile. Adding the bump function at a selected point in the original tile affects the selection of the next pixel in the grayscale. This method is specific for amplitude modulation halftoning and addresses the technical problem of internal moire that does not occur with frequency modulation halftoning. Furthermore, the method is specifically designed to improve the image quality of the current threshold matrix rather than it helps to calculate the original threshold matrix.

  A multilevel technique using a threshold mask is that technique, for example, described in US Pat. However, these techniques are mostly described in the context of periodic halftoning. Specific problems and solutions for probabilistic multi-level halftoning techniques are not addressed in prior art documents, however. One of these problems is the need to control the balance between changing the number of halftone dots and changing the density of halftone dots when reproducing different densities in the image.

  From the above analysis of the prior art, it can be seen that there is a need for an alternative and improved method of calculating a threshold matrix for the purpose of frequency modulation digital halftoning.

  In particular, there is a need for a method of creating a frequency modulation halftone tile that is efficient in terms of the number of calculations and produces a high image quality dot distribution with blue noise characteristics.

  There is also a need for an improved and alternative mask-based frequency modulation technique that optimizes the relative positions of halftone dots when printing with multiple colorants.

  In addition, there is a need for mask-based frequency modulation techniques that optimally support the multi-level capability of devices that can print at densities greater than two of the ink dot sizes.

US Pat. No. 6,637,851, Koen Van de Velde US Pat. No. 5,214,517, Sullivan et al. U.S. Pat. No. 5,111,310, Mitsa and Parker U.S. Pat. No. 5,535,020, Ulichney US Patent Application Publication No. US2003 / 210431 (A1) Specification US Pat. No. 6,833,933, Michael Woods US Pat. No. 5,822,451, Spalding U.S. Pat. No. 5,276,535, R.A. Levien US Pat. No. 5,903,713, Daels US Pat. No. 5,818,604 Robert Ulichney, Digital Halftoning, MIT press. Werner Purgafer, Robert F. Tobler and Manfred Geiler, Improved Threshold Matrices for Ordered Dithering (based on "Forced Random Dithering: Improved Threshold Matrices for Ordered Dithering", by the same authers and earlier published in the Proceeding of the First IEEE Conference on Image Processing, Nov.13- 16, 1994, Austin Texas, pp. 1032-1035)

  The invention is realized by a method having the specific features stated in claim 1. Specific features for preferred embodiments of the invention are set out in the dependent claims.

  Further advantages and embodiments of the present invention will become apparent from the accompanying description and drawings.

Printing
The method of the present invention is directed to use in ink jet printers, particularly drop-on-demand ink jet printers. The term printing, as used in the present invention, refers to the process of creating a structured pattern of ink markings on a substrate.

Ink
The ink is a conventional pigmented or dyed ink or colorant, but is also a wax, a water repellent substance, an adhesive or a plastic. Also good. Typically, the ink is not a pure compound, and each component has a specific function, such as dyes, pigments, surfactants, binders, fillers, solvents ( It is a complex mixture containing several components such as solvents, water and dispersants. The ink may also be a material whose viscosity or phase changes with temperature, such as wax. Also specifically mentioned are inks that polymerize under the influence of electromagnetic radiation, such as, for example, Ubuy (UV) light. This process is referred to as UV curing.

Substrate
The substrate is paper, but may be a textile, a synthetic foil or a metal plate or any other printable substrate known in the art. Examples include inkjet printing for creating a printing master for offset printing, printing on a cardboard or plastic package, printing decorative images {drop-on- Use of drop-on-demand and continuous}. Unprinted substrates are available on rolls or as separate, usually rectangular, sheet piles. The base is characterized by a width called substratewidth. When the base is on the roll, this width corresponds to the width of the roll. When the substrate is available as a rectangular sheet, this width corresponds to one of the two dimensions of the rectangular sheet, for example, the shortest dimension of said sheet.

System description (description of a system)
Single print head (single print head)
Referring to the embodiment shown in FIG. 1, the arrangement of transducers, ink chambers, and nozzles 130-etched into the nozzle plate-together constitute a print head 100. . The inkjet nozzles 130 operate in parallel to produce droplets having a fixed or variable volume.

Variable drop volume
According to one embodiment, the volume of the droplet is modulated by a variable number of micro-droplets. FIG. 2 shows the case where the number of microdroplets can vary between 0 and 7. Because the microdroplets are very small, air friction makes them subject to substantial deceleration, and as a result, the microdroplets land until they land on the substrate on which they are printed. In one single droplet having a variable volume. The volume of these droplets is called the number of micro droplets in which the droplets are formed. In particular, in the example shown in FIG. 2, the volume of a droplet having a variable size is called by an integer ranging from 0 to 7 (represented by the binary number in FIG. 2).

  Different drop volumes result in different sizes of ink spots and different measured densities on the printed substrate when printed.

  In the following, the tone resolution of a printing device refers to the number of density values that can be printed by a given device at a pixel location with one ink.

Printing device
According to the preferred embodiment and with reference to FIG. 3, the printing of the image of the material using the printer 300 is performed with an ink-receiving layer (ink-) for the page-wide printhead assembly 350 by a substrate transport mechanism. This is accomplished by moving a substrate 340 having a receiving layer 341 and selectively ejecting ink droplets onto the substrate in response to the image of the material.

  In FIG. 3, the infrastructure 340 rests on the infrastructure support 310.

  FIG. 3 shows a number of printing heads instead of one. In one embodiment, four print heads (360, 361, 362, 363) are used that print with cyan, magenta, yellow, and black inks. These heads are placed on a print head support (350). The ink reservoirs and tubes required to operate the printing device are not shown in FIG.

  In one embodiment, the ink spots on the substrate resulting from droplets ejected by various heads can be physically mixed before they cure. This technique of ejecting continuous droplets without intermediate curing is referred to as wet-on-wet printing.

  FIG. 3 also shows a curing station 370 that cures the printed ink after it has been printed onto the substrate. In one embodiment, the curing station is a UV source that emits energy having a spectrum that causes the ink to cure.

Slow and fast scan orientation and direction
Referring to FIG. 3, the orientation corresponding to the motion of the substrate relative to the print head is referred to as the fast scan orientation. The direction 330 in which the substrate moves relative to the print head will be referred to as the fast scan direction.

  An orientation that is perpendicular to the fast scan orientation is called a slow scan orientation. Depending on the preferred embodiment and with reference to FIG. 3, the orientation of the array of nozzles belonging to the print head is parallel to the slow scan orientation.

  The invention also works with different practices, but in the text below, by convention, i-orientation means the fast scan orientation and j-orientation is the slow scan orientation. Means.

Addressable grid of pixels
The location on the substrate where the droplets are printed forms an addressable grid of pixels. By convention, we define a set of pixels on the addressable grid aligned with a fast scan orientation as rows of pixels, and a set of pixels aligned with a slow scan orientation as a column of pixels ( called columns of pixels). An addressable grid of pixels having a raster line parallel to the fast scan orientation is also called a raster. In general, a raster line refers to a line of pixels parallel to a fast scan orientation and is indexed by [i] coordinates. The column of pixels is indexed with [j] coordinates. Pixels in the addressable grid are therefore addressed using a set of [i, j] coordinates.

Array of nozzles
Referring to FIG. 1, each nozzle of the printhead can be indexed by a nozzle index nozzleIndex ranging from 1 to nbrNozzels. The shortest distance between the two nozzles 130 along the slow scan orientation 140 is called the nozzle pitch 110 and is indicated by the parameter nozzlePitch. The length 150 of the nozzle array is expressed as a multiple of the length of slowScanPitch and is indicated by the parameter headSize.

  The set of pixels that can be addressed by the nozzles of a print head during one movement along the fast scan orientation is called a swath. The swath size is measured in an orientation parallel to the slow scan orientation and is referred to as swathSize. In a preferred embodiment, the swath size is greater than or equal to the image of the material so that all pixels of the material can be printed within one swath.

  Referring to FIG. 4, the array of nozzles 430 is zigzag along two or more rows (460, 461) for manufacturing reasons. In that case, the nozzle pitch 410 is defined as the shortest distance between two lines passing through the center of the zigzag nozzle perpendicular to the slow scan orientation.

  If the printhead has a zigzag array of nozzles, the timing of droplet firing from nozzles belonging to different rows is such that pixels belonging to the same line parallel to the slow scan orientation in the image of the material Adjustments are preferably made to land again on the same line parallel to the slow scan orientation on the printed image. By adjusting the timing in this way, the process of preparing the signal for the nozzle is essentially the same as all nozzles are on one and the same line (can be the same as if all the nozzles we virtually on one and the same line).

  In a preferred embodiment of the present invention, it is preferable to use a partial combination of two heads placed back to back rather than one head. Each individual printhead comprises a row of 318 nozzles with a nozzle pitch of 169 μm, which corresponds to 59.2 nozzles per cm or 150.3 nozzles per inch. The transducer in the printhead is of electromechanical type and is made using piezoelectric materials.

  Referring to FIG. 5, the second head displaces the second head 520 by exactly half the nozzle pitch 541 with respect to the first head 521, so that a subset that can be printed at twice the resolution of the original head. A sub-assembly 530 is obtained. The subassembly thus has two rows of 318 nozzles and is effectively equivalent to one printhead with a nozzle pitch of 85.5 μm, which pitch is 118.4 nozzles per cm or 1 This corresponds to 300.6 nozzles per inch.

  Depending on the preferred embodiment, and with reference to FIG. 5, the swath size is increased by zigzag placement of printhead subassemblies 530 to form printhead assembly 500. The subassembly is at a distance along the slow scan orientation such that any nozzle location is essentially separated from any other nozzle location by a distance along the slow scan orientation by a distance that is an integral multiple of the nozzle pitch 540. Moved. This is also indicated by the dotted grid parallel to the fast scan orientation of FIG.

  The subassembly 530 may or may not have overlapping nozzles. The timing of droplet ejection from nozzles belonging to different printhead subassemblies is the same for pixels belonging to the same line parallel to the slow scan orientation in the document image, parallel to the slow scan orientation on the printed image It is preferably adjusted to land on the line. By adjusting the timing in this way, the process of preparing the nozzle signal can be the same as if all nozzles were on the same line.

  According to the preferred embodiment, four prints are printed with different inks, such as with cyan, magenta, yellow and black inks, or with different tints and / or inks with different densities. A set of head assemblies is used.

  Also according to the preferred embodiment, nozzles belonging to different print head assemblies are arranged in a line along the fast scan orientation, so that nozzles belonging to different head assemblies but having the same nozzle index are the same at the same swath. Print on the raster line.

  There is a relationship between print resolution slowScanResolution along the slow scan orientation and nozzle pitch. In particular:

      slowScanResolution = 1 / nozzlePitch

  For a single printhead or printhead assembly, a constant speed of the substrate-said speed is represented by the parameter fastScanVelocity-for printing resolution in a fast scan orientation-the resolution is represented by the parameter fastScanResolution However,-is proportional to the firing frequency of the nozzle represented by the parameter firingFrequency. The parameter firingFrequency represents the time rate at which ink droplets are fired by the nozzle. Accordingly, the high-speed scanning resolution fastScanResolution is commanded by a ratio of the firing frequency firingFrequency divided by the speed fastScanVelocity in the high-speed scanning direction.

fastScanResolution
= FilingFrequency / fastScanVelocity

  Higher spatial resolution allows for the rendering of finer image details and thus generally allows higher image quality to be achieved.

Computer system
Depending on the preferred embodiment, and with reference to FIG. 6, a printer command is generated from a data processing system 600 such as a computer. The data processing system 600 includes network connection means 621, a central processing unit 622, and memory means 623 all connected through a computer bus 624. The data processing system 600 also typically includes a computer human interface 630, 631 for data input and a computer human interface 640 for data output. According to one embodiment, computer program code is stored on a computer readable medium such as a mass storage device 626 or a portable data carrier 650 that is read by portable data carrier reading means 625. Remembered.

Printer controller (printer controller)
Referring to FIG. 7, the substrate transport motor 740, the curing station 750 and the print head actuator 730 are controlled by the printer controller 700. The printer command 710 is received by the buffer memory 701. These printer commands have controller information sent to the control means 705 and image data sent to the image buffer 702. The image processor 703 can operate on the image data in the image buffer 702. The image processor 703 has access to one or more threshold masks 711 for halftoning purposes. The control means 705 controls a base transport driver 707 that drives the base transport motor 740. The control means 705 also controls a curing station driver 706 that drives the curing station 750. Information in the image buffer 702 can be used by a printhead driver 704 to drive the printhead actuator 730.

Halftoning processor
Binary Halftoning Processor According to one embodiment of the present invention, the image processor 703 in the printer controller 700 includes a halftoning processor that converts a continuous tone image into a halftone image.

  According to a preferred embodiment, the halftoning processor uses one or more threshold masks 711.

  FIG. 8 shows a first embodiment of a halftone processor. A continuous tone image 820 having pixels P [i, j] 860 represented by 8 unsigned bits per pixel is a half represented by 1 (unsigned) bits {1 (unsigned) bits} per pixel. Converted to a halftone image 830 having toned pixels H [i, j] 850. This is accomplished using a threshold mask 810 having a threshold S [i, j] 880 that is preferably represented on a scale of 1 to 255.

  Address generator 840 generates address [i, j] of halftone image 830. These addresses [i, j] are converted into converted addresses [i ′, j ′] by the address converter 870. These address translations may include position, scaling and rotational coordinate transformations. The address [i ′, j ′] is used to address the continuous tone pixel P [i ′, j ′] 860 of the image 820.

The continuous tone pixel value P [i′j ′] is multiplied by the integer coefficient K in the multiplier 861 to obtain a value K ^* P [i′j ′]. In the current embodiment, K is preferably equal to 1.

  The address [i, j] is also converted into a modulo converted address [i ″, j ″] by a modulo converter 871. These modulo translated addresses [i, j] address the threshold S [i ″, j ″] 880 in the threshold mask 810 having an iSize to jSize threshold. According to one embodiment, using a computer-like code:

i ″ = i modulo iSize;
j ″ = j modulo jSize;

The conversion of the continuous tone pixel P [i ′, j ′] 860 to the halftone pixel H [i, j] 850 is performed by applying a threshold S to the multiplied continuous tone pixel K ^* P [i ′, j ′] 860. [I ″, j ″] 880 is added, and the result K ^* P [i ′, j ′] + S [i ″, j ″] is quantized by the quantizer 891. According to this embodiment, the quantization is performed by the most significant bit of a 9-bit word representing the sum P [i ′, j ′] + S [i ″, j ″]. ). This most significant bit is convenient because it can be separated by shifting the result of the sum P [i ′, j ′] + S [i ″, j ″] right eight times. Using a computer-like code, the operation is expressed as follows:

      H [i, j] = (P [i ′, j ′] + S [i ″, j ″]) >> 8

  The operation of the halftoned image processor is now illustrated by FIG. 9, which has an image with pixel P [i ′, j ′] and a threshold S [i ″, j ″]. A one-dimensional sectional view of a threshold mask is shown. The image and threshold mask values are really discrete samples in space with discrete values, but for clarity, the image and threshold mask values are drawn as continuous lines. . Also for clarity, the threshold mask values shown represent periodic screening as opposed to frequency modulation screening. However, as in one of the other embodiments of the present invention, the halftoning processor also operates when a threshold mask calculated specifically for the purpose of frequency modulation halftoning is used. This will be apparent to those skilled in the art.

  FIG. 9 shows that the value P [i ″, j ″] corresponds to the spatially averaged density with the pixel value P [i ′, j ′] by the halftone processor. This is to be converted into a binary pattern of the halftoned pixel H [i, j].

  The embodiment shown in FIG. 9 corresponds to a halftoning process for a printer that can draw two density values that are binary coded as 0 and 1. There are also variations of the method that work for printers that can draw more than two density levels.

Multilevel halftoning processor—first embodiment FIG. 10 can draw four halftone values H [i, j] having values 0, 1, 2 or 3 instead of 2 1 shows a first embodiment of a halftone processor for a printer.

According to this embodiment, K is set to a value of 3 and the quantizer is the two most significant 10 bit words representing the sum 3 ^* P [i ′, j ′] + S [i ″, j ″]. It is preferably programmed to calculate the upper bits.

  In a code like a computer, this can be obtained, for example, by the following calculation.

H [i, j]
= {(0B11.100.0000) & (K ^* P [i, j] + S [i, j])} >> 8

  The calculation of this embodiment is also explained with reference to FIG. 10, which shows that the continuous tone value of the pixel P [i ′, j ′] is the pixel value P [i ', J'] and corresponding to a halftone pixel having a variable density (variable density).

The above example was given for a system capable of reproducing 4 density values. If the number of density values is, for example, M {M is an integer power of 2}, K is preferably set to M-1, and the quantizer is a sum K ^* P [i Preferably, it is programmed to calculate the most significant bit of log (M) / log (2) of ', j'] + S [i ″, j ″].

  A special case of the above embodiment is when K = 1. In that case, the multilevel screening can be attributed to the binary screening method of the above embodiment.

Multilevel halftoning processor-second embodiment Another embodiment of a multilevel halftoning processor is shown in FIG.

  According to this embodiment, multiple threshold masks are used instead of one. With particular reference to FIG. 11, three threshold values corresponding to threshold values S1 [i ″, j ″], S2 [i ″, j ″] and S3 [i ″, j ″]. A set of masks is used.

  The values H [i, j] generated for the halftone pixels are the three threshold values S1 [i ″, j ″], S2 [i ″, j ″] in the threshold mask. And depending on the value of pixel P [i, j] relative to the value in S3 [i ″, j ″].

  Referring to the embodiment of FIG. 12, using a computer-like language such as computer code, the calculation of H [i, j] proceeds as follows.

  First, an intermediate set of binary halftone values H1 [i, j], H2 [i, j] and H3 [i, j] is calculated using the following set of equations:

H3 [i, j]
= {(0B1.0000.0000) & (P [i, j] + S3 [i, j])} >> 8
H2 [i, j]
= {(0B1.0000.0000) & (P [i, j] + S2 [i, j])} >> 8
H1 [i, j]
= {(0B1.0000.0000) & (P [i, j] + S1 [i, j])} >> 8

  These equations are the same as those used in the previous embodiment, and the pixel value P [i ′, j ′] and the threshold values S1 [i ″, j ″], S2 [i ″, j ″] and S3 [i ″, j ″] with the aim of calculating the most significant bits H1 [i, j], H2 [i, j] and H3 [i, j] ing.

  A lookup table is then used to calculate H [i, j] based on the possible results of H1 [i, j], H2 [i, j] and H3 [i, j].

  The embodiment shown in FIG. 11 can be extended straight to the case of a printer with more or less printable density values per pixel.

  For example, if halftones are to be calculated for a printer that can draw 6 densities per pixel, then a set of 5 threshold masks is used, and a set of 5 intermediate binary halftone values is multi-level half The following table is useful for converting to tone values.

Halftone processor—implementation
According to one embodiment, the halftoning processor uses a software code that runs on a general purpose data processing system, such as the system shown in FIG. Realized.

  According to another embodiment, the halftone processor is implemented in a system embedded using, for example, ASIC or FPGA.

Description of embodiment to calculate threshold thresholds masks
1 Calculation of a threshold threshold mask (calculation of a single threshold mask)
A preferred embodiment of a method for obtaining a threshold mask that achieves the objectives of the present invention is described in the following paragraph, which is supported by the pseudo-programming language in the flow diagrams of FIGS. 12A and 12B. Yes.

Size of the threshold matrix (step 1210 of FIG. 12A)
The first step 1210 consists of defining the integer size of the threshold mask to be calculated. The threshold mask can be square or rectangular. If the frequency modulation technique is used for printers having the same resolution of iRes in i-orientation and jRes in j-orientation, the threshold mask is preferably square, Having a size of iMatrixSize in the i-orientation and jMatrixSize in the j-orientation. According to a first preferred embodiment, the mask size is a power of two. Good results can be obtained with a threshold mask having a size of 256 * 256 thresholds. According to the second embodiment, the mask size is a prime number, for example 491. Larger matrices produce slightly better results, but require more memory and more computation. If the print resolutions iRes and jRes are different, the mask is preferably rectangular and its size is preferably chosen to be iMatrixSize ^* iRes = jMatrixxSize ^* jRes.

Declaring two metrics (step 1220 of FIG. 12A)
According to a preferred embodiment, two matrices are declared in step 1220, both having a size of iMatrixSize * jMatrixSize and called rasterMountian and orderNumberMatrix.

  The matrix rasterMountain preferably has floating point values. It is preferably initialized with a small random number, such as a number generated by a white noise random generator. The number can be, for example, in the range between 0.0 and 10.0E-2. The matrix orderNumberMatrix preferably has an integer value that is preferably initialized to zero, although other initial values are possible.

Defining a single-valued function having a two-dimensional domain (process 1230 in FIG. 12A)
According to a preferred embodiment, the next step 1230 has a definition of a monovalent function z = f (x, y) having its domains in the i and j orientations. The domain of the function is preferably limited, and is preferably rectangular or elliptical.

  Various examples of domains are shown in FIGS. 14A, 14B, 14C and 14D.

  In the case of an elliptic domain (FIGS. 14A and 14B), the size of the domain is iDomainSize 1401 along a first axis parallel to the i-orientation and jDomainSize 1402 along a second axis parallel to the j-orientation, It is prescribed. In one embodiment, the ellipse is rotated at an angle α 1403 with respect to the i- and j-orientations. According to the preferred embodiment, this angle α is equal to zero degrees, and the long and short axes of the ellipse are parallel to the i-orientation and j-orientation.

  Preferably, iDomainSize is odd and greater than iMatrixSize, and preferably jDomainSize is also odd and greater than jMatrixSize, for example, twice the corresponding matrix size.

  In the case of a rectangular domain (FIGS. 14C and 14D), the size of the domain is determined by iDomainSize 1401 along a first axis parallel to the i-orientation of the rectangle and to a second axis parallel to the j-orientation of the rectangle. And is defined by jDomainSize1402. In one embodiment, the rectangle is rotated at an angle α 1403 relative to the i-orientation and j-orientation. According to a preferred embodiment, this angle α is equal to 0 degrees and the sides of the rectangle are parallel to the i-orientation and j-orientation.

  Preferably, iDomainSize is odd and greater than iMatrixSize, and preferably jDomainSize is also odd and greater than jMatrixSize, for example, twice the corresponding matrix size plus one.

  Depending on the preferred embodiment and with reference to FIG. 16, the monovalent function z = f (x, y) is preferably one at or near the origin (0,0) 1602 of the domain. It is preferable to have a maximum value fMax1601. The value should preferably gradually decrease away from the center and towards the domain boundary, preferably to zero. Near the origin is defined as being closer to the origin than 25% of the length of the shortest axis of the domain.

  Preferably, the monovalent function is symmetric with respect to the two principal axes of the domain. FIG. 16 shows an example of such a function. The monovalent function is preferably described by the profile P (d) 1603, which is from their maximum value fMax 1601 at the center of the domain, corresponding to d = 0.0, Explain how the function value decreases monotonically towards zero at the boundary 1604 of the domain where d = dMax. An example of a profile that can be used to define a monovalent function is:

Linear profile: P (d) = A ^* (1-x) + B; (see FIG. 15a)
Parabolic profile: P (d) = A ^* (1-x) ^* (1-x) + B ^* (1-x) + C;
Exponential function profile: P (d) = A ^* exp (B ^* x); (see FIG. 15c)
A profile defined by cubic interpolation between points (see FIG. 15b)
A profile defined by a Bezier segment defined by anchor points, where x = d / Max
D is an element from the interval [0.0, Max]
A, B, and C are constants.

  In accordance with a preferred embodiment of the present invention, an exponential profile was used for the exponential profile with A = 1.0 and B = -1 / 4 in the above equation.

  Another example of a profile is shown in FIGS. 15A, 15B and 15C. The present invention is not limited to these profiles, and there are many other profiles that produce satisfactory results.

Table (table representation) of function (function) z = f (x, y)
The above definition represents, for example, an explicit representation of the monovalent function z = f (i, j), as depicted in FIG. 13A. According to a preferred embodiment of the invention, a table version of the function z = x (i, j) is also calculated. According to a preferred embodiment, a rectangular matrix tableFunction with iDomainSize: * jDomainSize floating point elements is declared and initialized at step 1240 (FIG. 12).

  According to a preferred embodiment, the center element [(iDomainSize + 1) / 2, (jDomainSize + 1) / 2] of the matrix is the origin (0, 0) of the monovalent function z = f (x, y). ). The value of each component in the matrix tableFunction is preferably set to a number representing the volume integral of the function z = f (x, y) over the range of the corresponding matrix component. This is illustrated by FIG. 13B. The value of the component 1310 is set to a value representing the volume defined by the rectangular range 1320, the four ribs 1330, and the upper surface 1340. According to a preferred embodiment of the present invention, the value of the integral can be approximated using one of the techniques known in the art under the name of numerical integration or piecewise quadrature.

Description of the main loop of the computer implied method
According to a preferred embodiment, the next step 1250 (FIG. 12B) includes defining and declaring a loop counter loopCntr that should preferably be initialized with a value of 1. This loop counter is used to assign an order number to the components of the matrix orderNumberMatrix. The lowest order number is preferably 1, and the highest order number is preferably iMatrixSize ^* jMatrixSize. According to a preferred embodiment, one interpretation of the order number of a given position in the matrix orderNumberMatrix is that of the sequence it adds halftone dots to increase the tint density from zero to the maximum value. A dot indicates whether a halftone dot is added to the corresponding position of the order number.

  According to a preferred embodiment, the first step 1251 of the main loop has a global minimum value minRasterMountain, plus the coordinates of the component in the matrix rasterMountain whose corresponding component in orderNumberMatrix has not received an order number. Determining a set of [iMin, jMin]. Such determination of the global minimum is preferably performed using one of the numerical sorting techniques as is well known in the numerical calculation art.

  According to a preferred embodiment, the second step 1252 of the main loop comprises assigning the value of the loop counter loopCntr to the component having the position (iMin, jMin) in the matrix orderNumberMatrix.

  According to a preferred embodiment, the third step 1253 of the main loop includes the components of the matrix tableFunction such that the center of the matrix tableFunction corresponds to the global minimum position (iMin, jMin) in the rasterMountain. a process of shifting by an offset along the i and j coordinate axes. As part of the same process 1253, these components are added one-to-one to the components of the matrix rasterMountain.

  After the shift operation by the offset, if the position of the component of the matrix tableFunction is out of the position range of the matrix rasterMountain, the component is a rasterMountin that corresponds to a component in the repeater extension of rasterMounter. Is mapped onto the component of. The repeated extension of the matrix means that the components of the matrix are replicated repeatedly in a i-orientation over a distance corresponding to iDomainSize and in a j-orientation over a distance corresponding to jDomainSize. . In practice, this iterative extension of tableFunction is done on the [i, j] coordinate axis of the function using iDomainSize as the [i] coordinate modulo and jDomainSize as the [j] coordinate modulo. This is realized by applying a modulo operation.

  Addition of the components of the matrix tableFunction to the matrix rasterMountin results in an updated version of the matrix rasterMountain.

  According to a preferred embodiment, the third step 1254 of the main loop includes a step of incrementing the loop counter loopCntr of the main loop by one.

At the end of the main loop, the value of the loop counter loopCntr is checked. If the loop counter loopCntr is preferably equal to iMatrixSize ^* jMatrixSize, exit from the main loop, otherwise steps 1 through 4 of the main loop are repeated.

The method described above results in a matrix orderNumberMatrix with values ranging from 1 to iMatrixSize ^* jMatrixSize.

  According to a preferred embodiment of the present invention, step 1260 comprises the step of rescaling the values of the components of the matrix orderNumberMatrix to a tone range from 1 to preferably 65,536.

  According to a preferred embodiment, the process 1270 converts the value of the matrix orderNumberMatrix component expressed in the first higher number of bits to a value expressed in the second lower number of bits using the tone curve. The process of mapping. According to a preferred embodiment, the first higher number of bits is 16, and the second lower number of bits is 8.

  Methods for applying a tone to a threshold mask are known in the art and are described, for example, in US Pat. The result of step 1270 is a threshold mask that can be used for frequency modulation halftoning, for example by using the scheme shown in FIG.

  The presently presented method is effective depending on calculation, does not require conversion between frequency and spatial domain, and does not require a filter operation or multiple repetitions as in the conventional method. The method creates a halftone distribution with blue noise characteristics, but it is the average where the addition of a monovalent function to the components of rasterMountain adds a lower halftone dot position to increase the average density of the halftone dot distribution. This is because it has an effect that it is not close to the position of any dot that draws density.

  According to one embodiment, the method also includes the step of creating a printing plate precursor from the halftoned image (830). A printing plate precursor is an imaging material that can be used as a printing plate after one or more treatment steps including exposure as an image and possibly processing.

  According to another embodiment, the method also includes printing the halftoned image (830) on a substrate using an inkjet printing apparatus.

  The above description requires a threshold matrix for the purpose of frequency modulation halftoning, requiring fewer computations than prior art methods based on using void and cluster filter operations and transforming to the Fourier domain. Expresses how to calculate Instead, the method of the present invention uses a search algorithm (sorting) for the calculation of the position of each halftone dot, the addition of the components of the first matrix to the components of the second matrix and an efficient implementation therefor. (algorithm).

  The above method also results in a dot profile having desirable blue noise characteristics, where the addition of the matrix tableFunction to the matrix rasterMountain results in the position of the next minimum value in the matrix rasterMountine being the next minimum value. This is because it is designed to be moved away from the position of the previous minimum in the rasterMountain. This is also an ordering sequence in which halftone dots are placed for tints that become increasingly dark, and the position of the next halftone dot is preferably located at a distance from the position of the previous halftone dot. Means. As the result of the addition of the matrix tableFunction to the matrix rasterMount is accumulated during successive passes of the main loop, the position of the next halftone dot is preferably a distance from the previous position, but all It is also placed at a distance from the position of the halftone dot placed in front of. The use of the monovalent function z = f (x, y) of the method of the present invention thus has the effect of prohibiting the placement of halftone dots within the immediate area of the previously placed halftone dots. (Inhibiting effect). The monovalent function creates a repulsive force field that prohibits the placement of the next halftone dot in the range 30 in the immediate vicinity of the previously placed halftone dot. Thus, it counteracts the generation of halftone dot clusters and voids responsible for the power at low frequencies in the Fourier power spectrum. From this, the method of the present invention can create a dot distribution whose power spectrum has a blue noise characteristic.

Summary of Said Method to Calculate of a Single Threshold Mask (Abstract of Said Method to Calculate of a Single Threshold Mask)
Conceptually, the method described above for calculating a threshold mask for frequency modulation halftoning is abstracted as follows.
Initializing a force field having the same domain as that of the periodic extension of the threshold mask first;
-Initialize the order number;
Determining a position in the threshold mask corresponding to a local extremum of the force field, where the position has not been assigned an order number in the previous process;
-Assigning the order number to the position in the threshold mask;
-Adjusting the force field as a function of the position;
-Increment the order number;
-Repeat the above four steps until all positions in the threshold mask have been assigned an order number.

  While the invention will be described below in connection with preferred embodiments thereof, it will be understood that it is not intended to limit the invention to those embodiments.

  For example, equivalent results are obtained when the two-dimensional domain of the monovalent function is not limited, or if it has a different shape than the preferred embodiment rectangle or ellipse. There are alternative ways beyond the use of a function profile P (d) that defines the monovalent function z = f (x, y), which yields equivalent results. The use of the function z = f (x, y) that does not monotonously reduce the value away from the origin with the maximum value can result in an equivalent result. Equivalent results are also obtained when a monovalent function is used at or near its origin that does not have a global maximum but has a minimum. In that case, step 1251 of FIG. 12 is preferably replaced by finding a maximum value rather than a minimum value in the matrix rasterMountain.

Calculation of multiple threshold masks for color printing
When printing with multiple colorants, a set of threshold masks is used instead of one. For example, color image printing is typically performed by halftoning and printing of sub-images with cyan, magenta, yellow and black inks. According to one embodiment of the invention, a set of multiple threshold and distinct threshold masks is calculated using one of the previously described methods for calculating a single threshold mask. Is done.

  According to a preferred embodiment, the multiple threshold masks are calculated in such a way that the correlation between the dot distributions produced by the threshold masks is constrained.

  An embodiment of the present invention is illustrated by FIG. 20, which shows a main loop 2000 that is a further development of the main loop 1200 of FIG.

  In the main loop 2000, three variables called rasterMountainC, rasterMountainM, and rasterMountainY are used instead of one variable rasterMountain. These three variables correspond to the three sub-images cyan, magenta and yellow colors.

  In the main loop 200, three variables called orderNumberMatrixC, orderNumberMatrixM, and ordererNumberMatrixY are used instead of one variable orderNumberMatrix. These three variables correspond to the cyan, magenta and yellow colors of the three sub-images.

  These six matrices have the same size iMatrixSize * jMatrixSize and are declared and initialized in the previous step as in step 1200 of FIG. 12A.

  In accordance with an embodiment of the present invention and with reference to FIG. 20, the components of the matrix orderNumberMatrixC, orderNumberMatrixM, and orderNumberMatrixY are calculated in the same main loop 2000.

  The first process 2001 in the main loop has a global minimum value minasterMountine, and in addition, the corresponding components of the orderNumberMatrixC, orderNumberMatrixM and orderNumberMatrixY did not receive an order number [the component of the matrix rasterMountC] iMin, jMin] is determined.

  The second process in the main loop includes a process of assigning the value of the loop counter loopCntr to the component having the position [iMin, jMin] in the matrix orderNumberMatrixC.

  The third step of the main loop is to shift all the components of the matrix tableFunction by offsets along the i and j coordinate axes so that the center of the matrix tableFunction corresponds to the global minimum position [iMin, jMin] in rasterMountainC. Have a process to do. As part of the same process 2003, these components are added on a one-to-one basis to the components of the matrices rasterMountainC, rasterMountainM, and rasterMountainY.

The result of addition of the values of the components of the matrix tableFunction to the rasterMountainC, rasterMountainM, and rasterMountainY is the position of the next minimum value in the position where the position of the next minimum value in in any of the matrices rasterMountainC, rasterMounterM, and rasterMountY. It is moved away from. This is because the arrangement of the halftone dots at the position [iMin, jMin] in the cyan sub-image is within the closest range in any of the cyan, magenta, and yellow sub-images. It means having a forbidden effect on the arrangement of other halftone dots.

  The next steps 2011, 2012 and 2013 are the same as steps 2001, 2002 and 2003, and the same is true for the steps 2021, 2022 and 2023.

The main loop 2000 is executed iMatrixSize ^* jMatrixSize / 3 times. In that respect, the orderNumberMatrix component of the matrix rasterMountainC, rasterMounterM, and rasterMountainY takes iMatrixSize ^* jMatrixSize / 3 order number. The positions of these components in the three matrices are such that all of the components have different positions, and any such component position is common to the position of another component in the other of the two matrices. Complementary in the sense that it is not.

  The principles of the present invention described by the example shown in FIG. 20 can be generalized.

For example, an order number matrix (order numbermatrix), orderNumberMatrixC, orderNumberMatrixM, orderNumberMatrixY, and an element that belongs to the set of orderNumberMatrixK is an assignment from 1 to iMatrixSize ^* j.

Similarly, it is possible to calculate an allocation order number ranging from 1 to iMatrixSize ^* jMatrixSize / 2 for components belonging to a set of two order number matrices such as orderNumberMatrixC and orderNumberMatrixM.

Referring to FIG. 21, for example, components that belong to a set of two ordinal number matrices such as orderNumberMatrixC and orderNumberMatrixM are also assigned 1 + iMatrixSize ^* jMatrixSize / 3 to iMartizeSize ^* jMatrixSize / 2, and the number ranging from iMatrixSize ^* jMatrixSize / 2 Is possible. In that case, in each of these two ordinal number matrices, it is assumed that the component has already received an order number ranging from 1 to iMatrixSize ^* jMatrixSize / 3, and that the matrices rasterMountainC and rasterMountainM have been updated accordingly. ing. Any set of two matrices computed in loop 2000 of FIG. 20 can serve as a starting point for loop 2100 of FIG.

Many other variations are possible. The general condition is that at the end of the main loop 2000, 2100, the total number of components subjected to the order number should preferably be equal to iMatrixSize ^* jMatrixSize.

  FIG. 22 shows a first preferred embodiment of the present invention. We use the notations C, M, Y and K to refer to the respective order matrix.

In the first step, a set of four ordinal number matrices C, M, Y, and K receives an order number from 1 to iMatrixSize ^* jMatrixSize / 4 in one combined loop.

In the second step, the first set of components of the two ordinal number matrices C, M receives the order number from 1 + iMatrixSize ^* jMatrixSize / 4 to iMatrixSize ^* jMatrixSize / 2 in one combined loop. Also, in the second step, the second set of components of the two ordinal number matrices Y and K are subjected to an ordinal number from 1 + iMatrixSize ^* jMatrixSize / 4 to iMatrixSize ^* jMatrixSize / 2 in the third combined loop.

In the third step, each component of the order number matrix C, M, Y and K receives the order number from 1 + iMatrixSize ^* jMatrixSize / 2 to iMatrixSize ^* jMatrixSize in another loop.

  FIG. 23 shows another preferred embodiment.

In the first step, the components of the set of three order number matrices C, M, Y receive order numbers from 1 to iMatrixSize ^* jMatrixSize / 3 in one combined loop. In parallel, the fourth matrix K receives an order number from 1 to iMatrixSize ^* jMatrixSize / 3 in another loop.

In the second step, the first set of components of the two ordinal number matrices C and M are subjected to an ordered number from 1 + iMatrixSize ^* jMatrixSize / 3 to iMatrixSize ^* jMatrixSize / 2 in a combined loop. Also, in the second step, the second set of components of the two ordinal number matrices Y and K are subjected to an ordinal number from 1 + iMatrixSize ^* jMatrixSize / 3 to iMatrixSize ^* jMatrixSize / 2 in the third combined loop.

In the third step, each component of the order number matrix C, M, Y and K receives the order number from 1 + iMatrixSize ^* jMatrixSize / 2 to iMatrixSize ^* jMatrixSize in another loop.

Summary of a method to calculate of a set of threshold masks
Conceptually, the method described above for calculating a set of threshold masks for frequency modulation halftoning is based on the method described before calculating a single threshold mask using a force field, where Over at least part of the range, the assignment of ordinal numbers in different threshold masks is based on the same force field.

Calculation of a set of threshold masks for multilevel printing
In accordance with the preferred embodiment of the present invention, the set of child threshold masks is derived from a single threshold mask or an ordinal number matrix that operates as a parent matrix.

  FIG. 17 shows an example of a small threshold mask that operates as a parent matrix. In this case, the matrix size iMatrixSize 1702 along the i-orientation and the matrix size j-MatrixSize 1701 along the j-orientation are both equal to 3, so the parent matrix has nine thresholds. These thresholds are ordered from minimum to maximum and given an order number. The order number is assigned to each position in the threshold matrix. In FIG. 17, the values 1, 2, 3,. . . 9 indicates the order number of the thresholds in the series of thresholds in the parent matrix. The relationship between the order number and the corresponding threshold position is simply represented in Table 1 below.

  Instead, and in accordance with a preferred embodiment of the present invention, the order number of the series of thresholds is obtained directly from the matrix orderNumberMatrix calculated in the main loop 1200 in FIG. 12B.

  By way of example and for purposes of illustration only of the present invention, assume that the printing device is capable of regenerating 5 density levels (or equivalent dot area or droplet volume). In this example, the levels are coded as level 0, level 1, level 2, level 3 and level 4. However, the method works just as well at any other number of levels. In the general case, the number of printable levels is equal to the parameter nbrLevels ranging from level 0 to level nbrLevels-1.

  According to one embodiment. A two-dimensional table called an order number table having a level index x and a position index y can be created.

  The level index x in the table corresponds to the printable level of the printer. The order of the levels is arranged so that a larger level index preferably corresponds to printing at a higher density.

  The table position index y corresponds to the position in the parent matrix. The order of the positions is arranged so that the larger position index preferably corresponds to a larger threshold at the corresponding position in the threshold mask.

  Table 2 below shows an example in which the level index corresponds to the row index and the position index corresponds to the column index, corresponding to the threshold mask shown in FIG.

  According to one embodiment of the present invention, a monovalent function z = f (x, y) is used, where the first parameter (x) of the domain corresponds to the level index in the sequence number table. The second parameter (y) of the domain corresponds to the position index in the order number table.

According to a preferred embodiment, a function z = A ^* x + B ^* y is used in which A and B are constants. For example, A is 1 and B is 0.3: z = x + 0.3 ^* y.

According to a preferred embodiment of the present invention, the method comprises a process in which the function is used to populate the order number table. With reference to the example shown in FIG. 17, using the function z = x + 0.3 ^* y, this results in Table 3 below.

  According to a preferred embodiment of the present invention, the method is such that each function value F (x, y) in Table 3 corresponds to an ordered number of the function values F (x, y) in the series of function values. A unique order index 0 is assigned.

  For example, the position in Table 3 corresponding to level 0, position [0, 2] receives the smallest function value and is assigned the order index 1. Also, for example, the position in Table 2 corresponding to level 0, position [0, 1] receives the second smallest function value and is assigned the order index 2. This process of assigning order numbers is repeated until all positions in the order number table have received an order index.

  The result of the above process is an order number table filled with order indexes as shown in Table 4 below.

  According to a preferred embodiment of the present invention, the method comprises the step of converting the components of the sequence table obtained in the previous step into a threshold mask, which mask is for example shown in FIG. Can be used for multi-level halftoning using one such scheme described by.

  This transformation has two sub-steps. According to the first sub-process, the order numbers in the order number matrix are converted into an appropriate range of threshold values. For example, the order number range in Table 4 is from 1 to 45. For the purpose of digital halftoning a continuous tone image represented by 8 bits, the threshold range should preferably have a range from 1 to 255. The conversion of the range can be achieved in various ways. In one embodiment, the components of the ordinal number matrix are simply multiplied by an appropriate scale factor. In the example of the ordinal number table shown in Table 4, a suitable scale factor is 255/45. Multiplying all the components in Table 4 by the multiplication coefficients creates a new ordering table shown in Table 5.

  In another embodiment, a look-up table approach is used to map the range of values in the sequence table to the appropriate range of values for digital halftoning purposes. For example, FIG. 19 uses a mapping curve 1930 to represent the range of component values in the ordered number table represented on the axis 1910 and having the maximum value 1911, and the threshold value represented on the vertical axis 1920 and having the maximum value 1921. Map to the range of. In one embodiment, the mapping curve 1930 of FIG. 19 is not a linear mapping curve, but is designed to achieve a designed gradation of the halftoned image with the final printed result. A method of designing such a curve is described in Patent Document 10, for example.

  The result of this process is a converted ordinal number table.

  According to a preferred embodiment of the present invention, the method comprises the step of separating the transformed order number table into a set of nbrLevels threshold masks. This is accomplished by the process of assigning each component of the transformed sequence number table to the component having position [i, j] in the threshold mask, where the position [i, j] The threshold mask corresponds to the level index of the component in the transformed order number table, corresponding to the position index of each component in the converted order number table.

  18A to 18E show a set of threshold masks obtained by separating Table 5.

  The result of this process is a set of nbrLevels threshold masks that can be used for multi-level halftoning using the method described by FIG.

Abstraction of a method to calculate of a threshold masks for multiple level printing
While the invention will be described hereinafter in conjunction with its preferred embodiments, it will be understood that it is not intended to limit the invention to those embodiments.

For example, a different function z = f (i, j) may be used. In the above example, z = A ^* i + B ^* j is used with A = 1 and B = 0.3, but different values may be used.

  If B is greater than A, then for a higher density in the image, a frequency modulation half that prioritizes increasing the density of the already placed halftone dots before the additional halftone is printed Toning occurs. Table 6 below gives an ordinal number table corresponding to the case where B is much larger than A.

  If B is small compared to A, for higher densities in the image, frequency modulation halftoning prioritizing increasing the number of halftone dots printed before those densities are increased Happens. Table 7 gives an example of an ordinal number table corresponding to the case where B is much smaller than A.

  In general, by changing the parameters of the function z = f (i, j), the method of the present invention increases the number of halftone dots in order to increase the density of printed tints, It makes it possible to control the balance between increasing the density of the halftone dots.

Content summary
Having described the preferred embodiment of the present invention in detail, those skilled in the art will recognize that many other variations may be made therein without departing from the scope of the present invention as defined in the appended claims. it is obvious. For example, the present invention for generating a set of threshold masks for multi-level halftoning purposes has been described primarily in the context of a parent matrix that is a frequency modulated halftoned threshold mask. It can be used just as well when the matrix is an amplitude modulated halftoning threshold mask.

1 shows a printhead having nozzles. Fig. 4 illustrates the formation of an ink drop having a variable volume according to an embodiment of the present invention. 1 shows a printing system according to an embodiment of the present invention. 1 shows a printhead having multiple rows of nozzles. 1 shows a printhead assembly having multiple printheads. 1 shows a data processing system according to an embodiment of the present invention. 1 shows a printer controller according to an embodiment of the present invention. FIG. 6 illustrates an arrangement using a threshold array to convert a continuous tone image to a halftone image according to an embodiment of the present invention. FIG. The principle of converting a continuous tone image into a binary halftone image according to an embodiment of the present invention will be described. The principle of converting a continuous tone image into a multi-level halftone image according to the first embodiment of the present invention will be described. The principle of converting a continuous tone image to a multi-level halftone image according to a second embodiment of the present invention for calculating a single threshold mask will be described. 6 shows a flowchart illustrating a method according to an embodiment of the present invention. Fig. 4 shows a monovalent function z = f (x, y) having a two-dimensional (x, y) domain according to an embodiment of the present invention. FIG. 4 illustrates the use of a monovalent function z = f (x, y) to calculate a matrix representing a table function z = f (i, j) according to an embodiment of the present invention. Various embodiments of the present invention illustrate various shapes and orientations of the (x, y) domain of the monovalent function z = f (x, y). Fig. 4 illustrates various function profiles z = f (d) used to define a monovalent function z = f (x, y) according to various embodiments of the present invention. A preferred embodiment of the monovalent function z = f (x, y) is shown. An example of a parent matrix is shown according to an embodiment of the present invention. 18 shows an example of a set of child matrices derived from the parent matrix of FIG. 17 using one of the methods of the present invention. 4 shows a look-up table for converting values of an order number table according to an embodiment of the present invention. 6 shows a flowchart illustrating a method of an embodiment of the present invention for calculating a multi-threshold mask. 6 shows a flowchart illustrating a method of an embodiment of the present invention for calculating a multiple threshold mask. Various processes for calculating a multi-threshold mask according to an embodiment of the present invention will be described. Various processes for calculating a multi-threshold mask according to an embodiment of the present invention will be described.

Claims (7)

N-1 (N> 2) suitable for converting a matrix having the value M [i, j] at position [i, j] into a halftoned image comprising pixels that can have N density levels from a continuous tone image. ) Of threshold masks T [k] (k = 1,... N−1), wherein the method—in response to the values M [i, j], the matrix Assigning a first order number P [i, j] to each position of
Assigning a set of (N−1) levels L [k] (k = 1,... N−1) to each position of the matrix, the method further comprising: a function value F ( L [k], P [i, j]) to obtain the order number P [i, j] in combination with the level L [k]
Ordering the function values to obtain a second order number Q [k] [i, j];
-(N-1) the second order number Q [k] [i, j] to generate a threshold mask T [k] [i, j] for the set of threshold masks T [k]. And the method of scaling. In addition,
The method of claim 1, comprising using the set of threshold masks to convert a continuous tone image to a halftone image. In addition,
The method of claim 1, comprising the step of printing the halftone image. 4. An image processing system for converting a continuous tone image into a halftone image having N possible density levels, wherein the image processing system is generated using the process steps of any of claims 1-3. The system comprising a set of -1 threshold masks. A system for printing on a printable substrate, the printing system comprising an image processing system;
The printing system is:
A system for printing, wherein the image processing system is the image processing system of claim 4. A data processing system, wherein the data processing system is configured to perform the steps of the method of any of claims 1 to 3. A computer readable medium, wherein the computer readable medium comprises:
A medium comprising computer program code adapted to perform the method of any of claims 1 to 3 when executed on a computer.

Images