JP2005318543A - Image processing method, device and program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide half-toning process technology for attaining reduction of the processing amount and parallel operation, while making it possible to reproduce images with high quality. <P>SOLUTION: The image processing method includes a grouping process for separating pixels, which constitute a digital image, into a first group, in which an error generated by quantization is dispersed to an ambient pixel and a second group, in which an error generated by the quantization is not dispersed to the ambient pixel; a first quantization process for quantizing the pixel(pixel position A), belonging to the first group by using a threshold matrix; an error calculating process for seeking a quantization error generated at the first quantization process; an error dispersion process for dispersing the quantization error, which was obtained at the error calculating process, to at least one non-quantized pixel(pixel position B) adjoining to the pixel, in relation to the quantization process and belonging to the second group; and a second quantization process for quantizing the non-quantized pixel, to which the error value was added in the error diffusion process by using the threshold matrix. <P>COPYRIGHT: (C)2006,JPO&NCIPI

Description

  The present invention relates to an image processing method and apparatus and a program for realizing an image processing function by a computer, and more particularly to a halftone imaging technique suitable for an ink jet recording apparatus and other image forming apparatuses.

  In general, in an ink jet recording apparatus, gradation of an image is converted into an appropriate dot pattern by using a halftoning technique typified by an error diffusion method or a dither method (Patent Documents 1 to 3). etc).

Patent Document 1 proposes a method of quantitatively evaluating the conspicuousness of dots, quantizing the conspicuous dots using an error diffusion method, and quantizing the other dots using a systematic dither method. Patent Document 2 proposes a halftone processing method that sequentially repeats a process of diffusing an error generated by quantization using a threshold matrix to surrounding unprocessed pixels. Patent Document 3 proposes a method in which gray value images are grouped in units of rows of pixels (pixels), and a process of sequentially transferring a quantization error of a pixel of interest in a row to an adjacent pixel in the next row is proposed.
Japanese Patent Laid-Open No. 2001-157055 JP 2003-110852 A Japanese Patent Laid-Open No. 7-15606

  As shown in Patent Documents 2 and 3, quantization processing for diffusing errors generated by quantization using a threshold matrix to surrounding pixels has an advantage that high-quality image reproduction is possible. The calculation amount is large because the error is sequentially diffused. In addition, parallel processing is difficult because of sequential processing, and it is difficult to realize high-speed processing.

  The present invention has been made in view of such circumstances, and an image processing method and apparatus capable of achieving a reduction in the amount of computation and parallel processing while enabling high-quality image reproduction when halftoning a gradation image. An object of the present invention is to provide a program for realizing the function by a computer.

  In order to achieve the above object, the invention according to claim 1 is an image processing method for halftoning a digital image constituted by an array of pixels having gradation values corresponding to image contents, wherein the digital processing is performed. A first group to which a pixel at a pixel position that diffuses a quantization error generated by quantization to surrounding pixels and a pixel that does not diffuse the quantization error generated by quantization to the surrounding pixels for the pixels constituting the image A grouping process for dividing the pixel at the position into a second group, a first quantization process for performing quantization on the pixels belonging to the first group using a threshold matrix, and the first quantum An error calculation step for obtaining a quantization error generated in the quantization step, and the quantization error obtained in the error calculation step is adjacent to a pixel related to the quantization process and belongs to the second group An error diffusion step of diffusing to at least one unquantized pixel, and a second quantization for performing quantization using a threshold matrix on the unquantized pixel to which an error value is given by the error diffusion step And a process.

  A digital image is handled as a set (array) of pixels, which is a minimum information unit, and each pixel is given a gradation value (a value representing the degree of shading) according to the image content. When the original image is halftoned and converted into lower gradation image data (dot data), in the present invention, a pixel (first first) that diffuses its own quantization error with respect to adjacent surrounding pixels. Quantization is performed with a value that takes into account errors given from adjacent pixels, and the quantization error is processed separately for pixels that are not diffused to other pixels (second group). .

  The pixels belonging to the first group are quantized by a threshold matrix without considering the quantization error of other surrounding pixels, and the quantization error generated by this quantization is the second group's adjacent quantization error. Diffused into pixels. For the pixels belonging to the second group, the pixel value is corrected (error correction) by the error diffused from the pixels of the first group, and then quantized with the threshold matrix, and the error caused by this quantization Does not diffuse to other pixels.

  By doing so, it is possible to greatly reduce the amount of calculation related to quantization. Further, since it is not sequential processing, parallel processing can be easily configured, and the processing speed can be increased. Furthermore, since the error caused by the quantization is taken into consideration, it is possible to reproduce a high-quality image as compared with the conventional threshold matrix method.

  A second aspect of the present invention relates to an aspect of the image processing method according to the first aspect, wherein the pixels belonging to the first group include a plurality of columns and a plurality of columns in a two-dimensional array of pixels constituting the digital image. It is characterized by being distributed in a substantially uniform manner over a plurality of rows.

  As a distribution form of the pixels belonging to the first group, an aspect in which the pixels are sparsely distributed with a certain degree of uniformity and regularity over the entire screen is preferable. Thereby, high image quality and parallel processing can be realized. In consideration of the error diffusion method, the inter-pixel distance between the pixels belonging to the first group is more preferably about 1 to 2 pixels.

  A third aspect of the present invention relates to an aspect of the image processing method according to the first or second aspect, wherein the threshold value matrix has a blue noise characteristic.

  In general, regarding the spatial frequency of an image, an improvement in image quality can be achieved by diffusing an error over a wide range in a low frequency region. In the present invention, it is preferable to use a threshold value matrix that provides good results for low and medium frequency regions because of limiting the error diffusion range. Since the blue noise mask is a matrix mainly considering the low and medium frequency regions, it is suitable for use in the first quantization step and the second quantization step of the present invention.

  A fourth aspect of the present invention relates to an aspect of the image processing method according to the first, second, or third aspect, wherein the error diffusion step belongs to the second group the quantization error obtained in the error calculation step. It is characterized in that it is applied at a predetermined ratio to unquantized pixels.

  As a form for diffusing the quantization error, for example, the pixel position of the error diffusion destination and the distribution ratio of the error to each pixel position are determined in advance, and the pixel position having the predetermined adjacent relationship is determined at a predetermined ratio. There is an aspect that gives an error. Note that in the case where there is only one diffusion destination pixel position, the error is shifted at a rate of 100% with respect to the diffusion destination pixel position.

  A fifth aspect of the present invention relates to an aspect of the image processing method according to the first, second, or third aspect, wherein the error diffusion step includes a pixel position and an error of a quantization error diffusion destination obtained in the error calculation step. An error distribution control step of controlling at least one of the distribution ratios is included.

  The form in which the quantization error is diffused may be determined in advance, and as shown in claim 5, at least one of the diffusion destination pixel position and the error distribution ratio may be appropriately controlled.

  The invention according to claim 6 relates to an aspect of the image processing method according to claim 5, wherein the error distribution control step randomly determines at least one of the pixel position of the diffusion destination and the error distribution ratio. It is characterized by.

  By randomizing the error distribution, it is possible to suppress unevenness of the pattern unique to the image.

  A seventh aspect of the invention relates to an aspect of the image processing method according to the fifth aspect of the invention, wherein the error distribution control step includes the diffusion destination according to a gradation value gradient of surrounding pixels of the pixel related to the quantization process. In this case, at least one of the pixel position and the error distribution ratio is controlled.

  For example, a mode in which the error diffusion in the direction perpendicular to the inclination direction is promoted and the error diffusion in the inclination direction is suppressed as the inclination of the surrounding pixels (gradient of the gradation value) increases. This improves edge reproduction.

  The invention according to an eighth aspect relates to an aspect of the image processing method according to the fifth or seventh aspect, wherein the error distribution control step performs the diffusion destination according to a frequency characteristic of surrounding pixels of the pixel related to the quantization processing. In this case, at least one of the pixel position and the error distribution ratio is controlled.

  For example, when the threshold matrix having the blue noise characteristics described above is used, there is an aspect in which error diffusion is suppressed as the high frequency component is smaller. The amount of calculation can be further reduced by actively using the blue noise characteristics.

  The invention according to claim 9 relates to an aspect of the image processing method according to any one of claims 1 to 8, and is a representative value of a pixel range including a target pixel belonging to the first group and its peripheral pixels. A representative value calculating step for obtaining the representative value, using the representative value obtained in the representative value calculating step as the value of the target pixel, performing quantization on the representative value using a threshold matrix, and performing the quantization process The result is the quantization result of the pixel of interest, and the error calculation step finds an error between the gradation value of the pixel of interest and the quantization result of the pixel of interest.

  When quantization is performed on a certain target pixel belonging to the first group, a threshold value matrix may be applied to the pixel value of the target pixel. It is also possible to apply a threshold value matrix by calculating a representative value for a plurality of included pixels and regarding the representative value as the pixel value of the target pixel. Since the representative value is obtained, the amount of calculation is slightly increased, but further improvement in image quality can be achieved.

  Note that the “peripheral pixels” that are the calculation range of the representative value may belong to the first group or may belong to the second group. Of course, the peripheral pixels belonging to the first group and the peripheral pixels belonging to the second group may be mixed in the peripheral pixel group in the calculation range.

  A tenth aspect of the present invention relates to an aspect of the image processing method according to any one of the first to ninth aspects, wherein when the digital image is a color image including a plurality of color components, the threshold value matrix is determined depending on a color. It is characterized by different sizes.

  Interference between colors can be reduced by using an appropriate threshold value matrix for each color in consideration of dot overlap between colors.

  According to an eleventh aspect of the present invention, there is provided the image processing method according to any one of the first to tenth aspects, wherein when the digital image is a color image including a plurality of color components, the first image is determined depending on a color. The pixel positions of the pixels belonging to the group are made different.

  For example, in the case of a color image described with color components of cyan (C), magenta (M), and yellow (Y), “Cyan” and “magenta” that are visually noticeable are pixels of the first group. Different pixel positions. Thereby, interference between colors can be reduced.

  A twelfth aspect of the present invention relates to an aspect of the image processing method according to any one of the first to eleventh aspects, wherein when the digital image is a color image including a plurality of color components, an error is diffused depending on colors. It is characterized by different methods.

  Interference between colors can be reduced by varying the error diffusion pattern (diffusion method) such as the pixel position of the error diffusion destination and the error distribution ratio.

  As in the tenth to twelfth aspects, by varying the quantization processing for each color, the amount of calculation is slightly increased, but the image quality can be further improved.

  A thirteenth aspect of the present invention relates to an aspect of the image processing method according to any one of the first to ninth aspects, wherein when the digital image is a color image including a plurality of color components, a plurality of colors of one pixel. It is characterized in that signals are collectively subjected to vector quantization processing.

  According to this aspect, the amount of calculation can be further reduced.

  The invention according to claim 14 provides an apparatus embodying the method invention. That is, the image processing apparatus according to claim 14 is an image processing apparatus that performs halftoning of a digital image configured by an array of pixels having gradation values according to image content, and configures the digital image. A first group to which a pixel at a pixel position that diffuses a quantization error generated by quantization to surrounding pixels belongs, and a pixel at a pixel position that does not diffuse a quantization error generated by quantization to the surrounding pixels. Generated by grouping means for dividing into a second group belonging to the first group, first quantization means for performing quantization using a threshold matrix on pixels belonging to the first group, and the first quantization means An error calculating means for obtaining the quantized error, and a quantization error obtained by the error calculating means adjacent to a pixel related to the quantization process and belonging to the second group. Error diffusion means for diffusing to one unquantized pixel, and second quantization for performing quantization using a threshold matrix on the unquantized pixel to which an error value is given by the error diffusion means Means.

  A fifteenth aspect of the present invention provides an image forming apparatus including the image processing apparatus according to the fourteenth aspect, wherein the image forming apparatus includes a discharge head in which nozzles for discharging liquid droplets are formed; An ejection control means for controlling ejection from the ejection head based on halftone image data obtained by an image processing apparatus, and forming an image on a recording medium by droplets ejected from the nozzle To do.

  As a configuration example of the ejection head, a full-line type inkjet head having a nozzle row in which a plurality of nozzles for ejecting ink are arranged over a length corresponding to the entire width of the recording medium can be used.

  In this case, a combination of a plurality of relatively short ejection head blocks having nozzle rows that are less than the length corresponding to the entire width of the recording medium, and connecting them together, the nozzles having a length corresponding to the entire width of the recording medium as a whole There is an aspect that constitutes a column.

  A full-line type inkjet head is usually arranged along a direction perpendicular to the relative feeding direction (relative conveyance direction) of the recording medium, but at a certain angle with respect to the direction perpendicular to the conveyance direction. There may also be a mode in which the inkjet head is arranged along an oblique direction with a gap.

  A “recording medium” is a medium that can record an image by the action of an ejection head (a medium that can be called an ejection medium, a printing medium, an image forming medium, a recording medium, an image receiving medium, etc.), continuous paper, cut paper It includes various media regardless of materials and shapes, such as a sealing sheet, a resin sheet such as an OHP sheet, a film, a cloth, a printed circuit board on which a wiring pattern is formed by an ejection head, an intermediate transfer medium, and the like.

  The conveying means for relatively moving the recording medium and the ejection head is an aspect for conveying the recording medium to the stopped (fixed) ejection head, an aspect for moving the ejection head relative to the stopped recording medium, or Any of the modes in which both the ejection head and the recording medium are moved is included.

  The invention described in claim 16 provides a program for realizing the functions of the image processing apparatus described in claim 14 by a computer.

  That is, the image processing apparatus according to claim 14 can be realized by a computer, and the invention according to claim 16 is a program (or claim) for realizing each means of the above-described image processing apparatus by a computer. Item 14. A program for causing a computer to execute each step of the image processing method according to any one of Items 1 to 13 is provided.

  The image processing program according to the present invention can be applied as an operation program of a central processing unit (CPU) incorporated in an image output device such as a printer or a display device, and can also be applied to a computer system such as a personal computer. . Further, the image processing program according to the present invention may be configured as a single application software, or may be incorporated as a part of an application such as driver software or image editing software.

  According to the present invention, quantization is performed using a threshold matrix without performing error correction, pixels (first group) that diffuse the error generated by the quantization to the surroundings, and error correction using the diffused error is performed. Then, quantization is performed, and pixels (second group) that do not diffuse the error caused by the quantization to the surroundings are set, so that the amount of calculation related to quantization can be greatly reduced as compared with the prior art.

  In addition, since the halftoning process according to the present invention is not sequential processing, parallel processing is possible, and high-speed processing can be realized. Furthermore, since the present invention takes into account errors caused by quantization for some pixels, a higher quality image can be obtained as compared to the conventional threshold matrix method.

  Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the accompanying drawings.

[Overall configuration of inkjet recording apparatus]
FIG. 1 is an overall configuration diagram of an ink jet recording apparatus using an image processing apparatus according to an embodiment of the present invention. As shown in the figure, the ink jet recording apparatus 10 includes a plurality of ink jet heads (hereinafter referred to as “ink jet heads”) corresponding to black (K), cyan (C), magenta (M), and yellow (Y) inks. A printing unit 12 having 12K, 12C, 12M, and 12Y, an ink storage / loading unit 14 that stores ink to be supplied to each of the heads 12K, 12C, 12M, and 12Y, and a recording sheet as a recording medium 16 is disposed opposite to the decurling unit 20 for removing the curl of the recording paper 16 and the nozzle surface (ink ejection surface) of the printing unit 12 to improve the flatness of the recording paper 16. A suction belt conveyance unit 22 that conveys the recording paper 16 while holding it, a print detection unit 24 that reads a printing result by the printing unit 12, and a discharge that discharges the printed recording paper (printed matter) to the outside. And parts 26, and a.

  The ink storage / loading unit 14 has an ink tank that stores ink of a color corresponding to each of the heads 12K, 12C, 12M, and 12Y, and each tank has a head 12K, 12C, 12M, and 12Y through a required pipe line. Communicated with. Further, the ink storage / loading unit 14 includes notifying means (display means, warning sound generating means) for notifying when the ink remaining amount is low, and has a mechanism for preventing erroneous loading between colors. ing.

  In FIG. 1, a magazine for rolled paper (continuous paper) is shown as an example of the paper supply unit 18, but a plurality of magazines having different paper widths, paper quality, and the like may be provided side by side. Further, instead of the roll paper magazine or in combination therewith, the paper may be supplied by a cassette in which cut papers are stacked and loaded.

  When multiple types of recording paper are used, an information recording body such as a barcode or wireless tag that records paper type information is attached to the magazine, and the information on the information recording body is read by a predetermined reader. Thus, it is preferable to automatically determine the type of recording medium (media type) to be used and perform ink ejection control so as to realize appropriate ink ejection according to the media type.

  The recording paper 16 delivered from the paper supply unit 18 retains curl due to having been loaded in the magazine. In order to remove this curl, heat is applied to the recording paper 16 by the heating drum 30 in the direction opposite to the curl direction of the magazine in the decurling unit 20. At this time, it is more preferable to control the heating temperature so that the printed surface is slightly curled outward.

  In the case of an apparatus configuration that uses roll paper, a cutter (first cutter) 28 is provided as shown in FIG. 1, and the roll paper is cut into a desired size by the cutter 28. The cutter 28 includes a fixed blade 28A having a length equal to or greater than the conveyance path width of the recording paper 16, and a round blade 28B that moves along the fixed blade 28A. The fixed blade 28A is provided on the back side of the print. The round blade 28B is disposed on the printing surface side with the conveyance path interposed therebetween. Note that the cutter 28 is not necessary when cut paper is used.

  After the decurling process, the cut recording paper 16 is sent to the suction belt conveyance unit 22. The suction belt conveyance unit 22 has a structure in which an endless belt 33 is wound between rollers 31 and 32, and at least portions facing the nozzle surface of the printing unit 12 and the sensor surface of the printing detection unit 24 are horizontal ( Flat surface).

  The belt 33 has a width that is wider than the width of the recording paper 16, and a plurality of suction holes (not shown) are formed on the belt surface. As shown in FIG. 1, a suction chamber 34 is provided at a position facing the nozzle surface of the print unit 12 and the sensor surface of the print detection unit 24 inside the belt 33 spanned between the rollers 31 and 32. The recording paper 16 is sucked and held on the belt 33 by sucking the suction chamber 34 with a fan 35 to a negative pressure.

  When the power of the motor (reference numeral 88 in FIG. 7) is transmitted to at least one of the rollers 31 and 32 around which the belt 33 is wound, the belt 33 is driven in the clockwise direction in FIG. The held recording paper 16 is conveyed from left to right in FIG.

  Since ink adheres to the belt 33 when a borderless print or the like is printed, the belt cleaning unit 36 is provided at a predetermined position outside the belt 33 (an appropriate position other than the print area). Although details of the configuration of the belt cleaning unit 36 are not shown, for example, there are a method of niping a brush roll, a water absorbing roll, etc., an air blow method of blowing clean air, or a combination thereof. In the case where the cleaning roll is nipped, the cleaning effect is great if the belt linear velocity and the roller linear velocity are changed.

  Although a mode using a roller / nip conveyance mechanism instead of the suction belt conveyance unit 22 is also conceivable, if the roller / nip conveyance is performed in the print area, the image easily spreads because the roller contacts the printing surface of the sheet immediately after printing. There is a problem. Therefore, as in this example, suction belt conveyance that does not bring the image surface into contact with each other in the print region is preferable.

  A heating fan 40 is provided on the upstream side of the printing unit 12 on the paper conveyance path formed by the suction belt conveyance unit 22. The heating fan 40 heats the recording paper 16 by blowing heated air onto the recording paper 16 before printing. Heating the recording paper 16 immediately before printing makes it easier for the ink to dry after landing.

  Each of the heads 12K, 12C, 12M, and 12Y of the printing unit 12 has a length corresponding to the maximum paper width of the recording paper 16 targeted by the inkjet recording apparatus 10, and the nozzle surface has a recording medium of the maximum size. This is a full-line type head in which a plurality of nozzles for ink discharge are arranged over a length exceeding at least one side (full width of the drawable range) (see FIG. 2).

  The heads 12K, 12C, 12M, and 12Y are arranged in the order of black (K), cyan (C), magenta (M), and yellow (Y) from the upstream side in the recording paper 16 feed direction. 12K, 12C, 12M, and 12Y are fixedly installed so as to extend along a direction substantially orthogonal to the conveyance direction of the recording paper 16.

  A color image can be formed on the recording paper 16 by discharging different color inks from the heads 12K, 12C, 12M, and 12Y while transporting the recording paper 16 by the suction belt transporting section 22.

  As described above, according to the configuration in which the full-line heads 12K, 12C, 12M, and 12Y having nozzle rows that cover the entire width of the paper are provided for each color, the recording paper 16 and the printing unit in the paper feeding direction (sub-scanning direction). The image can be recorded on the entire surface of the recording paper 16 by performing the operation of moving the 12 relatively once (that is, by one sub-scan). Thereby, it is possible to perform high-speed printing as compared with a shuttle type head in which the recording head reciprocates in a direction orthogonal to the paper transport direction, and productivity can be improved.

  In this example, the configuration of KCMY standard colors (four colors) is illustrated, but the combination of ink colors and the number of colors is not limited to this embodiment, and light ink and dark ink may be added as necessary. Good. For example, it is possible to add an ink jet head that discharges light ink such as light cyan and light magenta. Also, the arrangement order of the color heads is not particularly limited.

  The print detection unit 24 shown in FIG. 1 includes an image sensor for imaging the droplet ejection result of the printing unit 12, and means for checking nozzle clogging and other ejection defects from the droplet ejection image read by the image sensor. Function as.

  The print detection unit 24 of this example is composed of a line sensor having a light receiving element array that is wider than at least the ink ejection width (image recording width) by the heads 12K, 12C, 12M, and 12Y. The line sensor includes an R sensor row in which photoelectric conversion elements (pixels) provided with red (R) color filters are arranged in a line, a G sensor row provided with green (G) color filters, The color separation line CCD sensor is composed of a B sensor array provided with a blue (B) color filter. Instead of the line sensor, an area sensor in which the light receiving elements are two-dimensionally arranged can be used.

  A test pattern or practical image printed by the heads 12K, 12C, 12M, and 12Y of each color is read by the print detection unit 24, and ejection determination of each head is performed. The ejection determination includes the presence / absence of ejection, measurement of dot size, measurement of dot landing position, and the like.

  A post-drying unit 42 is provided following the print detection unit 24. The post-drying unit 42 is means for drying the printed image surface, and for example, a heating fan is used. Since it is preferable to avoid contact with the printing surface until the ink after printing is dried, a method of blowing hot air is preferred.

  When printing on porous paper with dye-based ink, the weather resistance of the image is improved by preventing contact with ozone or other substances that cause dye molecules to break by pressurizing the paper holes with pressure. There is an effect to.

  A heating / pressurizing unit 44 is provided following the post-drying unit 42. The heating / pressurizing unit 44 is a means for controlling the glossiness of the image surface, and pressurizes with a pressure roller 45 having a predetermined surface uneven shape while heating the image surface to transfer the uneven shape to the image surface. To do.

  The printed matter generated in this manner is outputted from the paper output unit 26. It is preferable that the original image to be printed (printed target image) and the test print are discharged separately. The ink jet recording apparatus 10 is provided with a sorting means (not shown) for switching the paper discharge path in order to select the print product of the main image and the print product of the test print and send them to the discharge units 26A and 26B. Yes. Note that when the main image and the test print are simultaneously formed in parallel on a large sheet, the test print portion is separated by a cutter (second cutter) 48. The cutter 48 is provided immediately before the paper discharge unit 26, and cuts the main image and the test print unit when the test print is performed on the image margin. The structure of the cutter 48 is the same as that of the first cutter 28 described above, and includes a fixed blade 48A and a round blade 48B.

  Although not shown in FIG. 1, the paper output unit 26A for the target prints is provided with a sorter for collecting prints according to print orders.

[Head structure]
Next, the structure of the head will be described. Since the structures of the respective heads 12K, 12C, 12M, and 12Y for each color are common, the heads are represented by the reference numeral 50 in the following.

  FIG. 3A is a plan perspective view showing an example of the structure of the head 50, and FIG. 3B is an enlarged view of a part thereof. 3C is a plan perspective view showing another structure example of the head 50, and FIG. 4 is a cross-sectional view showing a three-dimensional configuration of one droplet discharge element (an ink chamber unit corresponding to one nozzle 51). FIG. 4 is a sectional view taken along line 4-4 in FIG.

  In order to increase the dot pitch printed on the recording paper 16, it is necessary to increase the nozzle pitch in the head 50. As shown in FIGS. 3A and 3B, the head 50 of this example includes a plurality of ink chamber units including nozzles 51 serving as ink droplet ejection openings, pressure chambers 52 corresponding to the nozzles 51, and the like. It has a structure in which (droplet discharge elements) 53 are arranged in a zigzag matrix (two-dimensionally), and is thereby projected so as to be arranged along the head longitudinal direction (direction perpendicular to the paper feed direction). High density of substantial nozzle interval (projection nozzle pitch) is achieved.

  The configuration in which one or more nozzle rows are configured over a length corresponding to the entire width of the recording paper 16 in a direction substantially orthogonal to the feeding direction of the recording paper 16 is not limited to this example. For example, instead of the configuration of FIG. 3 (a), short head blocks 50 ′ in which a plurality of nozzles 51 are two-dimensionally arranged are arranged in a staggered manner and connected as shown in FIG. 3 (c). A line head having a nozzle row having a length corresponding to the entire width of the recording paper 16 may be configured.

  The pressure chamber 52 provided corresponding to each nozzle 51 has a substantially square planar shape (see FIGS. 3 (a) and 3 (b)), and is connected to the nozzle 51 at both corners on a diagonal line. An outlet and an inlet (supply port) 54 for supply ink are provided. The shape of the pressure chamber 52 is not limited to this example, and the planar shape may have various forms such as a quadrangle (rhombus, rectangle, etc.), a pentagon, a hexagon and other polygons, a circle, and an ellipse.

  As shown in FIG. 4, each pressure chamber 52 communicates with a common flow channel 55 through a supply port 54. The common channel 55 communicates with an ink tank (not shown in FIG. 4, not shown in FIG. 6 and indicated by reference numeral 60) serving as an ink supply source, and the ink supplied from the ink tank 60 passes through the common channel 55 in FIG. Then, it is distributed and supplied to each pressure chamber 52.

  An actuator 58 having an individual electrode 57 is joined to a pressure plate (vibrating plate that also serves as a common electrode) 56 constituting a part of the pressure chamber 52 (the top surface in the drawing). By applying a driving voltage to the individual electrode 57, the actuator 58 is deformed to change the volume of the pressure chamber 52, and ink is ejected from the nozzle 51 due to the pressure change accompanying this. For the actuator 58, a piezoelectric body such as a piezoelectric element is preferably used. After ink discharge, new ink is supplied from the common channel 55 to the pressure chamber 52 through the supply port 54.

  As shown in FIG. 5, the ink chamber unit 53 having such a structure is latticed in a fixed arrangement pattern along a row direction along the main scanning direction and an oblique column direction having a constant angle θ not orthogonal to the main scanning direction. The high-density nozzle head of this example is realized by arranging a large number in the shape.

  That is, with a structure in which a plurality of ink chamber units 53 are arranged at a constant pitch d along a certain angle θ with respect to the main scanning direction, the pitch P of the nozzles projected so as to be aligned in the main scanning direction is d × cos θ. Thus, in the main scanning direction, each nozzle 51 can be handled equivalently as a linear arrangement with a constant pitch P. With such a configuration, it is possible to realize a high-density nozzle configuration in which 2400 nozzle rows are projected per inch (2400 nozzles / inch) so as to be aligned in the main scanning direction.

  When the nozzles are driven by a full line head having a nozzle row having a length corresponding to the entire printable width, (1) all the nozzles are driven simultaneously, (2) the nozzles are sequentially moved from one side to the other. (3) The nozzles are divided into blocks, and the nozzles are sequentially driven from one side to the other for each block, etc., and one line (1 in the width direction of the paper (direction perpendicular to the paper conveyance direction)) Driving a nozzle that prints a line of dots in a row or a line consisting of dots in a plurality of rows is defined as main scanning.

  In particular, when driving the nozzles 51 arranged in a matrix as shown in FIG. 5, the main scanning as described in (3) above is preferable. That is, nozzles 51-11, 51-12, 51-13, 51-14, 51-15, 51-16 are made into one block (other nozzles 51-21,..., 51-26 are made into one block, Nozzles 51-31,..., 51-36 as one block,...), And the nozzles 51-11, 51-12,. One line is printed in 16 width directions.

  On the other hand, by relatively moving the above-mentioned full line head and the paper, printing of one line (a line formed by one line of dots or a line composed of a plurality of lines) formed by the above-described main scanning is repeatedly performed. This is defined as sub-scanning.

  In implementing the present invention, the nozzle arrangement structure is not limited to the illustrated example. In the present embodiment, a method of ejecting ink droplets by deformation of an actuator 58 typified by a piezo element (piezoelectric element) is adopted. However, in the practice of the present invention, the method of ejecting ink is not particularly limited. Instead of the piezo jet method, various methods such as a thermal jet method in which ink is heated by a heating element such as a heater to generate bubbles and ink droplets are ejected by the pressure can be applied.

[Configuration of ink supply system]
FIG. 6 is a schematic diagram showing the configuration of the ink supply system in the inkjet recording apparatus 10. The ink tank 60 is a base tank that supplies ink to the head 50 and is installed in the ink storage / loading unit 14 described with reference to FIG. In the form of the ink tank 60, there are a system that replenishes ink from a replenishing port (not shown) and a cartridge system that replaces the entire tank when the remaining amount of ink is low. A cartridge system is suitable for changing the ink type according to the intended use. In this case, it is preferable that the ink type information is identified by a barcode or the like, and ejection control is performed according to the ink type. The ink tank 60 in FIG. 6 is equivalent to the ink storage / loading unit 14 in FIG. 1 described above.

  As shown in FIG. 6, a filter 62 is provided between the ink tank 60 and the head 50 in order to remove foreign matters and bubbles. The filter mesh size is preferably equal to or smaller than the nozzle diameter (generally about 20 μm). Although not shown in FIG. 6, a configuration in which a sub tank is provided in the vicinity of the head 50 or integrally with the head 50 is also preferable. The sub-tank has a function of improving a damper effect and refill that prevents fluctuations in the internal pressure of the head.

  Further, the inkjet recording apparatus 10 is provided with a cap 64 as a means for preventing the nozzle 51 from drying or preventing an increase in ink viscosity near the nozzle, and a cleaning blade 66 as a means for cleaning the nozzle surface 50A. . The maintenance unit including the cap 64 and the cleaning blade 66 can be moved relative to the head 50 by a moving mechanism (not shown), and is moved from a predetermined retracted position to a maintenance position below the head 50 as necessary.

  The cap 64 is displaced up and down relatively with respect to the head 50 by an elevator mechanism (not shown). The cap 64 is lifted to a predetermined raised position when the power is turned off or during printing standby and is brought into close contact with the head 50, thereby covering the nozzle surface 50 </ b> A with the cap 64.

  The cleaning blade 66 is made of an elastic member such as rubber, and can slide on the ink discharge surface (nozzle plate surface) of the head 50 by a blade moving mechanism (not shown). When ink droplets or foreign substances adhere to the nozzle plate, the nozzle plate surface is wiped by sliding the cleaning blade 66 on the nozzle plate to clean the nozzle plate surface.

  During printing or standby, when a specific nozzle is used less frequently and the ink viscosity in the vicinity of the nozzle increases, preliminary discharge is performed toward the cap 64 to discharge the deteriorated ink.

  When air bubbles are mixed in the ink in the head 50 (in the pressure chamber 52), the cap 64 is applied to the head 50, and the ink in the pressure chamber 52 (ink in which the air bubbles are mixed) is removed by suction with the suction pump 67. Then, the sucked and removed ink is sent to the collection tank 68. In this suction operation, the deteriorated ink that has increased in viscosity (solidified) is sucked out when the ink is initially loaded into the head 50 or when the ink is used after being stopped for a long time.

  If the head 50 is not ejected for a certain period of time, the ink solvent near the nozzles evaporates and the viscosity of the ink near the nozzles increases. Will not discharge. Therefore, before this state is reached (within the viscosity range in which ink can be discharged by the operation of the actuator 58), the actuator 58 is operated toward the ink receiver to discharge ink in the vicinity of the nozzle whose viscosity has increased. “Preliminary discharge” is performed. In addition, after the dirt on the surface of the nozzle plate is cleaned by a wiper such as a cleaning blade 66 provided as a cleaning means for the nozzle surface 50A, the foreign matter is prevented from entering the nozzle 51 by the wiper rubbing operation. Also, preliminary discharge is performed. Note that the preliminary discharge may be referred to as “empty discharge”, “purge”, “spitting”, or the like.

  In addition, if bubbles are mixed into the nozzle 51 or the pressure chamber 52 or if the viscosity increase of the ink in the nozzle 51 exceeds a certain level, ink cannot be ejected by the preliminary ejection, and the suction operation described below is performed.

  That is, when bubbles are mixed in the ink in the nozzle 51 or the pressure chamber 52, or when the ink viscosity in the nozzle 51 rises to a certain level or more, the ink can be ejected from the nozzle 51 even if the actuator 58 is operated. Disappear. In such a case, a suction means for sucking ink in the pressure chamber 52 with a pump or the like is brought into contact with the nozzle surface 50A of the head 50, and an operation of sucking ink mixed with bubbles or thickened ink is performed.

  However, since the above suction operation is performed on the entire ink in the pressure chamber 52, the amount of ink consumption is large. Therefore, when the increase in viscosity is small, it is preferable to perform preliminary discharge as much as possible.

[Explanation of control system]
FIG. 7 is a principal block diagram showing the system configuration of the inkjet recording apparatus 10. The inkjet recording apparatus 10 includes a communication interface 70, a system controller 72, an image memory 74, a ROM 75, a motor driver 76, a heater driver 78, a print control unit 80, an image buffer memory 82, a head driver 84, and the like.

  The communication interface 70 is an interface unit that receives image data sent from the host computer 86. As the communication interface 70, a serial interface such as USB, IEEE 1394, Ethernet, and wireless network, or a parallel interface such as Centronics can be applied. In this part, a buffer memory (not shown) for speeding up communication may be mounted. Image data sent from the host computer 86 is taken into the inkjet recording apparatus 10 via the communication interface 70 and temporarily stored in the image memory 74. The image memory 74 is a storage unit that temporarily stores an image input via the communication interface 70, and data is read and written through the system controller 72. The image memory 74 is not limited to a memory made of a semiconductor element, and a magnetic medium such as a hard disk may be used.

  The system controller 72 includes a central processing unit (CPU) and its peripheral circuits, and functions as a control device that controls the entire inkjet recording apparatus 10 according to a predetermined program, and also functions as an arithmetic device that performs various calculations. . That is, the system controller 72 controls each part such as the communication interface 70, the image memory 74, the motor driver 76, the heater driver 78, etc., performs communication control with the host computer 86, read / write control of the image memory 74, and the like. A control signal for controlling the motor 88 and the heater 89 of the transport system is generated.

  The ROM 75 stores programs executed by the CPU of the system controller 72 and various data necessary for control. The ROM 75 may be a non-rewritable storage unit, or may be a rewritable storage unit such as an EEPROM. The image memory 74 is used as a temporary storage area for image data, and is also used as a program development area and a calculation work area for the CPU.

  The motor driver 76 is a driver (drive circuit) that drives the motor 88 in accordance with an instruction from the system controller 72. The heater driver 78 is a driver that drives the heater 89 such as the post-drying unit 42 in accordance with an instruction from the system controller 72.

  The print control unit 80 has a signal processing function for performing various processing and correction processing for generating a print control signal from the image data in the image memory 74 according to the control of the system controller 72, and the generated print It is a control unit that supplies data (dot data) to the head driver 84. Necessary signal processing is performed in the print control unit 80, and the ejection amount and ejection timing of the ink droplets of the head 50 are controlled via the head driver 84 based on the image data. Thereby, a desired dot size and dot arrangement are realized.

  The print control unit 80 includes an image buffer memory 82, and image data, parameters, and other data are temporarily stored in the image buffer memory 82 when image data is processed in the print control unit 80. In FIG. 7, the image buffer memory 82 is shown in a mode associated with the print control unit 80, but it can also be used as the image memory 74. Also possible is an aspect in which the print controller 80 and the system controller 72 are integrated and configured with one processor.

  The head driver 84 drives the actuators of the heads 12K, 12C, 12M, and 12Y of the respective colors based on the print data given from the print control unit 80. The head driver 84 may include a feedback control system for keeping the head driving conditions constant.

  Image data to be printed is input from the outside via the communication interface 70 and stored in the image memory 74. At this stage, RGB image data is stored in the image memory 74.

  The image data stored in the image memory 74 is sent to the print controller 80 via the system controller 72, and is converted into dot data for each ink color by the print controller 80. That is, the print control unit 80 performs processing for converting the input RGB image data into KCMY four-color dot data. The dot data generated by the print controller 80 is stored in the image buffer memory 82.

  The head driver 84 generates a drive control signal for the head 50 based on the dot data stored in the image buffer memory 82. By applying the drive control signal generated by the head driver 84 to the head 50, ink is ejected from the head 50. An image is formed on the recording paper 16 by controlling the ink ejection from the head 50 in synchronization with the conveyance speed of the recording paper 16.

  As described with reference to FIG. 1, the print detection unit 24 is a block including a line sensor, reads an image printed on the recording paper 16, performs necessary signal processing, and the like to perform a print status (whether ejection is performed, droplet ejection And the detection result is provided to the print control unit 80.

  The print controller 80 performs various corrections on the head 50 based on information obtained from the print detector 24 as necessary.

[Description of image processing]
Next, image signal processing in the inkjet recording apparatus 10 configured as described above will be described.

  FIG. 8 is a block diagram showing an outline of an image processing function in the inkjet recording apparatus 10 of the present example. As shown in the figure, the inkjet recording apparatus 10 includes a color conversion unit 102 that generates KCMY data from input image data (RGB data) 100, a digital halftoning processing unit 104, and a head drive signal generation unit 106. A drive signal for the head 50 is generated based on dot data obtained as a result of digital halftoning, and a desired dot ejection 108 is performed.

  As described with reference to FIG. 7, image data (RGB data) 100 to be printed is input to the inkjet recording apparatus 10 through a predetermined image input unit such as the communication interface 70, and is input to the color conversion unit 102 shown in FIG. Sent. The color conversion unit 102 performs processing for converting RGB data of each pixel in the image into corresponding KCMY data. The KCMY data generated by the color conversion unit 102 is subjected to processing such as gradation correction and then sent to the digital halftoning processing unit 104.

  The digital halftoning processing unit 104 is a processing unit that halftones a KCMY gradation image and converts it into a dot pattern of a pseudo gradation image. As will be described in detail later, quantization processing using a threshold matrix and error diffusion are performed. A pseudo gradation image is generated by an algorithm combining the above. In the ink jet recording apparatus 10, a pseudo continuous tone image is formed by changing the droplet ejection density and dot size of fine dots with ink (coloring material) to the human eye. It is necessary to convert to a dot pattern that reproduces the gradation (shading of the image) as faithfully as possible. The digital halftoning processing unit 104 generates a dot pattern from input image data using a halftoning algorithm described later.

[Method of halftoning process]
Here, the method of the halftoning process in this example is demonstrated. FIG. 9 is a schematic diagram showing a part of a pixel array constituting a digital image, and each square cell represents a pixel 110. As illustrated, the pixels 110 are two-dimensionally arranged in the row direction and the column direction. All these pixels are grouped into a pixel at pixel position A shown in black in FIG. 9 and a pixel at pixel position B shown in white in FIG. For convenience of explanation, the pixel group at the pixel position A is referred to as a “first group”, and the pixel group at the pixel position B is referred to as a “second group”.

  The pixel position A is a pixel position where error correction from surrounding pixels is not performed, and an error (quantization error) generated by quantization by the threshold value matrix is diffused to the adjacent pixel position B of the second group. The arrows in FIG. 9 indicate the pixel position to which the quantization error generated by the quantization of each pixel position A is diffused.

  The pixel position B is a pixel position that is quantized after correction by the diffused error. The error caused by the quantization of the pixel position B is not calculated, and of course, the process of diffusing the error to other pixel positions is not performed.

  The pixel positions A are sparsely and substantially uniformly distributed over the entire surface of the image with regularity. Preferably, the pixel positions A are scattered apart from each other within a range of 1 to 2 pixels.

  In FIG. 9, a dispersion pattern in which pixel positions A are determined at a ratio of one pixel to five pixels in the row direction (that is, an array pattern in which four pixels belonging to the second group exist between the pixel positions A in the row direction). A distribution example is shown in which is repeated in the row direction while shifting by two pixel pitches in the column direction.

  In the figure, the quantization error at the pixel position A is diffused to four pixel positions B adjacent to the pixel position A in the vertical and horizontal directions. Further, the pixel position B to which the quantization error generated at the different pixel position A is diffused does not overlap. Note that the error distribution ratio may be equal distribution (in this case, ¼ each), may be weighted with respect to the diffusion direction, or may be randomly changed.

  The distribution form of the pixel positions A and the error diffusion form are not limited to the example of FIG. 9, and various forms are possible. 10 to 12 show other embodiments.

  In the example of FIG. 10, the pixel positions A and the pixel positions B are distributed in an arrangement form that forms a checkered pattern. Since the quantization error at each pixel position A is diffused to four adjacent pixel positions B in the upper, lower, left and right directions, the error distributed from the four pixel positions A in the upper, lower, left and right directions is superimposed on one pixel position B. Is added to

  In the example of FIG. 11, the dispersion pattern in which the pixel positions A and the pixel positions B are alternately arranged in the row direction is repeated every other row, and the pixel positions A are distributed every other pixel in the column direction and the row direction. It has become. In addition, two types of distribution patterns are set for the error distribution method from the pixel position A. That is, a pixel position (for example, reference numeral 111 in FIG. 11) that diffuses an error in four directions in the upper right, lower right, lower left, and upper left in an oblique direction, and a pixel position (for example, FIG. 11) and 112) are set to repeat alternately in the row direction and the column direction.

  The example in FIG. 12 has a distribution form in which the columns at the pixel positions A arranged in the column direction are shifted by one pixel pitch in the column direction as compared with the distribution form in FIG. In addition, an error generated by quantization of each pixel position A is diffused to adjacent pixel positions B in the three directions of upper, lower right, and lower left, and the diffusion destinations do not overlap.

  Quantization is performed for each pixel in the form illustrated in FIGS. 9 to 12 by the following procedure. FIG. 13A is a flowchart showing a quantization processing procedure.

  (Procedure 1): First, the pixel position A is quantized using a threshold matrix (step S1310). The threshold matrix is a matrix that defines thresholds that serve as determination criteria for determining whether or not to place a dot at the pixel position (x, y). The pixel value at the pixel position (x, y) is compared with the threshold value corresponding to that position. If the pixel value is larger than the threshold value, a dot is placed at that position (x, y). No dot is placed at the position. The handling when the gradation value is equal to the threshold value may be determined in either direction.

  The threshold value matrix used here preferably has blue noise characteristics. The blue noise mask is designed to provide a visually favorable dot arrangement for the DC component of the image. At this time, quantization noise (noise generated when the number of gradations decreases) is concentrated on the high-frequency side where it is difficult to see. When the image component of the input image is mainly a low-frequency component, the quantization result is high quality due to the blue noise characteristics.

  However, in the case of an image mainly including a high frequency component, the image component and the quantization noise are close to each other, so that a noisy quantization result is obtained. At this time, as shown in FIGS. 9 to 12, the image quality can be further improved by distributing the quantization error to the surrounding pixels.

  (Procedure 2): Next, an error generated in the quantization of the pixel position A in step S1310 described above is obtained (step S1320). Specifically, the difference between the quantization result and the pixel value is calculated as an error.

  (Procedure 3): The error thus obtained is distributed to the adjacent unquantized pixel positions B (step S1330). The pixel position of the error diffusion destination (distribution destination) and the distribution ratio are appropriately selected as described with reference to FIGS. The pixel value at the unquantized pixel position B is corrected (error correction) to the sum of the original gradation value and the error value distributed from the pixel position A.

  (Procedure 4): The unquantized pixel position B whose error has been corrected in step S1330 described above is quantized with a threshold matrix (step S1340). It should be noted that no error calculation is performed for the quantization of the pixel position B, and no error is diffused to surrounding pixels for an error caused by the quantization of the pixel position B.

  FIG. 13B shows a flowchart illustrating a more detailed process of the error distribution control process shown in step S1330 of FIG. In the figure, the same or similar steps to those in FIG. 13A are denoted by the same step numbers, and the description thereof is omitted.

  As shown in FIG. 13B, after obtaining the quantization error in step S1320, the process proceeds to step S1332. In step S1332, at least one of the error distribution position and the error distribution ratio is determined by a predetermined method. Next, the process proceeds to step S1333, and an error is distributed to adjacent unquantized pixel positions in accordance with the error distribution position and the error distribution determined in step S1332 (step S1333). After step S1333, the process proceeds to step S1340, and the unquantized pixel position B whose error has been corrected in step S1333 described above is quantized with a threshold matrix (step S1340).

  As described above, according to the quantization processing according to the present embodiment, since there are the pixel position B where no error is calculated and the pixel position A where no error correction processing is performed, the quantization is performed in comparison with the conventional method. The amount of computation related to is greatly reduced. Further, since it is not sequential processing, parallel processing can be easily configured. Furthermore, for some pixel positions (pixel positions A belonging to the first group), an error caused by quantization is taken into consideration, so that the average value of the input image is smaller than that of the conventional simple threshold matrix method. Since it is stored, the image quality is high, and the dot delay, which is a drawback of the conventional error diffusion method, does not occur.

  The “blue noise characteristics” will be outlined here. As a result of digital halftoning, a dot pattern is obtained. As a method for evaluating this dot pattern (dot arrangement), a method proposed by Robert and Ulichney is generally used ("Digital Halftoning"; published by The MIT Press).

  That is, the two-dimensional power spectrum of the dot arrangement is converted into polar coordinates, and an index corresponding to the average and variance of the spectrum at all angles is used for the spatial frequency fr corresponding to the radius of the polar coordinates as shown in FIG.

  The average index of the polar coordinate power spectrum is called “R.A.P.S (Radially Averaged Power Spectrum)” and is expressed by the following equation.

また、分散指標を「 Anisotropy 」と呼び、次式で表す。

The dispersion index is called “Anisotropy” and is expressed by the following equation.

ドット配置の視認性に関する指標であり、Anisotropyは、ドット配置の異方性に関する指標である。

An index relating to the visibility of dot arrangement, and Anisotropy is an index relating to anisotropy of dot arrangement.

FIG. 15 shows an example of RAPS calculated under certain conditions. In FIG. 15, σ _g is expressed by the following equation.

ただし、ｇは規格化された入力値を示し、０≦ｇ≦１である。

Here, g represents a standardized input value, and 0 ≦ g ≦ 1.

  In the graph shown in FIG. 15, the visual characteristics are not taken into consideration, but when the known visual characteristics (VTF) as shown in FIG. become. The VTF for calculating R.A.P.S and Anisotropy is not limited to the one proposed by Dooly & Shaw, and a known one can be used.

  FIG. 17 shows an example of Anisotropy calculated under a certain condition. According to Robert Ulichney, the anisotropy of dots is inconspicuous if Anisotropy is -10 dB [dB] or less.

  When the dot arrangement is evaluated using the method described with reference to FIGS. 14 to 17, the characteristics are such that RAPS is small in the low frequency range, has a peak at the medium frequency, and is constant at the high frequency, and Anisotropy is − A dot arrangement that is 10 decibels [dB] or less has a “blue noise characteristic”, and when the dot arrangement determined by the threshold matrix has a blue noise characteristic, the threshold matrix is referred to as a blue noise mask. The graph shown in FIG. 15 also generally shows blue noise characteristics, but a typical example of the graph is as shown in FIG.

  Next, an example of an aspect related to the error distribution method described in the above “procedure 3” (step S1330 in FIG. 13A) will be described. The error distribution method in “Procedure 3” can be configured as follows.

  (Additional configuration 1): The ratio of error distribution to adjacent unquantized pixel positions B is randomized. Thereby, generation | occurrence | production of the unique pattern (nonuniformity) unpreferable on image quality can be suppressed.

  FIG. 13C shows a flowchart thereof. In the figure, the same or similar steps to those in FIG. 13A are denoted by the same step numbers, and the description thereof is omitted. As shown in FIG. 13C, after obtaining the quantization error in step S1320, the process proceeds to step S1334. In step S1334, a parameter set for changing at least one of the error distribution position and the error distribution is prepared in advance, and one is randomly selected from the parameter set. Next, the process proceeds to step S1335, and an error is distributed to adjacent unquantized pixel positions according to the error distribution position and error distribution of the parameter selected in step S1334 (step S1335). After step S1335, the process proceeds to step S1340, and the unquantized pixel position B whose error has been corrected in step S1335 described above is quantized with a threshold matrix (step S1340).

  (Additional configuration 2): Error distribution is controlled according to the inclination of the surrounding pixels (gradient of the gradation value). For example, a difference in gradation value between adjacent pixels is obtained, and as the degree of the difference (inclination) increases, error diffusion is promoted in a direction perpendicular to the inclination direction, and error diffusion in the inclination direction is suppressed. This improves edge reproduction.

  FIG. 13D shows a flowchart thereof. In the figure, the same or similar steps to those in FIG. 13A are denoted by the same step numbers, and the description thereof is omitted. As shown in FIG. 13D, the quantization error is obtained in step S1320, while the gradation value gradient near the pixel position A is obtained (step S1322). Thereafter, the process proceeds to step S1336. In step S1336, a parameter set for changing at least one of the error distribution position and the error distribution corresponding to the gradation value gradient is prepared in advance, and 1 is selected from the parameter set based on the gradation value gradient calculated in step S1322. Select one.

  Next, the process proceeds to step S1337, and an error is distributed to adjacent unquantized pixel positions according to the error distribution position and error distribution of the parameter selected in step S1336 (step S1337). After step S1337, the process proceeds to step S1340, and the unquantized pixel position B whose error has been corrected in step S1337 described above is quantized with a threshold matrix (step S1340).

  (Additional configuration 3): Error distribution is controlled according to the frequency characteristics of surrounding pixels. For example, error diffusion is suppressed as the high frequency component is smaller. Thereby, the blue noise characteristic described above can be positively used. Further, the error diffusion destination may be changed according to the directionality of the frequency component. For example, the frequency component in the error diffusion direction is examined, and error diffusion in that direction is controlled.

  FIG. 13 (e) shows a flowchart thereof. In the figure, the same or similar steps to those in FIG. 13A are denoted by the same step numbers, and the description thereof is omitted. As shown in FIG. 13E, the quantization error is obtained in step S1320, while the spatial frequency characteristic in the vicinity of the pixel position A is obtained (step S1324). Thereafter, the process proceeds to step S1338. In step S1338, a parameter set for changing at least one of the error distribution position and the error distribution corresponding to the spatial frequency characteristic is prepared in advance, and one of the parameter sets is calculated based on the spatial frequency characteristic calculated in step S1324. select.

  Next, the process proceeds to step S1339, and an error is distributed to adjacent unquantized pixel positions according to the error distribution position and error distribution of the parameter selected in step S1338 (step S1339). After step S1339, the process proceeds to step S1340, and the unquantized pixel position B whose error has been corrected in step S1339 described above is quantized with a threshold matrix (step S1340).

  (Additional configuration 4): In the case of color processing, the size of the threshold matrix is varied depending on the color, or the pixel position A where the error is diffused and the pixel position B where the error is not diffused are varied depending on the color. In particular, it is preferable that visually conspicuous colors (for example, cyan and magenta) have pixel positions A and B different from each other. Alternatively, there is also an aspect in which an error diffusion pattern (FIGS. 9 to 12, etc.) is made different depending on colors. Specifically, an error diffusion pattern as shown in FIG. 10 is used for cyan and magenta that are visually noticeable, and an error diffusion pattern as shown in FIG. 9 is used for yellow that is not visually noticeable. Thus, interference between colors can be reduced by making the quantization process different for each color.

  (Additional configuration 5): Contrary to the above-mentioned “additional configuration 4”, in color processing, a mode in which a plurality of colors are collectively processed by a common threshold value matrix and error diffusion pattern is also possible.

  R, G, and B images are not separately quantized (by color plane) for an image that has two or more signals per pixel, such as RGB signals, and RGB signals for one pixel are not quantized. It is possible to perform quantization processing at the same time, and such a method is called vector quantization. When performing separate quantization processing for each color, it is difficult to control the quantization result between RGB (dot overlap), whereas one pixel RGB signal must be quantized simultaneously. Thus, the quantization result between RGB (dot overlap) can be easily controlled.

  The advantage of applying vector quantization in the embodiment of the present invention is as follows.

  When dot overlap is taken into account in error diffusion processing, the amount of computation becomes very large, and a desired result is not always obtained accurately (see Japanese Patent Laid-Open No. 8-279920).

  On the other hand, with threshold matrix processing, desirable results (desired dot overlap) are obtained (reference material: Modified Jointly Blue Noise Mask Approach Using S-CIELAB Color Difference., J Imaging Sci Technol, VOL. 46 NO. 6; PAGE.543). -551; (200111-200212) or Properties of Jointly-Blue Noise Masks and Applications to color Halftoning, J Imaging Sci Technol, VOL. 44 NO.4; PAGE.360-370) In such a case, there is a disadvantage that the image quality is inferior because the error is not taken into consideration.

  When the frequency component of an image is at a medium to high frequency by partially diffusing an error with respect to a threshold matrix in consideration of dot overlap (see the above reference material for the creation method) as in the embodiment of the present invention. In addition, high-quality processing can be performed in consideration of errors.

  As in the example shown in FIG. 10, the higher the number of pixels for which the error is calculated, the higher the image quality is, but the more the error calculation amount is. Therefore, it is preferable to select the distribution form of the pixel positions A and B and the error diffusion pattern (for example, FIGS. 9 to 12) in consideration of the balance between the image quality and the calculation amount.

  Instead of the flowchart described with reference to FIG. 13A, a mode in which quantization processing according to the flowchart illustrated in FIG. 19A is performed is also possible.

  (Procedure 1): First, representative values of peripheral pixels including the pixel position A are obtained (step S1410). For example, the representative value is calculated in a range of 3 × 3 pixels including the adjacent pixels of 8 pixels around the target pixel. As the representative value, an average value, a median value, or the like can be used.

  (Procedure 2): Next, the representative value is quantized using the threshold value matrix, and the quantization result is set as the quantization result of the pixel position A (step S1420). As in FIGS. 13A to 13E, the threshold value matrix preferably has blue noise characteristics.

  (Procedure 3): An error between the gradation value at the pixel position A and the quantization result at the pixel position A in step S1420 is obtained (step S1430).

  (Procedure 4): The error thus obtained is distributed to the adjacent unquantized pixel positions B (step S1440).

  (Procedure 5): The unquantized pixel position B whose error has been corrected in step S1440 described above is quantized with a threshold matrix (step S1450). With respect to the quantization of the pixel position B, the error calculation is not performed, and the error generated by the quantization of the pixel position B is not diffused to the surrounding pixels, as shown in the flowcharts of FIGS. 13 (a) to 13 (e). It is the same.

  Variations of the flowchart described in FIG. 19A are shown in FIGS. 19B to 19G. In these drawings, the same or similar steps as those in FIG. 19A are denoted by the same step numbers, and the description thereof is omitted.

  FIG. 19B is a flowchart showing a mode in which the size of the threshold matrix varies depending on the color. As shown in the figure, first, a color to be processed is set (step S1401), and a threshold value matrix having a different size for the color to be processed is set (step S1402). Next, representative values of peripheral pixels including the pixel position A are obtained (step S1410). The representative value thus obtained is quantized using the threshold value matrix set in step S1402, and the quantization result is set as the quantization result of the pixel position A (step S1420). The subsequent processing (steps S1430 to S1450) is the same as that in FIG. Note that the threshold value matrix used in the quantization in step S1450 shown in FIG. 19B is the threshold value matrix set in step S1402.

  FIG. 19C is a flowchart illustrating a mode in which the position of the pixel position A varies depending on the color. As shown in the figure, first, the color to be processed is set (step S1401), and the pixel position A is set so that the pixel position A differs depending on the color to be processed (step S1403). Next, representative values of peripheral pixels including the pixel position A are obtained (step S1410). The representative value thus obtained is quantized using the threshold matrix, and the quantization result is set as the quantization result at the pixel position A (step S1420). The subsequent processing (steps S1430 to S1450) is the same as that in FIG.

  FIG. 19D is a flowchart showing a mode in which an error diffusion method is changed depending on colors. As shown in the figure, first, a color to be processed is set (step S1401), and an error diffusion coefficient corresponding to the color to be processed is set (step S1404). Next, representative values of peripheral pixels including the pixel position A are obtained (step S1410). The representative value thus obtained is quantized using the threshold matrix, and the quantization result is set as the quantization result at the pixel position A (step S1420). Thereafter, the error obtained in step S1430 is distributed to adjacent unquantized pixel positions based on the error diffusion coefficient (coefficient set in step S1404) (step S1442). The unquantized pixel position B whose error has been corrected in step S1442 is quantized with a threshold matrix (step S1450).

  FIG. 19E is a flowchart showing another mode in which the error diffusion method varies depending on the color. As shown in the figure, first, a color to be processed is set (step S1401), and a pixel position A having an error distribution different from the arrangement pattern of the pixel position A is set for the color to be processed (step S1406). Next, a representative value of peripheral pixels including the pixel position A related to the setting in step S1406 is obtained (step S1410). The representative value thus obtained is quantized using the threshold matrix, and the quantization result is set as the quantization result at the pixel position A (step S1420). Thereafter, the error obtained in step S1430 is distributed to the adjacent unquantized pixel positions by the error distribution (according to the error distribution method set in step S1406) (step S1444). In this way, the unquantized pixel position B whose error has been corrected in step S1444 is quantized with the threshold matrix (step S1450).

  FIG. 19F is a flowchart showing a mode in which a vector quantization process is performed on a plurality of color signals of one pixel together. As shown in the figure, first, a threshold matrix corresponding to all ON / OFF combinations of each color in which at least one color is ON is set (step S1408). That is, a threshold matrix in which all color components of all colors correspond to ON, a threshold matrix in which one color component corresponds to OFF, a threshold matrix corresponding to OFF of two color components,..., Only one color component is ON A threshold matrix corresponding to is set.

  Next, the order of all ON / OFF combinations of each color in which at least one color is ON is set (step S1409). Thereafter, representative values of peripheral pixels including the pixel position A are obtained for each color (step S1411). The representative value obtained in step S1411 is vector quantized using the threshold matrix (threshold matrix set in step S1408), and this vector quantization result is used as the quantization result for each color at the pixel position A (step S1422).

  Then, the error of each color value at the pixel position A and the quantization result of each color at the pixel position A obtained at step S1422 is obtained (step S1433), and the error of each color is given to the adjacent unquantized pixel position. Distribution is performed according to the distribution method (step S1446). In this way, the unquantized pixel position B whose error has been corrected in step S1446 is quantized with the threshold matrix (step S1452).

  A flowchart of the vector quantization processing in steps S1422 and S1452 is shown in FIG. As shown in the figure, the vector quantization process is executed according to the order of the combination of ON and OFF (the order set in step S1409 in FIG. 19F) (step S2012). A value obtained by adding only the values of the respective colors that are ON corresponding to the combination of ON and OFF related to the corresponding order is obtained (step S2014). Next, it is compared with a threshold value matrix corresponding to the combination to determine whether or not the addition result satisfies a predetermined condition (step S2016).

  If the predetermined condition is satisfied in step S2016, the process proceeds to step S2018, and ON / OFF of the combination is set as a quantization result. On the other hand, if the predetermined condition is not satisfied in step S2016, the process proceeds to step S2020, and it is confirmed whether or not the unprocessed combination exists. If an unprocessed combination exists, the processing target is changed to the next ON / OFF combination according to the combination order (the order set in step S1409) (step S2022), and the process returns to step S2014.

  In the determination in step S2020, if there is no unprocessed combination, “all colors are OFF” is set as the quantization result (step S2024). Thus, when the quantization result is determined in step S2018 or step S2024, the vector quantization processing subroutine is exited, and the process returns to the flowchart of FIG.

  Similarly to FIGS. 13 (a) to 13 (e), when performing quantization by the quantization flow described in FIGS. 19 (a) to 19 (g), parallel processing can be easily configured because it is not sequential processing. . In addition, since the error caused by quantization is taken into consideration, the average value of the input image is preserved compared to the conventional threshold matrix method, so the image quality is high, and the dot delay is a disadvantage of the conventional error diffusion method. Does not occur.

  FIG. 20 shows a configuration example of the digital halftoning processing unit 104 that realizes the quantization processing described with reference to FIGS. 9 to 19G.

  The digital halftoning processing unit 104 includes a group division unit 202, a grouping control unit 204, a first quantization processing unit 206, a threshold matrix storage unit 208, an error calculation unit 210, a diffusion pattern control unit 212, an error addition unit 214, A two-quantization processing unit 216 is included. Each block (202 to 216) may be realized by hardware such as an electronic circuit or a signal processing circuit, or the function thereof may be realized by software, and of course, realized by an appropriate combination thereof. You can also.

  The group dividing unit 202 performs a process of dividing the constituent pixels of the input image (gradation image) 220 into a first group and a second group according to a predetermined distribution form as described with reference to FIGS. The grouping control unit 204 performs control to switch the distribution form of the pixel position A and the pixel position B.

  The first quantization processing unit 206 is a processing unit that performs quantization using a threshold value matrix on the pixel position A belonging to the first group among the pixel groups grouped by the group dividing unit 202. When obtaining the representative values described with reference to FIGS. 19A to 19F, the representative values are calculated in the previous stage of the first quantization processing unit 206.

  The threshold value matrix storage unit 208 shown in FIG. 20 stores threshold value matrix data used for quantization calculation. It is preferable that the threshold value matrix has a blue noise mask characteristic in consideration of a low and middle frequency range. Further, it is possible to store a plurality of threshold matrixes and switch the matrix to be used as necessary.

  The error calculator 210 calculates an error generated by the quantization of the first quantization processor 206. The difference between the original pixel value and the quantization result is calculated as an error.

  The diffusion pattern control unit 212 is a control unit that sets the quantization error distribution method obtained by the error calculation unit 210, and determines the pixel position of the error diffusion destination and the distribution ratio in cooperation with the grouping control unit 204. To do.

  The error addition unit 214 is a calculation unit that corrects the pixel value of the pixel position of the diffusion destination according to the pixel position of the diffusion destination determined by the diffusion pattern control unit 212 and the distribution ratio of the error to the pixel position. An error correction value obtained by multiplying the error obtained by the error calculation unit 210 by the distribution ratio is added to the pixel value of the target pixel to obtain a new pixel value (pixel value after error correction).

  The second quantization processing unit 216 is a processing unit that quantizes the second group of unquantized pixel positions B subjected to error correction by the error addition unit 214 using a threshold matrix. In the second quantization processing unit 216, the same threshold matrix as that of the first quantization processing unit 206 may be used, or a different threshold matrix may be used. For the pixel positions quantized by the second quantization processing unit 216, error diffusion is not performed, and therefore no error calculation is required.

  In this way, the halftone image 222 is generated through two-stage quantization processing for each group.

  In the above description, an inkjet recording apparatus is illustrated as an example of an image forming apparatus. However, the scope of application of the present invention is not limited to this, and dots such as a thermoautochrome (TA) printer, a sublimation printer, and a laser printer are used. The present invention can be applied to various output devices such as an image forming apparatus and a display device that express gradations.

  Further, the image processing apparatus of the present invention is not limited to an aspect incorporated in an image forming apparatus such as an ink jet recording apparatus, and can also be realized by a computer. A program for causing a computer to realize the above-described image processing function is recorded on a CD-ROM, a magnetic disk or other information storage medium, and the program is provided to a third party through the information storage medium, or a communication line such as the Internet. It is also possible to provide a download service for the program.

1 is an overall configuration diagram of an ink jet recording apparatus using a droplet discharge device according to an embodiment of the present invention. FIG. 1 is a plan view of a main part around a printing unit of the ink jet recording apparatus shown in FIG. Plane perspective view showing the configuration of the head Enlarged view of the main part of Fig. 3 (a) Plane perspective view showing another configuration example of a full-line head Sectional view along line 4-4 in Fig. 3 (a) Enlarged view showing the nozzle arrangement of the head shown in FIG. Schematic diagram showing the configuration of the ink supply system in the inkjet recording apparatus of this example Main block diagram showing the system configuration of the inkjet recording apparatus of this example Block diagram showing an outline of the image processing function in the ink jet recording apparatus of this example Schematic diagram showing a first example of pixel grouping and quantization error diffusion patterns Schematic diagram showing a second example of pixel grouping and quantization error diffusion patterns Schematic diagram showing a third example of pixel grouping and quantization error diffusion patterns Schematic diagram showing a fourth example of pixel grouping and quantization error diffusion patterns Flowchart showing quantization processing procedure Flowchart showing another example of quantization processing procedure Flowchart showing another example of quantization processing procedure Flowchart showing another example of quantization processing procedure Flowchart showing another example of quantization processing procedure Diagram showing coordinate system for calculating two-dimensional power spectrum Graph showing an example of the polar power spectrum average index (R.A.P.S) calculated under certain conditions Graph showing the visual characteristics (VTF) of the human eye Graph showing an example of polar power spectrum dispersion index (anisotropy) calculated under certain conditions Typical R.A.P.S graph of dot arrangement with blue noise characteristics Flowchart showing another example of quantization processing procedure Flowchart showing another example of quantization processing procedure Flowchart showing another example of quantization processing procedure Flowchart showing another example of quantization processing procedure Flowchart showing another example of quantization processing procedure Flowchart showing another example of quantization processing procedure A flowchart showing a subroutine of vector quantization processing Block diagram showing a configuration example of a digital halftoning processing unit

Explanation of symbols

DESCRIPTION OF SYMBOLS 10 ... Inkjet recording device, 12 ... Printing part, 12K, 12C, 12M, 12Y ... Head, 14 ... Ink storage / loading part, 16 ... Recording paper, 18 ... Paper feed part, 22 ... Adsorption belt conveyance part, 50 ... Head 50A ... Nozzle surface, 51 ... Nozzle, 52 ... Pressure chamber, 72 ... System controller, 75 ... ROM, 80 ... Print control unit, 104 ... Digital halftoning processing unit, 202 ... Group division unit, 204 ... Grouping control unit , 206 ... threshold matrix storage section, 206 ... first quantization processing section, 210 ... error calculation section, 212 ... diffusion pattern control section, 214 ... error addition section, 216 ... second quantization processing section

Claims (16)

An image processing method for halftoning a digital image composed of an array of pixels having gradation values according to image content,
For the pixels constituting the digital image, the first group to which the pixel at the pixel position that diffuses the quantization error generated by quantization to the surrounding pixels belongs, and the quantization error generated by quantization is diffused to the surrounding pixels. A grouping step for dividing the pixel group into a second group to which the pixel at the pixel position to be excluded belongs;
A first quantization step of performing quantization using a threshold value matrix on pixels belonging to the first group;
An error calculating step for obtaining a quantization error generated in the first quantization step;
An error diffusion step of diffusing the quantization error obtained in the error calculation step to at least one unquantized pixel adjacent to the pixel related to the quantization process and belonging to the second group;
A second quantization step of performing quantization using a threshold matrix on the unquantized pixels to which an error value is given by the error diffusion step;
An image processing method comprising:   The pixels belonging to the first group are arranged in a two-dimensional array of pixels constituting the digital image in a substantially uniform and distributed manner across a plurality of columns and a plurality of rows. The image processing method according to claim 1.   3. The image processing method according to claim 1, wherein the threshold value matrix has a blue noise characteristic.   4. The error diffusion step adds the quantization error obtained in the error calculation step to unquantized pixels belonging to the second group at a predetermined ratio. 5. The image processing method as described.   The error diffusion step includes an error distribution control step of controlling at least one of a pixel position of a quantization error diffusion destination and an error distribution ratio obtained in the error calculation step. Or the image processing method of 3.   The image processing method according to claim 5, wherein the error distribution control step randomly determines at least one of the pixel position of the diffusion destination and the error distribution ratio.   The error distribution control step controls at least one of a pixel position of the diffusion destination and a ratio of error distribution according to a gradation value gradient of surrounding pixels of the pixel related to the quantization processing. 5. The image processing method according to 5.   The error distribution control step controls at least one of a pixel position of the diffusion destination and an error distribution ratio according to frequency characteristics of surrounding pixels of the pixel related to the quantization process. 8. The image processing method according to 7. A representative value calculating step for obtaining a representative value of a pixel range including the target pixel belonging to the first group and its peripheral pixels;
The representative value obtained in the representative value calculation step is used as the value of the target pixel, the target value is quantized using a threshold matrix, and the result of the quantization process is the quantization result of the target pixel. age,
9. The image processing method according to claim 1, wherein the error calculation step obtains an error between a gradation value of the target pixel and a quantization result of the target pixel.   10. The image processing method according to claim 1, wherein when the digital image is a color image including a plurality of color components, the size of the threshold value matrix is varied depending on the color.   11. The image according to claim 1, wherein when the digital image is a color image including a plurality of color components, pixel positions of pixels belonging to the first group are made different depending on colors. Processing method.   The image processing method according to claim 1, wherein when the digital image is a color image including a plurality of color components, an error diffusion method is varied depending on colors.   The image processing method according to claim 1, wherein when the digital image is a color image including a plurality of color components, vector quantization processing is performed on a plurality of color signals of one pixel together. . An image processing apparatus that performs halftoning of a digital image configured by an array of pixels having gradation values according to image content,
For the pixels constituting the digital image, the first group to which the pixel at the pixel position that diffuses the quantization error generated by quantization to the surrounding pixels belongs, and the quantization error generated by quantization is diffused to the surrounding pixels. Grouping means for dividing into a second group to which a pixel at a pixel position to be excluded belongs,
First quantization means for performing quantization on the pixels belonging to the first group using a threshold matrix;
An error calculating means for obtaining a quantization error generated by the first quantizing means;
Error diffusion means for diffusing the quantization error obtained by the error calculation means to at least one unquantized pixel adjacent to the pixel related to the quantization processing and belonging to the second group;
Second quantization means for performing quantization using a threshold matrix on the unquantized pixels to which an error value is given by the error diffusion means;
An image processing apparatus comprising: An image forming apparatus comprising the image processing apparatus according to claim 14.
An ejection head in which nozzles for ejecting droplets are formed;
An ejection control means for controlling ejection from the ejection head based on halftone image data obtained by the image processing apparatus;
An image forming apparatus, wherein an image is formed on a recording medium by droplets ejected from the nozzle. A program for realizing the functions of the image processing apparatus according to claim 14 by a computer.

Images