JP5596913B2 - Image processing apparatus and image processing method - Google Patents

Description

The present invention relates to an image processing apparatus and an image processing method, and more particularly to an image processing apparatus and an image processing method for converting the number of gradations to a number of gradations smaller than the number of gradations of an input image.

  As information output devices in word processors, personal computers, facsimiles, and the like, there are various types of recording devices that record information such as desired characters and images on a sheet-like recording medium such as paper or film. Among them, an output system that forms text and images on a recording medium by attaching a recording agent to the recording medium has been put into practical use.

  In the ink jet recording apparatus which is one of the output methods described above, a nozzle in which a plurality of ink discharge ports (nozzles) capable of discharging ink of the same color and the same density are arranged in an integrated manner in order to improve recording speed and improve image quality. Use groups. Furthermore, in order to improve the image quality, there are provided nozzle groups that can eject ink with the same color and different density, and nozzles that can be ejected by changing the ejection amount of ink with the same color and same density in several stages. Some are also available.

  In such an image forming apparatus, as a means for converting multi-valued input image data into a binary image corresponding to a dot recording signal (or an image having two or more values and a smaller number of gradations than the number of input gradations), R. There is an error diffusion method by Floyd et al. In this error diffusion method, pseudo gradation expression is performed by diffusing a binarization error generated in a certain pixel to a plurality of subsequent pixels.

  In addition to the error diffusion method described above, dithering may be used as a method for converting multi-value input image data into a binary image corresponding to a dot recording signal (or an image having two or more values and a smaller number of gradations than the number of input gradations). The law is known. In the dither method, a threshold value matrix prepared in advance and multi-valued input data are compared, and binarization is performed to perform pseudo gradation expression. Since this dither method is simpler than the error diffusion method, it is known that high-speed processing can be performed.

  Further, Patent Document 1 proposes an image processing method for determining the formation order and arrangement when forming the above-mentioned tone-converted image with an image forming apparatus. According to such a technique, by processing the error diffusion method for each scan, it is possible to suppress a decrease in image quality due to density unevenness even when the registration of each scan varies.

  More specifically, main scanning is performed a plurality of times with different nozzle groups on the same main scanning recording area on a predetermined recording medium, and a binary image is formed by the error diffusion method for each main scanning. When the binary image is generated by performing the error diffusion method for each main scan, the dispersibility of the dot arrangement in the main scan is high and uniform. For this reason, even when physical registration changes such as the feeding amount of the recording medium and the position of the recording element when an image is formed by a plurality of main scans, the change in graininess hardly occurs. Furthermore, since there is little correlation between the dot arrangements between a plurality of scans, even if the registration fluctuates, the change in dot coverage on the paper surface is reduced, and density unevenness is considerably alleviated.

  Hereinafter, the cause of the above-described physical registration fluctuation will be specifically described.

  First, in the ink jet recording apparatus, when the carriage on which the recording head is mounted is scanned with respect to the recording area, the carriage speed fluctuates due to some influence.

  Further, even in a mechanism for transporting the recording medium in the sub-scanning direction after recording in the main scanning direction, the recording medium transport amount fluctuates due to, for example, a device manufacturing error.

  Furthermore, it is ideal that the recording medium and the recording head surface be mounted in parallel and perpendicular to the main scanning direction when the recording head is mounted on the carriage, but with a certain inclination (variation) from an accurate position due to manufacturing errors. May be installed. Further, since the recording head is scanned many times in the main scanning direction, the mounting state may gradually change.

  On the other hand, in the recording medium, the ink droplets ejected from the recording head penetrate into the inside of the paper, and the paper is swollen by absorbing the water contained in the ink into the paper. In multi-pass printing in which the same area is printed by a plurality of scans, the surface of the print medium is displaced for each scan (hereinafter referred to as cockling), that is, the distance between the print head and the print medium varies depending on the print position.

  As described above, in the ink jet recording apparatus, the above-described fluctuation occurs due to a manufacturing error or some reason during scanning. These fluctuation amounts are not constant depending on the apparatus, and change variously for each printing scan.

  Due to the physical variation between the recording apparatus and the recording medium as described above, the dot landing position (registration) is shifted and the graininess is changed. As a result, density unevenness occurs in each recording area, and the image quality of the output image deteriorates.

  A countermeasure is proposed in Patent Document 2 for image quality deterioration due to deviation of the landing position of ink due to speed fluctuation of the recording head. According to Patent Document 2, the movement amount and acceleration of the recording head are detected, and an ejection timing signal delayed by a predetermined time with respect to a reference timing corresponding to the predetermined acceleration is sent out to the set position on the recording surface with high accuracy. A way to reach it has been proposed.

  Further, Patent Document 3 proposes a countermeasure against landing position deviation due to cockling in which the surface of the recording medium is displaced for each scanning. According to Patent Document 3, a method is proposed in which a distance between a recording head and a recording material is detected by a paper interval sensor, and a dot position is corrected by delaying ejection timing based on distance displacement information.

Further, Patent Document 4 proposes a countermeasure against the problem that dots are not recorded at equal intervals in the sub-scanning direction due to a manufacturing error of the recording apparatus. According to Patent Document 4, a method for selecting a recording mode for preventing image quality deterioration by generating data representing a recording position shift and detecting (reading) the information is proposed.
JP 2000-103088 A JP 2000-71438 A JP-A-11-240146 JP 2001-322261 A Japanese Patent Laid-Open No. 2001-180057 R. Floyd et al., “An adaptive algorithm for spatial grayscale”, SID International Symposium Digest of Technical Papers, vol4.3, 1975, pp. 36-37

  However, according to the method described in Patent Document 1 described above, since the correlation is small in the dot arrangement between the main scans, there is a problem that the low frequency components of the image generated by the plurality of scans are emphasized. is there. This low-frequency component is emphasized as the number of scans is increased, and is recognized as graininess that is visually obtrusive.

  Further, according to Patent Document 2 and Patent Document 3 described above, fluctuations in the recording device and the recording medium are detected and the landing position for each dot is corrected. Is difficult to align. Further, in the above-described method, the correction amount at one dot level is determined for each variation value (carriage speed variation, transport amount, inter-paper distance, etc.). However, in an actual printing operation, these fluctuations may occur at the same time. Therefore, the correction of one dot level for a certain fluctuation amount is less robust with respect to other fluctuations.

  Further, in Patent Document 4, for example, a recording mode is determined based on a fluctuation amount detected at a predetermined timing such as at the time of shipment or at the time of activation. However, in such a method, when the amount of change greatly changes due to device deterioration or operation failure due to vibration or the like, such as time-dependent conversion, the robustness with respect to the landing position change becomes low and image deterioration occurs.

  The present invention has been made in view of the above problems, and maintains image quality against physical registration fluctuations that occur between a recording apparatus and a recording medium, and relies on low-frequency components of an image. An object of the present invention is to suppress the deterioration of graininess and make it possible to form a high-quality image.

In order to achieve the above object, an image processing apparatus according to an aspect of the present invention has the following arrangement. That is,
An image processing apparatus for an image forming apparatus that forms an image by recording and scanning a recording head including a plurality of recording elements a plurality of times for the same region on a recording medium,
An acquisition means for acquiring information indicating the amount of registration variation between the recording head and the recording medium during a plurality of recording scans;
Generating means for generating scan data for each recording scan by dividing the input image data for each recording scan by dividing the pixel value of each pixel constituting the image data;
Setting means for setting constraint information indicating whether or not dots are easily formed for each pixel according to the amount of variation;
N-value conversion means for performing N-value conversion processing (N is an integer of 2 or more) with reference to the constraint information with respect to the scan data, and creating a dot pattern for the print scan,
About the m-th printing scan and the n-th (m and n are natural numbers, n> m) -th printing scan in the area of the printing medium,
The setting means includes a dot pattern corresponding to the m-th printing scan and the n-th printing scan based on a result of filtering using a filter for the dot pattern printed by the m-th printing scan. The N-value conversion means refers to the constraint information to be referred to for N-value processing on the scan data corresponding to the n-th print scan so that the dot pattern corresponding to is in the opposite phase in the low frequency region. To set,
The setting means increases the size of the filter as the variation amount increases .

An image processing method according to another aspect of the present invention for achieving the above object is
An image processing method of an image forming apparatus for forming an image by performing a recording scan with a recording head including a plurality of recording elements a plurality of times on the same area of a recording medium,
An acquisition step in which the acquisition unit acquires information indicating a registration variation amount between the recording head and the recording medium while performing a plurality of recording scans;
A generating step of generating scan data for each recording scan by dividing the input image data into pixel values of each pixel constituting the image data for each recording scan;
A setting step in which the setting means sets constraint information indicating whether or not dots are easily formed for each pixel according to the amount of variation;
An N-value conversion step in which an N-value conversion unit applies an N-value conversion process (N is an integer of 2 or more) to the scan data with reference to the constraint information, and creates a dot pattern for the print scan. And
About the m-th printing scan and the n-th (m and n are natural numbers, n> m) -th printing scan in the area of the printing medium,
In the setting step, the dot pattern corresponding to the m-th print scan and the n-th print scan are based on the result of filtering using a filter for the dot pattern printed by the m-th print scan. The constraint information referred to N-value processing is set for the scan data corresponding to the n-th print scan so that the dot pattern corresponding to is in the opposite phase in the low frequency region,
You increase the size of the filter as the amount of the variation is large.

  According to the present invention, image quality is maintained against physical registration fluctuations that occur between the recording apparatus and the recording medium, and deterioration in graininess depending on low-frequency components of the image is suppressed. Image quality images can be formed.

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

<First Embodiment>
FIG. 1 is a block diagram showing the configuration of the image forming apparatus according to the first embodiment. In FIG. 1, 1 indicates an image processing apparatus, and 2 indicates a printer. The image processing apparatus 1 can be implemented by a printer driver installed in a general personal computer, for example. In that case, each part of the image processing apparatus 1 described below is realized by a computer executing a predetermined program. As another configuration, for example, the printer 2 may include the image processing apparatus 1.

  The image processing apparatus 1 and the printer 2 are connected by a printer interface or a circuit. The image processing apparatus 1 inputs image data to be printed from the image data input terminal 101 and stores the input image data in the image buffer 102. The color separation processing unit 103 separates the input color image into ink colors provided in the printer 2. In this color separation processing, a color separation lookup table 104 (hereinafter referred to as color separation LUT 104) is referred to. The scan duty setting unit 105 further converts each ink color value separated by the color separation processing unit 103 into each ink color value for each scan, and stores the data in the scan duty buffer 107. Note that, when converting to each ink color value, as described later, a scan duty setting lookup table 106 (hereinafter referred to as a scan duty setting LUT 106) is referred to. In this way, the scan duty setting unit 105 sets the ink color value indicating the recording amount of each recording element in each recording scan, and holds this in the scanning duty buffer 107.

  The halftone processing unit 108 binarizes the multi-gradation (three gradations or more) values of each color obtained by the scanning duty setting unit 105 based on values stored in the constraint information buffer 109 described later. Convert to image data. The constraint information buffer 109 stores information indicating whether or not dots are likely to be formed at addresses on the recorded image. It is assumed that the constraint information buffer is secured for each ink color.

  The halftone image storage buffer 110 stores binary image data of each color obtained by the halftone processing unit 108. The constraint information calculation unit 111 performs a predetermined calculation using the binary image data stored in the halftone image storage buffer 110 and the duty data value for each scan stored in the scan duty buffer 107, and updates the constraint information buffer 109. To do.

  The binary image data stored in the halftone image storage buffer 110 is output from the output terminal 112 to the printer 2. The printer 2 forms binary image data formed by the image processing apparatus 1 on the recording medium by moving the recording head 201 relative to the recording medium 202 vertically and horizontally.

  As the recording head 201, a thermal transfer method, an electrophotographic method, an ink jet method, or the like can be used, and each has one or more recording elements (nozzles in the case of an ink jet method). The moving unit 203 moves the recording head 201 under the control of the head control unit 204. The transport unit 205 transports the recording medium under the control of the head control unit 204. The ink color and ejection amount selection unit 206 also selects an ink color (and its ejection amount) from among the ink colors mounted on the recording head 201 based on the binary image data of each color formed by the image processing apparatus 1. ) Is selected.

  The fluctuation amount detection unit 207 detects the fluctuation amount of a main part (for example, the recording head 201) of the printer 2 by using various sensors. The fluctuation amount detection unit 207 detects a physical fluctuation amount between the recording head 201 and the recording medium, and the detected fluctuation amount is sent to the constraint information calculation unit 111 of the image processing apparatus 1 via the fluctuation amount input terminal 208. hand over.

  In the present embodiment, the variation detected by the variation detection unit 207 is the position variation due to the velocity variation in the main scanning direction of the recording head. Therefore, as a sensor used for detection, for example, a semiconductor type acceleration detection sensor described in Patent Document 2 can be cited, but is not limited thereto.

The fluctuation amount detected by the fluctuation amount detection unit 207 is the movement amount of the recording head. For example, as shown in FIG. 25, the fluctuation amount detection unit 207 detects the position x ′ (t _k ) of the recording head and the carriage member 2501 at every predetermined timing t _k (k = 0, 1,... K). At the timing t _k , the eccentric distance di = | x ′ (t _k ) −x (t _k ) | (i = 0,... K−1) from the position x (t _k ) where the dot originally landed fluctuates. Detected as a quantity. Note that the lower part of FIG. 25 shows a state in which dots are applied to the recording medium 2502.

  In the present embodiment, the fluctuation amount detection unit 207 detects the deviation σ_di of the deviation amount from the predetermined reference position described above in one scan as a fluctuation amount, and passes it to the constraint information calculation unit 111. Is not limited to this. For example, as another variation amount, the maximum value d_MAX of the displacement amount of the carriage in one scan or the displacement amount di detected at every predetermined timing may be used as the variation value used in the constraint information calculation unit 111.

  FIG. 2 is a diagram illustrating a configuration example of the recording head 201. In the first embodiment, in addition to four color inks of cyan (C), magenta (M), yellow (Y), and black (K), light cyan (Lc) and light magenta (Lm) having a relatively low ink density. Ink of 6 colors including is mounted on the recording head.

  In FIG. 2, a recording head having a configuration in which nozzles are arranged in a line in the paper conveyance direction is shown for simplicity of explanation, but the number and arrangement of nozzles are not limited to this. For example, there may be nozzle rows with the same color but different discharge amounts, there may be a plurality of nozzles with the same discharge amount, or the nozzles may be arranged in a zigzag manner. In FIG. 2, the ink color is arranged in a line in the head movement direction, but may be arranged in a line in the paper transport direction.

  Next, the operation of the image processing apparatus 1 of the present embodiment having the above-described functional configuration will be described with reference to the flowchart of FIG.

  First, multi-tone color input image data is input from the input terminal 101 and stored in the image buffer 102 (step S101). As input image data, color image data is constructed from three color components of red (R), green (G), and blue (B).

  Next, the color separation processing unit 103 performs color separation processing from RGB to CMYK and LcLm ink color planes on the multi-tone color input image data stored in the image buffer 102 using the color separation LUT 104. Perform (step S102). In the present embodiment, each pixel data after color separation processing is handled as 8 bits, but it may be converted with more gradations.

  As described above, the recording head 201 according to the first embodiment has six types of ink colors. Therefore, RGB color input image data is converted into image data of a total of 6 planes, each of CMYKLcLm planes. That is, six types of plane image data corresponding to six types of recording modes are generated.

  Hereinafter, the color separation processing of this embodiment will be described with reference to FIG. FIG. 4 shows the configuration of color separation processing of the color printer. The color separation processing unit 103 refers to the color separation LUT 104 and converts the image data R′G′B ′ into CMYKLcLm according to the following equation.

C = C_LUT — 3D (R ′, G ′, B ′) (1)
K = M_LUT — 3D (R ′, G ′, B ′) (2)
Y = Y_LUT — 3D (R ′, G ′, B ′) (3)
K = K_LUT — 3D (R ′, G ′, B ′) (4)
Lc = Lc_LUT — 3D (R ′, G ′, B ′) (5)
Lm = Lm_LUT — 3D (R ′, G ′, B ′) (6)
Here, each function defined on the right side of (1) to (6) corresponds to the color separation LUT 104. The color separation LUT 104 is a table for converting three input values of red, green, and blue into output values for each ink color. In the present embodiment, since the configuration includes six colors CMYKLcLm, an LUT configuration that obtains six output values from three input values is obtained. The color separation process of the first embodiment is completed through the above processing steps.

  In the next steps S102 to S103, the scan duty setting unit 105 calculates the recording amount of each of the plurality of recording elements for each recording scan of the recording head in accordance with the color-separated image data, and the scan duty buffer. 107. First, the scan duty setting unit 105 sets the scan number k and the color separation data cutout Y coordinate Ycut (k) (step S103). Note that the initial value of the scan number k is 1, and is incremented by 1 for each loop. Ycut (k) is the color separation data cutout position (nozzle upper end coordinates) at the scan number k.

  Here, as an example, in the case of four-pass printing having 16 nozzle rows and forming an image by four scans for the same main scanning recording area on the image, the color separation data cutout position Y coordinate Ycut The setting method of will be described. In general, in the case of 4-pass printing, as shown in FIG. 5, first, at the initial value k = 1 of the scan number, image formation is performed using only the nozzle lower end 1/4. At scan number k = 2, image formation is performed after the paper is fed by ¼ nozzle length with respect to scan number k = 1. Further, at scan number k = 3, image formation is performed after feeding the paper for 1/4 nozzle length with respect to scan number k = 2. The above operation is repeated to form a final output image. That is, in the case of performing four-pass printing and feeding a quarter of the nozzle length, as shown in FIG. 9, the recording of one main scanning recording area is completed at k = 1 to 7, and this is repeated. Thus, the final output image is formed. Therefore, when the scan number k = 1, the color separation data cutout position Ycut = −12.

When Ycut (k) is generalized, the color separation data cut-out position Ycut (k) is given by the following equation when the number of nozzle rows: Nzzl, the number of passes: Pass, and the scan number: k.
Ycut (k) = − Nzzl + (Nzzl / Pass) × k (7)
Next, the scan duty setting unit 105 refers to the scan duty setting LUT 106 and sets a data value for each scan from the image data of each color separation processing plane (step S104).

The scan duty setting LUT 106 is given in the form shown in FIG. 6 in the case of four passes. FIG. 6 shows an example of 16 nozzles and 4 passes. The inflection points of P1 to P4 are set for every 4 nozzles, and the duty division ratio for 16 nozzles obtained by linearly interpolating each inflection point is set as the scan duty. LUT. Here, the setting method of the numerical values of P1 to P4 is as follows.
P1 + P2 + P3 + P4 = 1.0 (8)

  The scan duty setting LUT 106 is not limited to the above setting method. For example, the inflection point may be set more finely or may be directly specified for each nozzle. The scan duty is set by the product of the scan duty setting LUT 106 and color separation data as shown in FIG. However, when the corresponding nozzle is outside the area of the image Y address, the scanning duty is set to 0. For example, when the scan number is k = 1, as shown in FIG. 8, the image Y address becomes negative at the upper 3/4 of the nozzle row, so the scan duty value = 0 is substituted, and the nozzle row lower end 1/4 is significant. Is assigned a valid value.

  Since the color separation data cutout position Ycut (k) is determined by the scanning number k, the scanning duty is determined as shown in FIG. 9 when the scanning number k = 1 to 7. Each scanning duty in FIG. 9 is determined by the product of the color separation data and the scanning duty setting LUT. Therefore, when the product is taken with the LUT while feeding paper, the scanning number k = 1 to 4 in the area 1 portion is 4. The total value of one raster formed by one scan is the same as the color separation data. Similarly in the areas 2, 3, and 4, the total value of one raster is the same as the color separation data.

  However, in the scan duty separation, when the color separation data value is smaller than the duty separation threshold D_Th3, image formation is not performed by four scans, and image formation is performed by one to three scans. More specifically, when the color separation data value is D_Th3 or less, the image is scanned three times, when the color separation data value is D_th2 or less, twice, and when the color separation data value is D_th1 or less, the image is scanned once. Let the formation take place.

  When the scan duty setting in the scan duty setting unit 105 is expressed in detail by an expression only for cyan C (X, Y), for example, it is given as follows. C (X, Y) is a scan duty at the address (X, Y), and S_LUT (Y) is a scan duty setting LUT at the address Y.

When C (nx, Ycut (k) + ny)> D_Th3 C_d (nx, ny) = C (nx, Ycut (k) + ny) × S_LUT (ny) (9)
When C (nx, Ycut (k) + ny) ≦ D_Th1, “scan number k = 1, 5,..., 4n + 1 (n is an integer of 0 or more)”
C_d (nx, ny) = C (nx, Ycut (k) + ny) (10)
"Scan number other than above"
C_d (nx, ny) = 0 (11)
When C (nx, Ycut (k) + ny) ≦ D_Th2, “scan number k = 1, 5,..., 4n + 1”
“Scan number k = 2, 6,..., 4n + 2”
C_d (nx, ny) = C (nx, Ycut (k) + ny) / 2 (12)
"Scan number other than above"
C_d (nx, ny) = 0 (13)
When C (nx, Ycut (k) + ny) ≦ D_Th3 “Scan number k = 1, 5,..., 4n + 1”
“Scan number k = 2, 6,..., 4n + 2”
“When scan number k = 3, 7,..., 4n + 3”
C_d (nx, ny) = C (nx, Ycut (k) + ny) / 3 (14)
"Scan number other than above"
C_d (nx, ny) = 0 (15)
In the above formulas (9) to (15), 0 ≦ nx <X size of the image and 0 ≦ ny <Nzzl.

  Similarly, Lc (X, Y), M (X, Y), Lm (X, Y), Y (X, Y), and K (X, Y) are also decomposed into scanning duty by the above formula.

  In this embodiment, since four passes are taken as an example, the Duty factorization of exceptions is set as in (10) to (15). However, for example, D_Th1 to D_Th7 are similarly set for eight passes, and Duty factorization of exceptions is performed. Do.

  Next, the scan duty setting unit 105 stores the set scan duty data in the scan duty buffer 107 (step S105). As shown in FIG. 10, the scan duty buffer 107 stores, for each color, band-shaped data values for vertical: number of nozzles and horizontal: image Xsize.

  Next, the halftone processing unit 108 performs halftone processing on the total value of the data in the scan duty buffer 107 and the data in the constraint information buffer 109 to convert it into a two-level gradation value (binary data) ( Step S106). As shown in FIG. 11, the constraint information buffer 109 also stores, for each color, band-like data values (constraint information) for vertical: number of nozzles and horizontal: image Xsize.

  In the first embodiment, in the constraint information buffer 109, constraint information indicating whether or not a binary image is easily formed at an address on a recorded image is updated for each scan number k. However, the initial value of all 0s is substituted at the start of the processing of the scan number k = 1. That is, the constraint information of each color at the address (X, Y) is C_r (X, Y), Lc_r (X, Y), M_r (X, Y), Lm_r (X, Y), Y_r (X, Y), K_r. Assuming that (X, Y), the scan number k = 1 is described as follows.

C_r (nx, ny) = 0 (16)
Lc_r (nx, ny) = 0 (17)
M_r (nx, ny) = 0 (18)
Lm_r (nx, ny) = 0 (19)
Y_r (nx, ny) = 0 (20)
K_r (nx, ny) = 0 (21)
“(0 ≦ nx <image X size) (0 ≦ ny <Nzzl)”
For this reason, significant constraint information is updated when the scan number k ≧ 2. As the value of the constraint information is smaller, the dot is less likely to be formed at the location, and conversely, the larger the value is, the easier the dot is formed. In addition, the value updated in the constraint information buffer stores an average value of 0 regardless of the scan number. That is, a positive value is stored when a dot is likely to be formed at that location, and a negative value is stored when a dot is difficult to form at that location. Details of the constraint information update method will be described later.

  The halftone process according to the first embodiment is a process for converting multi-valued input image data into a binary image (or an image having two or more values and a smaller number of gradations than the number of input gradations), for example, by R. Floyd et al. An error diffusion method (see Non-Patent Document 1) is used.

  Here, in order to simplify the description, the details of the cyan halftone process in four-pass printing and scan number k = 1 will be described with reference to the block diagram of the halftone processing unit 108 in FIG. 12 and the flowchart in FIG. To do.

First, in step S <b> 201, C_d that is cyan scanning duty and C_r that is cyan constraint information are input to the halftone processing unit 108. The adding unit 401 calculates the total value data Ic of C_d and C_r as follows.
Ic = C_d + C_r (22)
(However, the constraint information C_r of the scan number k = 1 is all 0)
In this embodiment, it is assumed that error diffusion coefficients for error diffusion processing have four coefficients K1 to K4 as shown in FIG. For example, K1 = 7/16, K2 = 3/16, K3 = 5/16, and K4 = 1/16. However, the diffusion coefficient is not fixed as described above, and may be changed according to the gradation (for example, according to C_d), or may be more than the above four coefficients. In order to diffuse and accumulate errors using such error diffusion coefficients, four sets of accumulated error line buffers in the halftone processing unit 108 are secured, and the accumulated error line buffers to be used are switched for each scan number.

In particular,
“When scan number k = 1, 5,..., 4n + 1 (n is an integer of 0 or more)”
(4n + 1) cumulative error buffer 402 is used “when scan number k = 2, 6,..., 4n + 2”
(4n + 2) cumulative error buffer 403 is used “when scan number k = 3, 7,..., 4n + 3”
(4n + 3) cumulative error buffer 404 is used “when scan number k = 4, 8,..., 4n + 4”
The accumulated error buffer 405 of (4n + 4) is switched to use.

  As shown in FIG. 15, the accumulated error buffers 402, 403, 404, and 405 have (a) “Ec1_0, Ec1 (x)”, (b) “Ec2_0, Ec2 (x)”, and (c) “ Ec3_0, Ec3 (x) ", (d)" Ec4_0, Ec4 (x) ".

  The (4n + 1) cumulative error buffer 402 has one storage area Ec1_0 and the same number of storage areas Ec1_ (x) (x = 1 to W) as the number of horizontal pixels W of the input image. The other accumulated error buffers 403, 404, and 405 also have the same recording area. Also, it is assumed that the accumulated error buffers 402, 403, 404, and 405 are all initialized with the initial value 0 only when the processing of the scan numbers k = 1, 2, 3, and 4 is started. For example, at the start of processing for scan number k = 5, (4n + 1) cumulative error buffer 402 is not initialized.

  In this embodiment, the above-described four sets of accumulated error line buffers are required for each color. That is, since this embodiment has 6 colors, a total of 4 × 6 = 24 sets of line buffers are required.

  In this description, since the scan number is k = 1, a case where error diffusion processing is performed using the (4n + 1) cumulative error buffer 402 will be described.

In step S202, the error Ec1 (x) corresponding to the horizontal pixel position x of the input pixel data is added by the cumulative error adding unit 406. That is, if the input data of interest is Ic and the data after adding the cumulative error is Ic ′,
Ic ′ = Ic + Ec1 (x) (23)
It becomes.

Next, in step S203, the threshold selection unit 407 selects the threshold T. In the selection of the threshold T, for example,
T = 128 (24)
Is set. Alternatively, in order to avoid dot generation delay, the average quantization error is reduced.
T = f (C_d) (25)
In this way, the threshold value T may be finely changed according to C_d.

Next, in step S204, the quantization unit 408 compares the pixel data Ic ′ after the error addition with the threshold value T, and determines the binarization result Out_c of the dots. The rules are as follows: That is,
When Ic ′ <T Out_c = 0 (26)
When Ic '≧ T
Out_c = 255 (27)
Next, in step S205, the error calculation unit 409 calculates a difference Err_c between the pixel data Ic ′ after the error addition to the target pixel Ic and the output pixel value Out_c.
Err_c (x) = Ic′−Out_c (28)
Calculate as follows.

Next, in step S206, the error diffusion unit 410 diffuses the error. That is, using the (4n + 1) cumulative error buffer 402, the error Err_c (x) is diffused as follows according to the horizontal pixel position x.
Ec1 (x + 1) ← Ec1 (x + 1) + Err_c (x) x 7/16 (x <W)
Ec1 (x-1) ← Ec1 (x-1) + Err_c (x) x 3/16 (x> 1)
Ec1 (x) ← Ec1_0 + Err_c (x) x 5/16 (1 <x <W)
Ec1 (x) ← Ec1_0 + Err_c (x) x 8/16 (x = 1)
Ec1 (x) ← Ec1_0 + Err_c (x) x 13/16 (x = W)
Ec1_0 ← Err_c (x) x 1/16 (x <W)
Ec1_0 ← 0 (x = W)
... (29)
This completes the binarization (quantization value 0, 255) for one cyan pixel of scan number k = 1.

  The processing of steps S201 to S206 described above is performed for addresses (0, 0) to (W-1, Nzzl-1) in the band (step S207), thereby determining the dot position of the halftone image data. Can do.

  When the scan number k = 2, the (4n + 2) cumulative error buffer 403 is used, when the scan number k = 3, the (4n + 3) cumulative error buffer 404 is used, and when the scan number k = 4, the (4n + 4) cumulative error buffer 405 is used. Process. In the process of scan number k = 5, the same (4n + 1) cumulative error buffer 402 as scan number k = 1 is used as it is without being initialized (all 0s are not substituted). This is because the print areas of scan number k = 1 and scan number k = 5 are vertically adjacent (see FIG. 16), so that the accumulated error is stored in the adjacent lower area. If the accumulated error buffer is initialized, the error is not stored at the adjacent boundary portion, and the dot continuity cannot be maintained.

  Next, the binary image data after the halftone process is stored in the halftone image storage buffer 110 (step S107). FIG. 17 is a diagram showing a state where the scan duty of scan number k = 1 is halftone processed and stored in the halftone image storage buffer. (Nzzl) × (W) binary image data corresponding to the pixel position of the scan duty is stored.

  As described above, the halftone process at the scan number k = 1 is completed. As a result, the binary image formed by one head operation for each color is stored in the halftone image storage buffer 110 for each color. It will be.

  Next, band data of vertical: Nzzl and horizontal image: W stored in the halftone image storage buffer 110 is output from the image output terminal 112 (step S108).

  Next, constraint information is calculated (step S109). Here, for the sake of simplicity of explanation, details of the calculation of the constraint information for cyan in four-pass printing, the scan number k = 1, and the number of nozzles Nzzl = 16 will be described with reference to the block diagram of the constraint information calculation unit 111 in FIG. This will be described with reference to the flowchart of FIG. As described above, the constraint information is information indicating whether or not dots are likely to be shot when determining the dot arrangement of the halftone image at the next scan number k = 2 after the current scan number k = 1. . When the current scan number is k, the next scan number is k + 1.

First, the scan duty filter processing unit 501 performs a filter process by applying a predetermined filter F_m to the cyan scan Duty and C_d in the scan duty buffer 107 and calculates C_df (step S301).
C_df = C_d * F_m (30)
*: Convolution FIG. 20 shows a schematic diagram of data filtered by the filter F_m.

  The constraint information calculation unit 111 according to the present embodiment sets the size (M, N) and coefficient of F_m based on the variation value var detected by the variation amount detection unit 207.

  For example, the filter size M × N is determined by the deviation σ_di of the landing position fluctuation value according to the carriage speed detected by the fluctuation amount detection unit 207. As shown in FIG. 26, the relationship between the fluctuation value var (= σ_di) and the filter size (M, N) may be set in advance. Alternatively, these relationships may be set as a lookup table (LUT). Further, this table may be changed for each resolution in the main scanning direction. As an example, FIG. 27 shows an isotropic load-averaging filter in which M = N = 5 squares and coefficients are arranged almost concentrically.

  At this time, the larger the filter size, the higher the robustness against fluctuations. Therefore, as shown in FIG. 26, the larger the fluctuation value var, the larger the filter size. Similarly, when the variation amount var is set to the maximum value d_max of the carriage variation amount di, it is desirable to increase the filter size as var increases. In order to satisfy this condition, for example, the filter size (N, M) may be set as a monotonically increasing function with the variation amount var as a variable, as in equations (31) and (32).

M = α _M · var + β _{M where} α _M > 0, β _M > 0 (31)
N = α _N · var + β _{N where} α _N > 0, β _N > 0 (32)
When the amount of position variation due to carriage speed variation is as close to 0 as possible, setting the filter size to 0 improves the graininess when there is no landing variation of the printing apparatus. However, since the fluctuation of the printing apparatus due to the image quality deterioration is not limited to the carriage fluctuation, it is not preferable to set N = M = 0 when the fluctuation value var = 0.

  Further, when the variation amount var is the carriage variation amount di (i = 0,... K−1), the variation amount detection unit may change the filter shape according to the position x at which the carriage position is detected. Good.

  For example, at a location where the speed is not stable at the start of scanning, there is a possibility that the fluctuation amount di is larger than the central portion of the image where the speed is stable. In that part, the filter size is increased to give priority to robustness against fluctuations, and in other parts, the filter size is made relatively small, so that higher image quality is given priority.

  The shape of the filter may be a square with M = N or a rectangle with M ≠ N. Since the carriage speed fluctuation causes a registration shift in the main scanning direction, it is effective to set a filter (M = 7, N = 5) that spreads in the main scanning direction, that is, N> M. . An example of the filter F_m in this case is shown in FIG. Note that the filter of FIG. 28 is an isotropic load-average filter in which coefficients are arranged substantially concentrically.

  Further, the filter coefficient may be an elliptical anisotropic filter, and is not limited to a low-pass characteristic, and may be a band-pass characteristic or a high-pass characteristic filter. In order to give directionality to the filter characteristics by the filter coefficient, if the long axis of the ellipse is set to be in the x direction, a filter that is robust against registration misalignment in the main scanning direction for the reason described above is used as constraint information. This is desirable. At this time, anisotropy is more easily expressed when the filter size is increased. FIG. 29 shows an anisotropic filter with N = M = 7 in which the major axis of the ellipse is in the x direction.

Next, the halftone data filter processing unit 502 performs filtering on Out_c using a predetermined low-pass filter LPF_b (step S302). FIG. 21 shows a schematic diagram of the filtered data.
Out_c_LPF = Out_c * LPF_b (33)
The coefficient of LPF_b in the present embodiment is also set by the fluctuation value var detected by the fluctuation amount detection unit 207, similarly to the aforementioned F_m. However, LPF_b preferably has a low-pass characteristic. In this embodiment, F_m and LPF_b are the same, but different coefficients may be used.

  Next, the scan duty data shift unit 503 shifts the data C_df of the scan duty filter processing unit 501 upward by one paper feed amount LF = Nzzl / Pass = 16/4 = 4 (step S303). FIG. 22 shows a schematic diagram of the shifted data. If the shifted scanning duty data is C′_df, the calculation is as follows. Note that 0 is substituted for the lower four nozzles after the shift.

C′_df (nx, ny) = C_df (nx, ny + LF) (34)
“(0 ≦ nx <image X size) (0 ≦ ny <Nzzl)”
“Substitute 0 when (ny + LF ≧ Nzzl) (substitute 0 for the lower end LF nozzle)”
Similarly, the halftone data shift unit 504 also shifts the data Out_c_LPF to the paper feed amount (step S304). FIG. 23 shows a schematic diagram of the shifted data. If the shifted data is Out'_c_LPF, the calculation is as follows.

Out′_c_LPF (nx, ny) = Out_c_LPF (nx, ny + LF) (35)
“(0 ≦ nx <image X size) (0 ≦ ny <Nzzl)”
“Substitute 0 when (ny + LF ≧ Nzzl) (substitute 0 for the lower end LF nozzle)”
Further, the pre-update constraint information data shift unit 505 also shifts the pre-update constraint information data C_r onto the paper feed amount (step S305). If the shifted scanning duty data is C′_r, the calculation is as follows.

C′_r (nx, ny) = C_r (nx, ny + LF) (36)
“(0 ≦ nx <image X size) (0 ≦ ny <Nzzl)”
“Substitute 0 when (ny + LF ≧ Nzzl) (substitute 0 for the lower end LF nozzle)”
As described above, in the constraint information calculation of the next scanning number, the paper feed amount LF is relatively shifted upward. Further, 0 is substituted for the lower end LF nozzle.

  The reason why the buffer data is shifted by the paper feed amount LF in this manner is that the halftone dot arrangement formed by the next scanning number is formed on the recording medium with a relative shift of the paper feed amount LF.

  Further, the subtraction unit 506 subtracts the data calculated by the halftone data shift unit 504 from the data calculated by the scanning duty data shift unit 503 (step S306). Then, the weight accumulation unit 507 accumulates the weight coefficient h (real number) on the subtraction result from the subtraction unit 506 (step S307).

Next, the adding unit 508 adds the data calculated by the weight integrating unit 507 and the cyan constraint information shifted by the pre-update constraint information data shift unit 505 (step S308), and the post-update constraint information C_r. And The updated constraint information C_r is stored as constraint information for halftone processing after the next scan number k = 2 (or the next scan number is k + 1 if the current scan number is k).
C_r ← (−Out′_c_LPF + C′_df) × h + C′_r (37)
Here, the equation (37) will be described. Out'_c_LPF in the equation is data obtained by extracting only a low frequency component from the halftone image of scan number k using LPF_b in the halftone data filter processing unit 502. In the equation, by subtracting from -Out'_c_LPF, the data from which only the low frequency component is extracted is set to a negative value. This “negative value” effect means that dots are less likely to be shot after the next scan number k + 1. That is, this process has an effect that the low frequency component of the halftone image printed before the scan number k after the next scan number k + 1 has an opposite phase.

  As described above, -Out'_c_LPF has an effect that the low-frequency component of the previously arranged dot has an opposite phase, but the meaning that only the low-frequency component has the opposite phase is described based on FIG.

  FIG. 24 is a diagram schematically showing a change in a print image with respect to registration shift (registration shift) divided into a high frequency region and a low frequency region. First, considering the high-frequency region of the printed image, when the density distributions of scan number k and scan number k + 1 are in opposite phases, the peaks and valleys of the density distribution are interpolated when there is no registration, and the dots fill the paper evenly. Therefore, it is difficult to generate white paper. For this reason, the density becomes high, but for example, even if it is slightly shifted (20 μm in the figure), the density distributions tend to overlap each other, and the white paper portion tends to be visible, and the density tends to be low. That is, when the printed image has an anti-phase with a high-frequency component, the density tolerance against the registration is lowered. However, when there is no correlation between the high-frequency components, the density is hardly changed and the graininess is hardly deteriorated even if there is a slight misregistration.

  On the other hand, in the low-frequency area of the printed image, the density distribution is in reverse phase, which has the effect of reducing low-frequency components that are visually unsightly, and more conveniently, even if there is some registration error Concentration tolerance is strong because the relationship between the upper and lower valleys is almost unchanged. However, when the density distribution is made uncorrelated, the low frequency component appears on the image regardless of whether or not the registration error occurs, and the graininess deteriorates.

To summarize the above, in order to increase the density unevenness resistance against the registration error on the printed image and improve the graininess,
・ Low-frequency components that are visually unsightly are in anti-phase.
・ High-frequency components should not be out of phase. Make it uncorrelated.
It becomes important.

  In the above example, only the scan numbers k and k + 1 have been mentioned, but the above characteristics are also valid between arbitrary scans.

  Subtracting Out'_c_LPF calculated by the halftone data shift unit 504 using the equation (37) has an effect of setting the low frequency component in the opposite phase, and further, does not consider the high frequency component as the constraint information. There is an effect of making the components uncorrelated.

  Further, attention is paid to C′_df in the equation (37). C′_df in Expression (37) is data obtained by filtering the Duty data at the scanning number k using F_m in the scanning duty filter processing unit 501. In the equation, C′_df is added, and there are two reasons.

  The first point is to always keep the constraint information C_r at 0 by setting the average value in parentheses to 0, so that the scanning duty density and the density of the halftone process are the same.

  This is because if only the data having a negative value at −Out′_LPF is treated as constraint information, Ic in Expression (22) becomes smaller than C_d, so that the number of dots in the halftone process is small. Because it becomes. Since the error diffusion method operates so that the input density and the output density are always the same, C_r must always be kept at an average of zero. Fortunately, since C′_df has the same density as Out′_c_LPF, C_r can always maintain an average of 0 by adding C′_df.

  The second reason is the control of the spatial frequency characteristics of the output image.

  As described above, the average value in parentheses in Equation (37) must be 0, but when only Out'_c_LPF is multiplied by LPF_b, Out'_c_LPF and C'_df in parentheses differ in terms of spatial frequency. As a result, edge enhancement occurs. In other words, if the subtraction is performed without performing the blurring process only on Out_c and without performing the blurring process on C_d, the process becomes equivalent to unsharp masking and edge enhancement occurs. Therefore, edge enhancement can be suppressed by applying the same filter to C_d as Out_c.

  However, when plain paper or uncoated paper such as matte paper is used as the recording medium, the outline of the image tends to be blurred, so it is not always necessary to perform the same processing on both sides. For example, the degree of spread of F_m may be narrowed.

  Furthermore, the strength of the constraint information can be adjusted with h. For example, when h = 1.0, the constraint information calculated by the filter can be output as it is, and when h = 0.0, an output equivalent to not calculating the constraint information is obtained. Since the average in parentheses is 0, the average value 0 is maintained no matter what value is entered for h.

  This completes the constraint information calculation in step S109.

  Next, the constraint information calculation unit 111 updates the constraint information buffer 109 with the calculation result (step S110). This restriction information is held as information for determining dot arrangements after the next scanning number (if the current scanning number is k, the next scanning number is k + 1).

  Here, the meaning of updating the constraint information buffer 109 will be described in detail with reference to the area 1 in FIG. In the present embodiment, the halftone process at the scan number k = 2 requires that the low frequency components of the dot pattern of the region 1 formed at the scan number k = 1 be in reverse phase. Therefore, as described in the equations (30) to (37), by shifting the filter processing result or the like at the scan number k = 1 by one paper feed amount LF = Nzzl / Pass = 16/4 = 4, The constraint information for the scan number k = 2 is created as described above.

  However, Expressions (30) to (37) are not limited to “turning the low frequency component in reverse phase” in the relationship of the scan numbers k = 1 and 2. For example, in the halftone process at the scan number k = 3, the low frequency component is reversed in phase with respect to the total dot pattern of the region 1 formed at the scan numbers k = 1 and 2. This is because, when the scanning number is incremented by 1 in Expressions (30) to (37), the constraint information buffer is always held while being shifted by one paper feed amount LF = Nzzl / Pass = 16/4 = 4. This is because the update is repeated.

  Similarly, the halftone process at scan number k = 4 has the effect of making the low frequency component in reverse phase with respect to the total dot pattern formed at scan numbers k = 1, 2, and 3.

  In other words, in this embodiment, the dot pattern for each scan is formed so that the low frequency component of the already recorded total dot pattern is always in the opposite phase in the same area on the recording medium. As a result, it is possible to ensure good graininess regardless of the occurrence of registration deviation and to reduce density unevenness due to registration deviation.

  Next, in the printer 2 that has received the halftone image data, the ink color and ejection amount selection unit 206 selects an ink color that matches the transferred image data, and the head control unit 204 starts the printing operation by the recording head 201. To do. Then, along with the start of the printing operation, the fluctuation amount detector 207 starts fluctuation amount detection (step S111).

  In the printing operation, the recording head 201 drives each nozzle at a constant driving interval while moving from the left to the right with respect to the recording medium. When the process is completed, the sub-scan, which is a scan in the vertical direction with the main scan, is performed once. In the fluctuation amount detection, as described above, the movement speed fluctuation amount of the recording head in the main scanning direction is detected. The detected fluctuation amount is sent to the constraint information calculation unit 111 in order to determine halftone image data to be recorded in the next scan.

  Next, it is determined whether or not all scanning has been completed (step S112). If completed, a series of image forming processes are completed, and if not completed, the process returns to step S103. Thus, all of the processing is completed.

  The process of the first embodiment has been described above. In the first embodiment, for the same color, as shown in Expression (22), the input data C_d reflects the constraint information obtained by extracting the low-frequency component of the dot pattern already formed at the same location on the recording medium. Thus, the low frequency component is antiphase and the high frequency component is uncorrelated.

  However, this embodiment is not the only method for realizing “low frequency components are in anti-phase” and “high frequency components are uncorrelated”. As another method, the constraint information obtained by extracting the low frequency component may be reflected in, for example, a threshold value or a quantization error. In this embodiment, the error diffusion method is used for halftone processing, but the average error minimum method may be used.

    In the above embodiment, the binarization calculation is performed in the halftone processing unit 108. However, when each recording element performs N-level multi-gradation recording, the halftone processing unit 108 performs the N-leveling process. It is clear that it will be done. This also applies to the following embodiments.

  As described above, according to the first embodiment, in the recording method in which the N-value calculation is performed for each main scanning, the amount of change in the moving speed of the recording head is the following main scanning in the same area on the recording medium. This is reflected in the N-value calculation. Therefore, it is possible to reduce registration shift due to changes in the recording apparatus and the recording medium, and to realize high image quality over the entire surface of the recording medium.

Second Embodiment
Next, a second embodiment will be described. In the first embodiment, the speed fluctuation amount of the recording head is used as the physical fluctuation amount between the recording head and the recording medium. In the second embodiment, the conveyance fluctuation amount of the recording medium is used. That is, in the second embodiment, the fluctuation amount detected by the fluctuation amount detection unit 207 is set as the conveyance fluctuation amount of the recording medium. Operations other than the constraint information calculation unit 111 and the fluctuation amount detection unit 207 are the same as those in the first embodiment.

  In the second embodiment, the fluctuation amount detection unit 207 performs sub-scanning, which is scanning in the direction perpendicular to main scanning, after the recording head 201 completes one main scanning for recording an image on a recording medium. Changes in the conveyance amount of the recording medium are detected.

  Examples of the sensor used for detection include a sensor that measures the reference position of the transport roller and the rotation angle of the transport roller, such as a mechanism described in Patent Document 5 (Japanese Patent Laid-Open No. 2001-180057). It is not limited to. In the present embodiment, it is assumed that the amount of deviation d_lf from the reference position in one transport operation (paper feed) is detected by the sensor.

  Next, the constraint information calculation (step S109) according to the second embodiment will be described. Here, for the sake of simplicity of explanation, details of the calculation of the constraint information for cyan in four-pass printing, the scan number k = 1, and the number of nozzles Nzzl = 16 will be described with reference to the block diagram of the constraint information calculation unit 111 in FIG. This will be described with reference to the flowchart of FIG. As described above, the constraint information is information indicating whether or not dots are likely to be shot when determining the dot arrangement of the halftone image at the next scan number k = 2 after the current scan number k = 1. (Note that when the current scan number is k, the next scan number is k + 1).

First, the scan duty filter processing unit 501 applies a predetermined filter F_m to the cyan scan Duty and C_d in the scan duty buffer 107, performs filter processing, and calculates C_df (step S301). FIG. 20 shows a schematic diagram of the filtered data.
C_df = C_d * F_m (38)
*: Convolution In the second embodiment, the size (M, N) and coefficient of F_m are set by the fluctuation value var detected by the fluctuation amount detection unit 207.

  For example, the filter size is determined by the conveyance variation amount d_lf of the recording medium detected by the variation amount detection unit 207. As shown in FIG. 26, the relationship between the fluctuation value var (= d_lf) and the filter size (M, N) may be set in advance. At this time, the larger the filter size, the higher the robustness against fluctuations. Therefore, as shown in FIG. 26, the larger the fluctuation value var, the larger the filter size. Alternatively, it may be set as an LUT. Furthermore, an LUT may be set for each resolution in the sub-scanning direction or for each pass of multi-pass printing. As an example, FIG. 30 shows an isotropic load-average filter F_m in which M = N = 3 squares and coefficients are arranged in a substantially concentric manner.

  In addition, since the conveyance fluctuation amount of the recording medium causes a registration shift in the sub-scanning direction, a filter having a spread in the sub-scanning direction, that is, M> N (for example, M = 5, N = 7) is set. It is effective. An example of the filter F_m in this case is shown in FIG. Note that the filter of FIG. 31 is an isotropic load-average filter in which the coefficients are arranged substantially concentrically.

  Further, the filter coefficient may be an elliptical anisotropic filter, and is not limited to a low-pass characteristic, and may be a band-pass characteristic or a high-pass characteristic filter. In order to give directionality to the filter characteristics by the filter coefficient, if the ellipse's major axis is set to be in the y direction, a filter that is robust against registration deviation in the sub-scanning direction is constrained for the reason described above. It is desirable because it becomes information. At this time, anisotropy is more easily expressed when the filter size is increased. FIG. 32 shows an anisotropic filter with N = M = 9 in which the major axis of the ellipse is in the y direction.

Next, the halftone data filter processing unit 502 performs filtering on Out_c with a predetermined low-pass filter LPF_b. (Step S302) FIG. 21 shows a schematic diagram of the filtered data.
Out_c_LPF = Out_c * LPF_b (39)
The coefficient of LPF_b in the second embodiment is also set by the fluctuation value var detected by the fluctuation amount detection unit 207, similarly to the aforementioned F_m. However, LPF_b preferably has a low-pass characteristic. In this embodiment, F_m and LPF_b are the same, but different coefficients may be used.

Next, the scan duty data shift unit 503 shifts the data C_df of the scan duty filter processing unit 501 upward by one paper feed amount LF = Nzzl / Pass = 16/4 = 4 (step S303). FIG. 22 shows a schematic diagram of the shifted data. If the shifted scanning duty data is C′_df, the calculation is as follows. Note that 0 is substituted for the lower four nozzles after the shift.
C′_df (nx, ny) = C_df (nx, ny + LF) (40)
“(0 ≦ nx <image X size) (0 ≦ ny <Nzzl)”
“Substitute 0 when (ny + LF ≧ Nzzl) (substitute 0 for the lower end LF nozzle)”
Similarly, the halftone data shift unit 504 also shifts the data Out_c_LPF to the paper feed amount (step S304). FIG. 23 shows a schematic diagram of the shifted data. If the shifted data is Out'_c_LPF, the calculation is as follows.
Out'_c_LPF (nx, ny) = Out_c_LPF (nx, ny + LF) (41)
“(0 ≦ nx <image X size) (0 ≦ ny <Nzzl)”
“Substitute 0 when (ny + LF ≧ Nzzl) (substitute 0 for the lower end LF nozzle)”
Further, the pre-update constraint information data shift unit 505 also shifts the pre-update constraint information data C_r onto the paper feed amount (step S305). If the shifted scanning duty data is C′_r, the calculation is as follows.
C′_r (nx, ny) = C_r (nx, ny + LF) (42)
“(0 ≦ nx <image X size) (0 ≦ ny <Nzzl)”
“Substitute 0 when (ny + LF ≧ Nzzl) (substitute 0 for the lower end LF nozzle)”
As described above, in the constraint information calculation of the next scanning number, the paper feed amount LF is relatively shifted upward. Further, 0 is substituted for the lower end LF nozzle.

  The reason why the buffer data is shifted by the paper feed amount LF in this manner is that the halftone dot arrangement formed by the next scanning number is formed on the recording medium with a relative shift of the paper feed amount LF.

  Further, the subtraction unit 506 subtracts the data calculated by the halftone data shift unit 504 from the data calculated by the scanning duty data shift unit 503 (step S306). Then, the weight accumulation unit 507 accumulates the weight coefficient h (real number) on the subtraction result of the subtraction unit 506 (step S307).

Next, the adder 508 adds the data calculated by the weight integrator 507 and the cyan constraint information shifted by the pre-update constraint information data shift unit 505 to obtain updated constraint information C_r (step S308). . The post-update constraint information C_r obtained in this way is stored as constraint information for halftone processing after the next scan number k = 2 (or, if the current scan number is k, the next scan number is k + 1).
C_r ← (−Out′_c_LPF + C′_df) × h + C′_r (43)
This completes the constraint information calculation in step S109.

As described above, according to the second embodiment, the change in the conveyance amount of the recording medium detected by the sensor is reflected in the N-value calculation in the subsequent main scanning in the same area on the recording medium. As a result, fluctuations between the recording apparatus and the recording medium, in particular, registration shift due to fluctuations in the recording medium conveyance can be reduced, and high image quality can be realized on the entire surface of the recording medium.
Hereinafter, a third embodiment according to the present invention will be described. In the first embodiment, the change in the moving speed of the recording head is reflected in the N-value calculation in the second embodiment, and the change in the transport amount of the recording medium is reflected in the N-value calculation. In the third embodiment, the fluctuation amount detected by the fluctuation amount detection unit 207 is both the fluctuation amount of the moving speed of the recording head and the conveyance fluctuation amount of the recording medium. Operations other than the constraint information calculation unit 111 and the fluctuation amount detection unit 207 are the same as those in the first embodiment.

  In the third embodiment, the fluctuation amount detection unit 207 detects the positional fluctuation amount due to the speed fluctuation of the recording head 201 in the main scanning direction and the main scanning after the recording head 201 records an image on the recording medium once. The amount of change in conveyance of the recording medium in the sub-scanning direction is detected.

  The fluctuation amount var (cr, lf) in the third embodiment is a pair of the fluctuation amount of the moving speed of the recording head 201 and the conveyance fluctuation amount of the recording medium. As described above, examples of the speed fluctuation amount of the recording head 201 include the deviation σ_cr of the positional fluctuation caused by the carriage, or its maximum value d_cr_MAX, and the deviation amount d_cr (i) detected every predetermined period. Further, the conveyance fluctuation amount d_lf (j) in the sub-scanning direction is exemplified as the conveyance fluctuation amount of the recording medium. However, the amount of variation is not limited to these.

  The constraint information calculation in the third embodiment will be described (step S109). Here, for the sake of simplicity of explanation, details of the calculation of the constraint information for cyan in four-pass printing, the scan number k = 1, and the number of nozzles Nzzl = 16 will be described with reference to the block diagram of the constraint information calculation unit 111 in FIG. This will be described with reference to the flowchart of FIG. As described above, the constraint information is information indicating whether or not dots are likely to be shot when determining the dot arrangement of the halftone image at the next scan number k = 2 after the current scan number k = 1. (Note that when the current scan number is k, the next scan number is k + 1).

First, the scan duty filter processing unit 501 performs a filter process by applying a predetermined filter F_m to the cyan scan Duty and C_d in the scan duty buffer 107 and calculates C_df (step S301). FIG. 20 shows a schematic diagram of the filtered data.
C_df = C_d * F_m (44)
*: Convolution In this embodiment, the size (M, N) and coefficient of F_m are set by the fluctuation value var (cr, lf) detected by the fluctuation amount detection unit 207.

For example, the filter size is set by the deviation σ_cr of the position fluctuation due to the carriage speed fluctuation and the fluctuation amount d_lf (j) in the conveyance of the recording medium in the sub-scanning direction. More specifically, the magnitude relationship between σ_cr and d_lf (j) is set so as to satisfy the expressions (45) and (46).
When σ_cr> d_lf (j): N> M (45)
When σ_cr ≦ d_lf (j): N ≦ M (46)
The reason for determining the filter size as described above is that fluctuations in the transport amount of the recording medium cause registration shift in the sub-scanning direction, and positional fluctuations due to carriage speed fluctuations cause registration deviation in the main scanning direction. is there. It is possible to reduce image quality degradation by increasing the filter size in the direction of large variation and increasing the robustness against mechanical variation. Note that the relationship between the fluctuation amount var (cr, lf) = (σ_cr, d_lf (j)) and the filter size (M, N) may be set. Another example is a method in which the relationship between both variables is set as a multidimensional LUT. Furthermore, the LUT may be set for each of the resolution in the main scanning direction, the resolution in the sub-scanning direction, and the number of passes in multi-pass printing. As an example, FIG. 33 shows an isotropic load average filter F_m in which M = 5 and N = 3 squares and coefficients are arranged in a substantially concentric manner.

  Furthermore, in the fourth embodiment, a filter coefficient is set based on the variation amount var (cr, lf) using a two-dimensional Gaussian filter such as Expression (47) as the filter F_m.

... (47)

However, F_m = F'_m / Sum_m (48)
Sum_m is the total value of the coefficients of F′_m.
In Equation (46), (σ _x , σ _y ) is a standard deviation of a two-dimensional normal distribution, and ρ is a correlation coefficient of x, y.

In the third embodiment, when the standard deviation (σ _x , σ _y ) is set so that the relationship between the variation var (cr, lf) = (σ_cr, d_lf (j)) satisfies the expressions (49) and (50). Good.
When σ_cr> d_lf (j): σ _x > σ _y (49)
When σ_cr ≦ d_lf (j): σ _x ≦ σ _y (50)
For example, FIG. 34 shows the filter F_m when σ _x = 1.0, σ _y = 2.5, ρ = 0, and M = N = 7. As shown in FIG. 34, when σ _x ≦ σ _y is satisfied, the filter coefficient has anisotropic characteristics. By using this filter, it is possible to increase the robustness against registration shift in the sub-scanning direction. In addition, the relationship between the standard deviation (σ _x , σ _y ) and the fluctuation amount var (cr, lf) = (σ_cr, d_lf (j)) may be set in advance using an LUT or the like.

Next, the halftone data filter processing unit 502 performs filtering on Out_c using a predetermined low-pass filter LPF_b (step S302). FIG. 21 shows a schematic diagram of the filtered data.
Out_c_LPF = Out_c * LPF_b (51)
The coefficient of LPF_b in the third embodiment is also set by the fluctuation value var (cr, lf) detected by the fluctuation amount detection unit 207 as in the case of F_m described above. However, LPF_b preferably has a low-pass characteristic. In this embodiment, F_m and LPF_b are the same, but different coefficients may be used.

Next, the scan duty data shift unit 503 shifts the data C_df of the scan duty filter processing unit 501 upward by one paper feed amount LF = Nzzl / Pass = 16/4 = 4 (step S303). FIG. 22 shows a schematic diagram of the shifted data. If the shifted scanning duty data is C′_df, the calculation is as follows. Note that 0 is substituted for the lower four nozzles after the shift.
C′_df (nx, ny) = C_df (nx, ny + LF) (52)
“(0 ≦ nx <image X size) (0 ≦ ny <Nzzl)”
“Substitute 0 when (ny + LF ≧ Nzzl) (substitute 0 for the lower end LF nozzle)”
Similarly, the halftone data shift unit 504 also shifts the data Out_c_LPF to the paper feed amount (step S304). FIG. 23 shows a schematic diagram of the shifted data. If the shifted data is Out'_c_LPF, the calculation is as follows.
Out'_c_LPF (nx, ny) = Out_c_LPF (nx, ny + LF) (53)
“(0 ≦ nx <image X size) (0 ≦ ny <Nzzl)”
“Substitute 0 when (ny + LF ≧ Nzzl) (substitute 0 for the lower end LF nozzle)”
Further, the pre-update constraint information data shift unit 505 also shifts the pre-update constraint information data C_r onto the paper feed amount (step S305). If the shifted scanning duty data is C′_r, the calculation is as follows.
C′_r (nx, ny) = C_r (nx, ny + LF) (54)
“(0 ≦ nx <image X size) (0 ≦ ny <Nzzl)”
“Substitute 0 when (ny + LF ≧ Nzzl) (substitute 0 for the lower end LF nozzle)”
As described above, in the constraint information calculation of the next scanning number, the paper feed amount LF is relatively shifted upward. Further, 0 is substituted for the lower end LF nozzle.

  The reason why the buffer data is shifted by the paper feed amount LF in this manner is that the halftone dot arrangement formed by the next scanning number is formed on the recording medium with a relative shift of the paper feed amount LF.

  Further, the subtraction unit 506 subtracts the data calculated by the halftone data shift unit 504 from the data calculated by the scanning duty data shift unit 503 (step S306). Then, the weight accumulating unit 507 accumulates the weight coefficient h (real number) on the empty subtraction result of the subtracting unit 506 (step S307).

  Next, the adding unit 508 adds the data calculated by the weight integrating unit 507 and the cyan constraint information shifted by the pre-update constraint information data shift unit 505 (step S308), and outputs post-update constraint information C_r. To do. The post-update constraint information C_r obtained in this way is stored as constraint information for halftone processing after the next scan number k = 2 (or, if the current scan number is k, the next scan number is k + 1).

  Next, the constraint information buffer 109 is updated with the calculation result (step S110). This restriction information is held as information for determining dot arrangements after the next scanning number (if the current scanning number is k, the next scanning number is k + 1).

As described above, according to the third embodiment, the fluctuations in the position fluctuation amount and the recording medium conveyance amount due to the carriage speed fluctuation detected by the sensor are converted into N values in the subsequent main scanning in the same area on the recording medium. It can be reflected in the calculation. For this reason, it is possible to achieve high image quality over the entire surface of the recording medium even when there are variations in the recording apparatus and a plurality of variations that cause registration shifts having different characteristics. <Fourth Embodiment>
The fourth embodiment according to the present invention will be described below. In the fourth embodiment, the fluctuation amount detected by the fluctuation amount detection unit 207 is the fluctuation amount of the distance between the recording head and the recording medium due to cockling in which the surface of the recording medium is displaced for each scan. Operations other than the constraint information calculation unit 111 and the fluctuation amount detection unit 207 are the same as those in the first embodiment.

  As a method for detecting the distance between the recording head and the recording medium, for example, as disclosed in Patent Document 3, a sensor irradiates the recording medium with laser light and detects the reflected light to measure the distance between the sheets. Of course, but not limited to this.

In the fourth embodiment, the variation amount detected by the variation amount detection unit 207 is the distance between the recording head and the recording medium. For example, as shown in FIG. 35, the sensor 3503 detects the distance d_p (x _k ) between the recording head 3501 and the recording medium 3502 at every predetermined position x _k (k = 0, 1,... K) in the scanning direction. Note that the distance d_p0 between the surfaces of the recording head 3501 and the recording medium 3502 when cockling or the like does not occur is measured in advance before starting the printing scan.

Note that the interval | x _k -x _k-1 | for detecting the distance between the recording head and the recording medium is preferably set so that it can be detected at an interval finer than the installation width of the platen that sucks the paper.

  Further, in FIG. 35, the position of the sensor is set to the head traveling direction side when the carriage scanning direction is changed from the left to the right in the figure, but the present invention is not limited to this.

  When performing reciprocal printing, as shown in FIG. 40, a sensor is also attached to the opposite side of the sensor 3503 shown in FIG. 35, and the distance between the recording head and the recording medium is detected. Also, as shown in FIG. 40, when the carriage scanning direction is from left to right in the figure, the sensor 4001 may be set to use the sensor 4002 when scanning from right to left.

  The process of the constraint information calculation unit 111 in step S109 in the fourth embodiment will be described. Here, for the sake of simplicity of explanation, details of the calculation of the constraint information for cyan in four-pass printing, the scan number k = 1, and the number of nozzles Nzzl = 16 will be described with reference to the block diagram of the constraint information calculation unit 111 in FIG. This will be described with reference to the flowchart of FIG. As described above, the constraint information is information indicating whether or not dots are likely to be shot when determining the dot arrangement of the halftone image at the next scan number k = 2 after the current scan number k = 1. (Note that when the current scan number is k, the next scan number is k + 1).

First, the scan duty filter processing unit 501 performs a filter process by applying a predetermined filter F_m to the cyan scan Duty and C_d in the scan duty buffer 107 and calculates C_df (step S301). FIG. 20 shows a schematic diagram of the filtered data.
C_df = C_d * F_m (55)
*: Convolution The filter F_m in the fourth embodiment will be described as an isotropic weighted average filter having a filter size of 3 × 3 squares and coefficients almost concentrically arranged as shown in FIG. 30, but is not limited to this form. is not. For example, the filter size may be a square such as 5 × 5, 7 × 7, or 9 × 9, or a rectangle such as 3 × 5, 5 × 7, or 5 × 9, or an anisotropic filter coefficient may be an ellipse. It may be a filter. Further, the filter is not limited to the low-pass characteristic, and may be a band-pass characteristic or a high-pass characteristic filter.

Next, the halftone data filter processing unit 502 performs filtering on Out_c using a predetermined low-pass filter LPF_b (step S302). FIG. 21 shows a schematic diagram of the filtered data.
Out_c_LPF = Out_c * LPF_b (56)
The coefficient of the LPF_b in the fourth embodiment is also described as an isotropic weighted average filter having a filter size of 3 × 3 squares and coefficients arranged substantially concentrically as shown in FIG. It is not limited to form. However, LPF_b preferably has a low-pass characteristic. In this embodiment, F_m and LPF_b are the same, but different coefficients may be used.

Next, the scan duty data shift unit 503 shifts the data C_df of the scan duty filter processing unit 501 upward by one paper feed amount LF = Nzzl / Pass = 16/4 = 4 (step S303). FIG. 22 shows a schematic diagram of the shifted data. If the shifted scanning duty data is C′_df, the calculation is as follows. Note that 0 is substituted for the lower four nozzles after the shift.
C′_df (nx, ny) = C_df (nx, ny + LF) (57)
“(0 ≦ nx <image X size) (0 ≦ ny <Nzzl)”
“Substitute 0 when (ny + LF ≧ Nzzl) (substitute 0 for the lower end LF nozzle)”
Similarly, the halftone data shift unit 504 also shifts the data Out_c_LPF to the paper feed amount (step S304). FIG. 23 shows a schematic diagram of the shifted data. If the shifted data is Out'_c_LPF, the calculation is as follows.
Out'_c_LPF (nx, ny) = Out_c_LPF (nx, ny + LF) (58)
“(0 ≦ nx <image X size) (0 ≦ ny <Nzzl)”
“Substitute 0 when (ny + LF ≧ Nzzl) (substitute 0 for the lower end LF nozzle)”
Further, the pre-update constraint information data shift unit 505 also shifts the pre-update constraint information data C_r onto the paper feed amount (step S305). If the shifted scanning duty data is C′_r, the calculation is as follows.
C′_r (nx, ny) = C_r (nx, ny + LF) (59)
“(0 ≦ nx <image X size) (0 ≦ ny <Nzzl)”
“Substitute 0 when (ny + LF ≧ Nzzl) (substitute 0 for the lower end LF nozzle)”
As described above, in the constraint information calculation of the next scanning number, the paper feed amount LF is relatively shifted upward. Further, 0 is substituted for the lower end LF nozzle.

  The reason why the buffer data is shifted by the paper feed amount LF in this manner is that the halftone dot arrangement formed by the next scanning number is formed on the recording medium with a relative shift of the paper feed amount LF.

  Further, the subtraction unit 506 subtracts the data calculated by the halftone data shift unit 504 from the data calculated by the scanning duty data shift unit 503 (step S306). Then, the weight accumulation unit 507 accumulates the weight coefficient h (real number) on the subtraction result obtained by the subtraction unit 506 (step S307).

Next, the adding unit 508 adds the data calculated by the weight integrating unit 507 and the cyan constraint information shifted by the pre-update constraint information data shift unit 505 (step S308) to obtain post-update constraint information C_r. . The post-update constraint information C_r obtained in this way is stored as constraint information for halftone processing after the next scan number k = 2 (or, if the current scan number is k, the next scan number is k + 1).
C_r ← (−Out′_c_LPF + C′_df) × h + C′_r (60)
In Expression (59), h is a variable for adjusting the strength of the constraint information. For example, when h = 1.0, the constraint information calculated by the filter can be output as it is, and when h = 0.0, an output equivalent to not calculating the constraint information is obtained.

In the fourth embodiment, the variable h in Expression (59) is set by the fluctuation value d_p (x _k ) detected by the fluctuation amount detection unit 207.

  As shown in FIG. 36, the cockling phenomenon is a phenomenon in which the paper is swept in the main scanning direction by the penetration and swelling of the ink. That is, due to cockling, the dots become dense at the peaks 3601 and the dots become plain at the valleys 3602.

  The reason for this will be described with reference to FIGS. In FIG. 37, the ink ejected from the nozzles of the recording head 3701 to the recording medium 3702 lands on the position 3703 when cockling has not occurred. However, when cockling has occurred as shown in the figure, it lands at a position 3704. FIG. 38 shows the landing position deviation when the scanning direction is opposite to that in FIG. The ink ejected from the nozzles of the recording head 3801 to the recording medium 3802 lands on the position 3803 when cockling has not occurred. However, as shown in the figure, when cockling occurs, it lands at a position 3804. As in the case of FIG. 37, it is easy for dots to be struck at locations close to the peak. As described above, when the reciprocating scanning is repeated, the dots become dense at the peak portion 3601 and the dots become plain at the valley portion 3602 due to the influence of cockling.

When the fluctuation value d_p (x _k ) at the position x _k is small, or when the difference Δd = | d_p-d_p (x _k ) | from the reference value d_p0 is large, the state of the recording medium at the position x _k is It becomes a peak 3601 in FIG. Therefore, in such a case, h is increased to make it difficult for dots to be hit.

On the other hand, when the fluctuation value d_p (x _k ) is large or when the difference Δd = | d_p−d_p (x _k ) | from the reference value d_p0 is small, the state of the recording medium at the position x _k is as shown in FIG. It becomes a valley 3602. Therefore, in such a case, h is reduced in order to make it easier to hit dots.

That is, the relationship between the difference value d_p (x _i ) (i = 0, 1,..., K,... K) and the reference value d_p0 Δd = | d_p-d_p (x _i ) | For example, the setting is made as shown in FIG. By setting in this way, it is possible to set constraint information that is robust against fluctuations between sheets due to cockling. However, h takes a value of 0.0 to 1.0. Note that the relationship between the fluctuation value d_p and the intensity adjustment variable h may be set in advance using an LUT.

  This is the end of the constraint information calculation in step S109 according to the fourth embodiment.

  As described above, according to the fourth embodiment, the fluctuation amount of the distance between the recording head and the recording medium detected by the sensor is reflected in the N-value calculation in the subsequent main scanning in the same area on the recording medium. Is possible. For this reason, fluctuations in the recording apparatus and the recording medium, in particular, registration shift due to cockling can be reduced, and high image quality can be realized over the entire surface of the recording medium.

<Fifth Embodiment>
The fifth embodiment according to the present invention will be described below.

  In the fifth embodiment, the fluctuation amount detection unit 207 detects the inclination of the recording head. Operations other than the constraint information calculation unit 111 and the fluctuation amount detection unit 207 are the same as those in the first embodiment.

  The fluctuation amount detection unit 207 detects the inclination of the recording head. FIG. 41 is a view of the recording head 4101 as viewed from the top of the head with the surface on which the nozzles are disposed facing down. In the fifth embodiment, the sensor 4102 and the sensor 4103 measure the distance (dx_a, dy_a) from the sensor at a predetermined point A (4104) of the recording head. Note that the distances (dx0_a, dy0_a) from each sensor in a situation where the recording head is not tilted are measured in advance. From the distance (dx_a, dy_a) from each sensor and the reference distance (dx0_a, dy0_a), the tilt amount (Δdx_a, Δdy_a) from the reference position in the xy direction of the recording head is calculated (FIG. 42). 42, the dotted line indicated by 4202 is the position of the recording head when there is no inclination. Similarly, for the predetermined point B (4107) of the recording head, the distance (dx_b, dy_b) is measured using the sensor 4105 and the sensor 4106, and the inclination amount (Δdx_b, Δdy_b) of the recording head from the reference position in the xy direction is measured. Is calculated. The inclination amounts (Δdx_a, Δdy_a) and (Δdx_b, Δdy_b) are values set by the following equations (61) to (64).

Δdx_a = | dx0_a-dx_a | (61)
Δdy_a = | dx0_a-dy_a | (62)
Δdx_b = | dx0_b-dx_b | (63)
Δdy_b = | dy0_b-dy_b | (64)
The detection is performed at the start position of each print scan, or after the scan (before the start of the next scan) at the standby position of the print head. The place where the detection is performed is not limited to the illustrated form as long as the distance sensor as shown in FIG. 41 can be installed.

  Alternatively, as shown in FIG. 43, the z-direction inclination may be detected with the direction perpendicular to the recording medium as the z-direction. As for the inclination of the recording head in the z direction, for example, as shown in FIG. 43, the distance from the sensor is detected at two locations on the front side (paper transport direction (y)) and the back side of the recording head, and Δd_z = d_zb-d_zf may be used as the fluctuation amount. As described above, the detection timing is performed at the standby position of the recording head at the start of one scan or after the scan (before the start of the next scan). The place where the detection is performed may be any place where the distance sensor can be installed.

  Next, the process of step S109 performed by the constraint information calculation unit 111 in the fifth embodiment will be described. Here, for the sake of simplicity of explanation, details of the calculation of the constraint information for cyan in four-pass printing, the scan number k = 1, and the number of nozzles Nzzl = 16 will be described with reference to the block diagram of the constraint information calculation unit 111 in FIG. This will be described with reference to the flowchart. As described above, the constraint information is information indicating whether or not dots are likely to be shot when determining the dot arrangement of the halftone image at the next scan number k = 2 after the current scan number k = 1. (Note that when the current scan number is k, the next scan number is k + 1).

  First, the scan duty filter processing unit 501 performs a filter process by applying a predetermined filter F_m to the cyan scan Duty and C_d in the scan duty buffer 107 and calculates C_df (step S301). FIG. 20 shows a schematic diagram of the filtered data.

C_df = C_d * F_m (65)
*: Convolution In the fifth embodiment, the size (M, N) and coefficient of F_m for each scan are set by the fluctuation value var detected by the fluctuation amount detection unit 207.

  Here, let us consider a case where the relationship between the tilt amounts is Δdx_a> Δdy_a as in the recording head 4401 in FIG. 44, that is, the tilt in the x direction is large. In this case, when ink is ejected using all nozzles in one scan and a ruled line is output, a ruled line having a large shift in the x direction is formed with respect to the accurate landing position (4403) (FIG. 44). . As can be seen from FIG. 44, the amount of deviation in the x direction increases toward the end of the nozzle.

  Therefore, when the recording head is tilted as shown in FIG. 44, it is necessary to make a dot arrangement that is more robust to variations as the nozzle ends. In other words, it is desirable to increase the filter used when generating the constraint information for determining the dot pattern to be hit at the end of the nozzle. That is, the filter F_m used in the equation (64) is selectively used as the filter F_m in which the size and coefficient are set according to the position of the nozzle used and the tilt (variation) amount of the print head.

  For example, in the recording head 4101 of FIG. 41, the filter F_m is set by the variation value var = (Δdx_a, Δdy_a, Δdx_b, Δdy_b) so that the size (M, N) or coefficient of the filter F_m satisfies the following condition. . For simplification, four-pass printing is used here, and the nozzle group used when scanning a predetermined area is divided into four (4501, 4502, 4503, 4504) as shown in FIG. Further, filters used when generating constraint information for determining a dot pattern to be hit using each nozzle group are F_m1, F_m2, F_m3, and F_m4. The respective filter sizes are (M1, M1), (M2, N2), (M3, M3), (M4, N4).

... (66)
... (67)

(conditions)
(1) When Δd_a> Δd_b (where Δd_a> 0, Δd_b> 0),
M1>M4> M2 ≧ M3 and N1>N4> N2 ≧ N3,
(2) When Δd_a ≦ Δd_b (where Δd_a> 0, Δd_b> 0),
M4 ≧ M1> M2 ≧ M3 and N4 ≧ N1> N2 ≧ N3,
(3) When Δdx_a> Δdy_a, M1> N1,
(4) When Δdx_a ≦ Δdy_a, M1 ≦ N1
(5) Similar to (3) and (4), if the amount of tilt in the x direction is relatively large compared to the amount of tilt in the y direction, Mk (k = 0, ... 4: number of passes) ≧ Nk. Further, when the amount of inclination in the y direction is large, Mk ≦ Nk.

  In order to satisfy these relationships, the relationship between the inclination amount var and the filter size (Mk, Nk) may be held as an LUT. Alternatively, it is possible to calculate and use the filter size (Mk, Nk) for each detection by using a function that satisfies the above-described conditions. Absent.

  Further, a filter coefficient is set based on the amount of variation var (cr, lf) using a two-dimensional Gaussian filter such as Expression (67) as the filter F_mk (k is the number of passes).

... (68)

However, F_mk = F'_mk / Sum_mk (69)
Sum_mk is the total value of the coefficients of F′_mk.
In equation (68), (σ x — _k , σ _{y — k} ) is a standard deviation of a two-dimensional normal distribution, and ρ is a correlation coefficient of x, y.

In the present embodiment, the standard deviation (σ _{x_k} , σ _{y_k} ) is changed _relative to the variation value var = (Δdx_a, Δdy_a, Δdx_b, Δdy_b) in the same manner as the conditions for setting the filter size (Mk, Nk) described above. It is good to set by.

For example, when the amount of inclination in the x direction is relatively larger than the amount of inclination in the y direction, σ _{x_k} ≧ σ _{y_k} is satisfied. In addition, when the amount of inclination in the y direction is large, by _setting σ _{x_k} ≦ σ _{y_k} , the filter coefficient can have anisotropic characteristics. If the filter coefficient has an anisotropic characteristic such as an elliptic filter, setting the ellipse's major axis to be in the y direction will restrict the constraint information using a filter that is robust against registration deviation in the sub-scanning direction. Is calculated. On the other hand, if the major axis of the ellipse is set to be in the x direction, the constraint information is calculated using a filter that is robust against registration shift in the main scanning direction. By using a filter that satisfies the above-described conditions, the robustness with respect to landing fluctuations due to the tilt of the recording head is increased, and deterioration of the output image can be reduced. Note that anisotropy is more easily expressed when the filter size is increased. The filter is not limited to the low-pass characteristic, and may be a band-pass characteristic or a high-pass characteristic filter.

Further, similarly to the relationship between the inclination amount var and the filter size (Mk, Nk) described above, the values of the inclination amount var and the standard deviation (σ _{x_k} , σ _{y_k} ) may be held as an LUT. Alternatively, using a function that satisfies the above-described conditions, the slope amount var may be calculated and used for each detection by calculating the standard deviation (σ _{x_k} , σ _{y_k} ). It is not restricted to these.

  Next, the halftone data filter processing unit 502 performs filtering on Out_c with a predetermined low-pass filter LPF_b. (Step S302) FIG. 21 shows a schematic diagram of the filtered data.

Out_c_LPF = Out_c * LPF_b (70)
The coefficient of LPF_b in the fifth embodiment is also set by the fluctuation value var detected by the fluctuation amount detection unit 207, similarly to the aforementioned F_m. However, LPF_b preferably has a low-pass characteristic. In this embodiment, F_m and LPF_b are the same, but different coefficients may be used.

  Next, the scan duty data shift unit 503 shifts the data C_df of the scan duty filter processing unit 501 upward by one paper feed amount LF = Nzzl / Pass = 16/4 = 4 (step S303). FIG. 22 shows a schematic diagram of the shifted data. If the shifted scanning duty data is C′_df, the calculation is as follows. Note that 0 is substituted for the lower four nozzles after the shift.

C′_df (nx, ny) = C_df (nx, ny + LF) (71)
“(0 ≦ nx <image X size) (0 ≦ ny <Nzzl)”
“Substitute 0 when (ny + LF ≧ Nzzl) (substitute 0 for the lower end LF nozzle)”
Similarly, the halftone data shift unit 504 also shifts the data Out_c_LPF to the paper feed amount (step S304). FIG. 23 shows a schematic diagram of the shifted data. If the shifted data is Out'_c_LPF, the calculation is as follows.

Out′_c_LPF (nx, ny) = Out_c_LPF (nx, ny + LF) (72)
“(0 ≦ nx <image X size) (0 ≦ ny <Nzzl)”
“Substitute 0 when (ny + LF ≧ Nzzl) (substitute 0 for the lower end LF nozzle)”
Further, the pre-update constraint information data shift unit 505 also shifts the pre-update constraint information data C_r onto the paper feed amount (step S305). If the shifted scanning duty data is C′_r, the calculation is as follows.

C′_r (nx, ny) = C_r (nx, ny + LF) (73)
“(0 ≦ nx <image X size) (0 ≦ ny <Nzzl)”
“Substitute 0 when (ny + LF ≧ Nzzl) (substitute 0 for the lower end LF nozzle)”
As described above, in the constraint information calculation of the next scanning number, the paper feed amount LF is relatively shifted upward. Further, 0 is substituted for the lower end LF nozzle.

  The reason why the buffer data is shifted by the paper feed amount LF in this manner is that the halftone dot arrangement formed by the next scanning number is formed on the recording medium with a relative shift of the paper feed amount LF.

  Further, the subtraction unit 506 subtracts the data calculated by the halftone data shift unit 504 from the data calculated by the scanning duty data shift unit 503 (step S306). Then, the weight accumulation unit 507 accumulates the weight coefficient h (real number) on the subtraction result of the subtraction unit 506 (step S307).

  Next, the adding unit 508 adds the data calculated by the weight integrating unit 507 and the cyan constraint information shifted by the pre-update constraint information data shift unit 505 (step S308). The addition result by the adding unit 508 obtained in this way is set as post-update constraint information C_r, and the constraint information of the halftone process after the next scan number k = 2 (or the next scan number is k + 1 if the current scan number is k). Save as.

C_r ← (−Out′_c_LPF + C′_df) × h + C′_r (74)
This completes the constraint information calculation in step S109.

  Next, the constraint information buffer 109 is updated with the calculation result (step S110). This restriction information is held as information for determining dot arrangements after the next scanning number (if the current scanning number is k, the next scanning number is k + 1).

  As described above, according to the fifth embodiment, the fluctuation amount of the installation position of the recording head detected by the sensor is reflected in the N-value calculation in the subsequent main scanning in the same area on the recording medium. As a result, it is possible to reduce registration deviation due to fluctuations in the recording apparatus, particularly the inclination of the recording head, and realize high image quality over the entire surface of the recording medium.

  As described above, according to each of the above-described embodiments, in the recording method in which the N-value calculation is performed for each main scan, the variation amount of the recording apparatus and the recording medium detected by various sensors is the same area on the recording medium. This is reflected in the N-value conversion operation in the subsequent main scanning. As a result, registration shift due to fluctuations in the recording apparatus and the recording medium is reduced, and high image quality can be realized over the entire surface of the recording medium.

  Further, according to each of the above embodiments, for each main scan, the update is performed in real time based on the N-value conversion result (N is an integer of 2 or more) in the previous main scan and the fluctuation amount of the printing apparatus and printing medium. N-ary processing based on the constraint information is performed. As a result, it is possible to suppress the graininess deterioration depending on the low frequency component of the image and form a high quality image.

  The object of the present invention can also be achieved by supplying a recording medium on which a program code of software is recorded to a system or apparatus, and reading and executing the program code stored in the recording medium by a computer of the system or apparatus. It is. Here, the software program code realizes the functions of the above-described embodiments in a computer (or CPU or MPU) of a system or apparatus. In this case, the program code itself read from the storage medium realizes the functions of the above-described embodiment, and the storage medium storing the program code constitutes the present invention.

  Examples of the computer-readable storage medium for supplying the program code include the following. That is, a flexible disk, hard disk, optical disk, magneto-optical disk, CD-ROM, CD-R, magnetic tape, nonvolatile memory card, ROM, DVD, or the like can be used.

  In addition to the functions of the above-described embodiment being realized by the computer executing the read program, the embodiment of the embodiment is implemented in cooperation with an OS or the like running on the computer based on an instruction of the program. A function may be realized. In this case, the OS or the like performs part or all of the actual processing, and the functions of the above-described embodiments are realized by the processing.

  Furthermore, the program read from the recording medium is written in a memory provided in a function expansion board inserted into the computer or a function expansion unit connected to the computer, so that part or all of the functions of the above-described embodiments are realized. May be. In this case, after a program is written in the function expansion board or function expansion unit, the CPU or the like provided in the function expansion board or function expansion unit performs part or all of the actual processing based on the instructions of the program.

1 is a block diagram illustrating a configuration of an image forming system according to an embodiment of the present invention. FIG. 3 is a diagram illustrating a configuration example of a recording head in the printer of the embodiment. 6 is a flowchart illustrating image forming processing in the embodiment. It is a figure which shows the detail of the input / output data in the color separation process part of embodiment. It is a figure which shows the outline | summary of the image formation by 16 nozzles and 4-pass printing in embodiment. It is a figure which shows an example of the Duty division ratio hold | maintained at the scanning duty setting LUT in embodiment. It is a figure which shows the outline | summary of the setting method of the scanning duty in embodiment. It is a figure which shows the outline | summary of the setting method of the scanning duty in embodiment. It is a figure which shows the outline | summary of the setting method of the scanning duty in embodiment. It is a figure which shows the band structural example of the scanning duty buffer in embodiment. It is a figure which shows the band structural example of the restriction | limiting information buffer in embodiment. It is a block diagram which shows the detailed structure of the halftone process part in embodiment. It is a flowchart which shows the halftone process in embodiment. It is a figure which shows an example of the error diffusion coefficient in embodiment. It is a figure which shows the storage area of each cyan accumulation error line buffer in embodiment. It is a figure which shows the mode of the adjoining of the printing area | region in embodiment. It is a figure which shows the example of storage of the halftone process result in embodiment. It is a block diagram which shows the detailed structure of the constraint information calculating part in embodiment. It is a flowchart which shows the constraint information calculation process in embodiment. It is a schematic diagram of the scan duty data after the filter process in the constraint information calculation of the embodiment. It is a schematic diagram of the halftone data after the filter process in the constraint information calculation of the embodiment. It is a schematic diagram of the scanning duty data after the shift in the constraint information calculation of the embodiment. It is a schematic diagram of the halftone data after the shift in the constraint information calculation of the embodiment. It is a figure which shows the resisting characteristic with respect to the frequency domain of the printed image in embodiment. FIG. 6 is a diagram illustrating a variation amount and a landing position deviation related to a moving amount of a recording head. It is a graph which shows the relationship between the filter size and variation | change_quantity in the constraint information calculation of embodiment. It is a figure which shows the specific example of the filter in the constraint information calculation of embodiment. It is a figure which shows the specific example of the filter in the constraint information calculation of embodiment. It is a figure which shows the specific example of the filter in the constraint information calculation of embodiment. It is a figure which shows the specific example of the filter in the constraint information calculation of 2nd Embodiment. It is a figure which shows the specific example of the filter in the constraint information calculation of 2nd Embodiment. It is a figure which shows the specific example of the filter in the constraint information calculation of 2nd Embodiment. It is a figure which shows the specific example of the filter in the constraint information calculation of 3rd Embodiment. It is a figure which shows the specific example of the filter in the constraint information calculation of 3rd Embodiment. It is a figure explaining the relationship between the structure of the paper gap sensor in 4th Embodiment, and the variation | change_quantity to detect. It is explanatory drawing of a cock ring. It is explanatory drawing of the dot position shift which generate | occur | produces by cockling. It is explanatory drawing of the dot position shift which generate | occur | produces by cockling. It is a graph which shows the relationship between the restriction | limiting intensity | strength variable and variation | change_quantity in the restriction | limiting information calculation in 4th Embodiment. It is a figure which shows another structure of the paper gap sensor in 4th Embodiment. It is a figure explaining the relationship between the structure of the inclination detection sensor in 5th Embodiment, and the variation | change_quantity to detect. It is a figure explaining the head inclination amount in 5th Embodiment. It is a figure explaining the relationship between another structure of the inclination detection sensor in 5th Embodiment, and the variation | change_quantity to detect. It is a figure explaining the relationship between the head inclination and the landing position of a dot in 5th Embodiment. It is a figure explaining the nozzle group in the restrictions information calculation in 5th Embodiment.

Claims (7)

An image processing apparatus for an image forming apparatus that forms an image by recording and scanning a recording head including a plurality of recording elements a plurality of times for the same region on a recording medium,
An acquisition means for acquiring information indicating the amount of registration variation between the recording head and the recording medium during a plurality of recording scans;
Generating means for generating scan data for each recording scan by dividing the input image data for each recording scan by dividing the pixel value of each pixel constituting the image data;
Setting means for setting constraint information indicating whether or not dots are easily formed for each pixel according to the amount of variation;
N-value conversion means for performing N-value conversion processing (N is an integer of 2 or more) with reference to the constraint information with respect to the scan data, and creating a dot pattern for the print scan,
About the m-th printing scan and the n-th (m and n are natural numbers, n> m) -th printing scan in the area of the printing medium,
The setting means includes a dot pattern corresponding to the m-th printing scan and the n-th printing scan based on a result of filtering using a filter for the dot pattern printed by the m-th printing scan. The N-value conversion means refers to the constraint information to be referred to for N-value processing on the scan data corresponding to the n-th print scan so that the dot pattern corresponding to is in the opposite phase in the low frequency region. To set,
The image processing apparatus, wherein the setting means increases the size of the filter as the variation amount increases . The image processing apparatus according to claim 1, wherein the acquisition unit detects a fluctuation amount related to a change in a moving speed of the recording head in a recording scan. The image processing apparatus according to claim 1, wherein the acquisition unit detects a fluctuation amount related to a conveyance amount of the recording medium in the conveyance of the recording medium performed during recording scanning by the recording head. The image processing apparatus according to claim 1, wherein the acquisition unit detects a fluctuation amount related to a distance between the recording head and the recording medium. The image processing apparatus according to claim 1, wherein the acquisition unit detects a variation amount of inclination of the recording head for each recording scan. An image processing method of an image forming apparatus for forming an image by performing a recording scan with a recording head including a plurality of recording elements a plurality of times on the same area of a recording medium,
An acquisition step in which the acquisition unit acquires information indicating a registration variation amount between the recording head and the recording medium while performing a plurality of recording scans;
A generating step of generating scan data for each recording scan by dividing the input image data into pixel values of each pixel constituting the image data for each recording scan;
A setting step in which the setting means sets constraint information indicating whether or not dots are easily formed for each pixel according to the amount of variation;
An N-value conversion step in which an N-value conversion unit applies an N-value conversion process (N is an integer of 2 or more) to the scan data with reference to the constraint information, and creates a dot pattern for the print scan. And
About the m-th printing scan and the n-th (m and n are natural numbers, n> m) -th printing scan in the area of the printing medium,
In the setting step, the dot pattern corresponding to the m-th print scan and the n-th print scan are based on the result of filtering using a filter for the dot pattern printed by the m-th print scan. The constraint information referred to N-value processing is set for the scan data corresponding to the n-th print scan so that the dot pattern corresponding to is in the opposite phase in the low frequency region,
The image processing method of the to Rukoto wherein increasing the size of the filter as the amount of the variation is large. A computer program for causing a computer to function as each unit of the image processing apparatus according to any one of claims 1 to 5 by being read and executed by a computer.