JP4543796B2 - Calibration apparatus, calibration program, and calibration method - Google Patents

Calibration apparatus, calibration program, and calibration method Download PDF

Info

Publication number
JP4543796B2
JP4543796B2 JP2004205068A JP2004205068A JP4543796B2 JP 4543796 B2 JP4543796 B2 JP 4543796B2 JP 2004205068 A JP2004205068 A JP 2004205068A JP 2004205068 A JP2004205068 A JP 2004205068A JP 4543796 B2 JP4543796 B2 JP 4543796B2
Authority
JP
Japan
Prior art keywords
noise
pattern
patch
region
printing apparatus
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Expired - Fee Related
Application number
JP2004205068A
Other languages
Japanese (ja)
Other versions
JP2006026943A (en
Inventor
義彦 松沢
Original Assignee
セイコーエプソン株式会社
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by セイコーエプソン株式会社 filed Critical セイコーエプソン株式会社
Priority to JP2004205068A priority Critical patent/JP4543796B2/en
Publication of JP2006026943A publication Critical patent/JP2006026943A/en
Application granted granted Critical
Publication of JP4543796B2 publication Critical patent/JP4543796B2/en
Application status is Expired - Fee Related legal-status Critical
Anticipated expiration legal-status Critical

Links

Images

Description

  The present invention relates to a technique for detecting noise based on image data.

In a printing apparatus such as an ink jet printer, calibration is performed by a process before product shipment or an end user operation so that printing can be executed with the least noise. In order to perform calibration, it is necessary to detect the occurrence of noise, but there are various types of noise generated in the printing apparatus, for example, some periodicity in the main scanning direction (direction perpendicular to the paper feed). A technique for detecting banding noise that accompanies this is known. (For example, see Patent Document 1)
JP 2003-219158 A

In the conventional technology described above, it is not possible to reliably detect noise to be detected, particularly banding noise. In addition, the noise detection accuracy varies significantly depending on the conditions for acquiring the noise detection target image.
That is, according to the conventional technique, since the one-dimensional Fourier transform is used to detect the banding noise, the reading result of the test patch is averaged in one direction to be one-dimensional. In this configuration, noise other than banding noise, for example, noise caused by dust attached to the test patch and noise during reading are equalized. Therefore, in one-dimensional Fourier transform, banding noise and other noise cannot be distinguished, and noise cannot be accurately detected and removed for any noise.

Furthermore, the main scanning direction and sub-scanning direction when reading the test patch and the main scanning direction and sub-scanning direction when printing the test patch may be relatively inclined. At this time, since the banding noise parallel to the main scanning direction is also inclined with respect to the main scanning direction at the time of reading, if the test patch reading results are averaged in one direction as described above, the influence of the banding noise is equalized. And its periodicity becomes ambiguous. Therefore, it becomes practically impossible to detect banding noise by Fourier transform.
The present invention has been made in view of the above problems, and is a case where noise is extracted reliably so as to remove irregularly (non-periodically) noise in a certain direction, and a tilt occurs. detectably to Ruki catcher calibration device reliably banding noise also an object to provide a calibration program and calibration methods.

  In order to achieve the above object, in the present invention, a patch printed on a recording medium is divided into smaller areas, and the distribution information of the recording material is compared for each area to extract an area having a specific distribution. When some non-periodic noise is included in the distribution information together with a value indicating the distribution of the recording material, the distribution information in the area including the non-periodic noise is different from the distribution information in the area not including the non-periodic noise. Therefore, by comparing the distribution information for each divided region and extracting a singular region having a singular distribution, a region including noise that occurs aperiodically can be extracted.

  Therefore, if this area is excluded and patch noise is detected based on the distribution information of the remaining areas, periodic noise such as banding noise is reliably detected without being affected by non-periodic noise. can do. Non-periodic noise includes noise caused by dust attached to the patch (noise formed by reading while dust is attached), noise during reading, and noise recorded randomly during patch printing. Etc.

  Here, the patch only needs to include noise to be detected by the noise detection unit, but it is preferable to determine and print the recorded content of the patch so that the noise is easily detected. The distribution information acquisition unit only needs to be able to acquire distribution information indicating the distribution of the recording material, and it is only necessary to acquire data reflecting the distribution of the recording material recorded on the patch.

  Therefore, the patch may be scanned by the optical reading device, or data indicating the distribution of the recording material in the patch may be generated in advance and the data may be acquired. In any case, when acquiring distribution information indicating the distribution of the recording material, it is difficult to acquire data that accurately indicates only the distribution of the recording material, and recording is performed as intended before calibration. It is not easy to generate a patch by accurately recording the material. Therefore, the distribution information includes the aperiodic noise described above. Of course, if it is a situation where periodic noise can occur, that noise is also included.

  The division information acquisition unit only needs to be able to acquire distribution information of a plurality of divided regions corresponding to regions smaller than the patch based on the distribution information. That is, it is only necessary to divide the patch into a plurality of areas and perform a process substantially equivalent to the process of acquiring distribution information in each area. Periodically generated noise is expected to be included in each divided area, so if you divide the area and analyze and compare each area, it is a singular area, that is, an area that contains non-periodic noise Can be extracted.

  It should be noted that various methods can be employed as a method of performing division by the divided region acquisition means, and the division may be performed by a straight line parallel to either the main scanning direction or the sub-scanning direction of the patch. , It may be divided by straight lines parallel to both. Of course, it is not indispensable to divide by a straight line parallel to the main scanning direction and the sub-scanning direction, but it is preferable to be parallel for easy processing. Further, since the noise period only needs to exist in a direction parallel to the boundary at the time of division, the width of the divided area may be periodic or aperiodic. It can be adopted.

  As a suitable example for effective analysis, a configuration in which the distribution in one direction of the recording material is averaged for each divided region can be employed. That is, noise that occurs periodically in a certain direction on a two-dimensional plane, such as banding noise, and does not have a period related to a direction orthogonal to this direction can be averaged in this orthogonal direction. The influence on the periodicity of the noise itself is small. On the other hand, if averaging is not performed, it is easy to be affected by reading errors and fine noise when acquiring individual distribution information.

  Furthermore, when the image data is composed of a plurality of pixels, if an analysis is performed for each pixel, a complicated cycle occurs for each pixel. In particular, an inkjet printer that records recording material grains is affected by the recording material grains that are not visible to the human eye. Therefore, by averaging the distribution in one direction for each divided area, these effects can be suppressed, and the above distribution is included in a state where information caused by periodic noise is reliably included while suppressing these effects. It becomes possible to acquire information.

  Here, it is important to average for each divided region. That is, when averaged over the entire patch as in the conventional example described above, the influence of periodic noise is equalized and the period becomes ambiguous. However, if averaging is performed for each divided region, averaging is performed in a smaller range, and even if the period becomes somewhat ambiguous, the influence is remarkably small, and the periodicity is surely included in the distribution information.

  In addition, in the configuration in which periodic noise is detected by the noise detection means and a region including non-periodic noise is extracted by the singular region extraction means, this noise is reflected in the distribution information to reflect the periodic noise period. It is preferable to divide in a direction substantially perpendicular to a direction in which is periodically generated. That is, it is preferable to divide so that this noise still periodically occurs in each divided region. As a result, while removing the non-periodic noise by removing the divided areas, the periodically generated noise can be detected by the noise detecting means.

  As a more specific example, the division may be performed so that an area obtained by cutting the patch in a direction substantially perpendicular to the banding noise generation direction when the patch is printed corresponds to the divided area. That is, banding noise is generally generated as a line parallel to the main scanning direction, and further, this line is periodically generated in the sub-scanning direction. Therefore, if the banding noise is generated in a direction perpendicular to the main scanning direction at the time of patch printing (that is, parallel to the sub-scanning direction), the periodicity of the banding noise remains in each divided area. The area can be divided. Therefore, it is possible to surely remove noise other than banding noise by the singular region extraction means and reliably detect the banding noise by the noise detection means.

  The size of the divided areas is preferably set so that each divided area includes noise other than the noise to be detected by the noise detecting means. According to such a configuration, it is possible to reliably remove noise other than noise to be detected by the noise detection means by extracting and removing the singular region. As a specific configuration for this purpose, various configurations can be adopted. For example, a typical dust size that can be attached to a recording medium on which a patch is printed and a typical noise generated when the above distribution information is acquired. It may be determined based on the size of.

  In other words, the dust adhering to the recording medium can have various sizes. If the influence of the dust to be removed and the dust is observed, the typical size of the dust to be removed can be specified. In the distribution information, the data size (number of pixels) for including garbage differs depending on the resolution set when acquiring the distribution information. Even in this case, the resolution is specified along with the size of the garbage. Then, the size of the divided area can be determined so that typical garbage is included in one divided area. That is, by setting the length of the divided area in the short direction to be longer than the size of the garbage, it is possible to set typical garbage to be included in one divided area.

  The same applies to typical noise generated when acquiring distribution information, and a typical noise magnitude (number of pixels) may be acquired at a certain resolution. That is, if the length of the divided area in the short direction is longer than the magnitude of the noise, the size of the divided area can be set so that typical noise is included in one divided area. Of course, the size in the short direction of the divided area may be determined so that both the typical garbage and the typical noise are included in one divided area. In this case, the larger one of the above-mentioned typical dust and typical noise is selected, and this is set to be shorter than the length of the divided area in the short direction. Note that the length in the short direction of the divided areas may be determined based on the magnitude of typical noise that may occur when the patch is printed. A typical noise level may be determined in advance by observing the noise.

  In the above configuration, it is assumed that a rectangular divided region is set. That is, since the patch is a two-dimensional image, the distribution information is also two-dimensional data, and this data is divided into rectangles to form divided areas. At this time, since noise having a periodicity is detected by the noise detection means, data having a sufficient length for analysis is secured in the direction in which the period is generated. Therefore, it is preferable to ensure a length in one direction sufficient to detect periodic noise. However, even in this case, since it is not necessary to consider the period in the periodic noise when determining the other length, it is sufficient to ensure data large enough to detect aperiodic noise. .

  As another example of an index for determining the length in the short direction of the divided area, a configuration that prevents the influence of the inclination of the recording medium at the time of reading the patch can be employed. That is, in the configuration in which the distribution information is acquired by reading a patch recorded on a recording medium by a scanner, tilt may occur when the recording medium is set in the scanner. More specifically, the main scanning direction and sub-scanning direction at the time of printing a patch and the main scanning direction and sub-scanning direction at the time of reading may be relatively inclined.

  Banding noise that is parallel to the main scanning direction at the time of printing and periodically generated in the sub-scanning direction is inclined with respect to the main scanning direction at the time of reading when the recording medium is inclined. When projected in the sub-scanning direction, the area occupied by the banding noise increases as the tilt angle increases. In particular, when the inclination is large or the banding noise period is short, the projection of the banding noise in the sub-scanning direction may overlap or continue depending on the inclination of the recording medium.

  However, since the patch is divided in the present invention, even if a considerably large inclination occurs when the recording medium is read, in most cases, the projection of banding noise in the longitudinal direction of each divided area (range in the longitudinal direction) ) Do not overlap. Therefore, banding noise exists in a certain period with respect to the longitudinal direction, and analysis based on this period can be performed. Therefore, in order to ensure that banding noise can be detected even when such an inclination occurs, the recording medium is present in each divided area in a state where the recording medium has an inclination during reading. What is necessary is just to determine so that the range (projection to a longitudinal direction) of the banding noise in the longitudinal direction may not overlap. A typical inclination may be determined in advance by, for example, an average of inclinations obtained by reading the recording medium a plurality of times.

  The singular region extraction means only needs to be able to extract a singular region based on the distribution information for each divided region, and can adopt a configuration in which the characteristics of each region are compared with the distribution characteristic value indicating the characteristic of the recording material distribution. is there. In other words, when extracting a singular region, it is only necessary to extract a region having a singular distribution compared to other regions in each divided region, and if a distribution characteristic value in each region is obtained, recording can be performed. It is possible to determine whether the material distribution is unique. As a result, it is possible to easily extract the unique region. The distribution characteristic value may be a value that reflects the manner in which the recording material is distributed. If this value is calculated, it can be easily determined whether or not the region is a specific region by a predetermined threshold value.

  As specific examples for calculating the distribution characteristic value, various configurations can be employed. For example, a configuration using a value obtained by converting distribution information into a spectrum with respect to a spatial frequency can be employed. According to the spectrum value, a value reflecting the distribution cycle of the recording material included in the divided area can be acquired regardless of the width of the divided area. The effect of periodic banding noise contributes in the same way for each divided region, but non-periodic noise makes the spectral value of the region containing the non-periodic noise different from the spectral value of other regions. Contribute. Therefore, it is possible to extract a region including non-periodic noise as a singular region by comparing spectral values. As a preferred example of calculating the distribution characteristic value using the spectrum value, a configuration in which the spectrum value is integrated with respect to the spatial frequency can be employed. According to such a configuration, the distribution characteristic for each divided region can be expressed by one value, and comparison for each region becomes very easy.

  Furthermore, a constant multiple of the standard deviation of the distribution characteristic value can be adopted as a suitable example of a threshold used when performing comparison for each region. That is, according to the standard deviation, it is possible to evaluate how far from the statistical average for a plurality of distribution characteristic values. Accordingly, if a constant multiple of the standard deviation is used as a threshold value as to whether or not the region is a singular region, a plurality of divided regions can be compared with each other and a region having a singular distribution can be easily extracted. Of course, various values can be adopted as the constant here, and various values such as a variance value can be adopted as a threshold in addition to the standard deviation.

  According to the above configuration, the singular region can be easily extracted, and the noise detection unit only needs to be able to detect noise included in the divided region excluding the extracted singular region. In the present invention, since periodic noise characteristics can be taken into account by dividing the patch, the noise detection means preferably detects the periodic noise. Again, it is preferable to use a value obtained by converting the distribution information into a spectrum with respect to the spatial frequency in order to detect periodic noise. If a spectral value obtained by conversion to spatial frequency or a value obtained by extracting the influence of periodic noise from the spectral value is used, a value (for example, an integral value) for evaluating the degree of occurrence of the periodic noise is obtained. Can do.

  Periodic noise is considered to cause a cause of the noise, and it is highly possible that the noise can be removed by calibration of the printing apparatus. In other words, if the occurrence of periodic noise can be evaluated, it is possible to grasp which setting suppresses noise most by performing printing with a plurality of settings in a printing apparatus that prints patches. In view of this, the noise can be suppressed by adjusting the setting so that the patch having the smallest value indicating the degree of occurrence of periodic noise is printed.

  The patch in the above configuration may be a patch that can include aperiodic and periodic noise, but it is preferable to employ a patch suitable as a patch for performing the calibration. In the sense of detecting noise, it is preferable to employ a single color and uniform patch in order to surely reveal whether it is noise or not.

  Furthermore, it is preferable that the patch includes a color in which noise is easily recognized by human eyes. The colors recognized by the human eye vary from person to person, but in general, achromatic colors are easier to recognize color differences (color differences) than chromatic colors. Therefore, if an achromatic color is included in the patch and noise is detected by the achromatic color, it is possible to detect noise corresponding to the perception by human eyes. Of course, it is preferable to include a color other than an achromatic color in the patch if it is a color in which noise is easily recognized.

  Moreover, you may employ | adopt other than a human as a reference | standard which judges whether noise is conspicuous. That is, depending on the type of printing apparatus and the type of recording material mounted on the printing apparatus, noise may be noticeable in a specific color. In such a case, a patch of a different color is printed for each printer type and each recording material type. As a result, it is possible to reliably detect noise regardless of the type of printing apparatus and the type of recording material. In determining this color, a plurality of colors may be printed for each printer type and each type of recording material, and a color that generates noise relatively frequently among them may be selected. Of course, the patch may be printed with only one color, or may be printed with a plurality of colors.

  Further, when printing one or more colors of patches, it is preferable to use recording materials of all colors that can be used in the printing apparatus. For example, all colors may be used for printing one color patch, or a recording material of all colors may be included in any one of two or more patches. That is, a specific color can greatly affect noise generation, so if you print a patch using all colors in a situation where such noise can occur, the noise will be reduced in any patch. Can be detected. Therefore, the generation of noise is not overlooked.

  As an example where a specific color greatly affects noise generation, the flying direction of an ink droplet flying from a specific nozzle for ejecting a specific color in an inkjet printer is different from the flying direction of an ink droplet from another nozzle. For example, there may be a case where one of the nozzles ejecting ink of a certain color has a different pitch from the other nozzles due to nozzle manufacturing errors. Of course, not only an ink jet printer but also a laser printer or the like, a specific color can greatly contribute to the generation of noise as long as it is configured to express multiple colors by combining a plurality of colors.

  Furthermore, it is possible to employ a configuration in which a patch including two colors, a color having a predetermined brightness and a color having a lower brightness, is printed. That is, a part that becomes unintentionally low in the high brightness color becomes noise, and a part that becomes unintentionally high in the low brightness color becomes noise. Therefore, if a high-lightness patch and a low-lightness patch are employed as the above-described patches, any noise can be detected. Here, as an example in which a part having a low lightness unintentionally occurs in a high lightness color, there is a case where recording materials are unintentionally overlapped and recorded. On the contrary, as an example in which a part having a high lightness unintentionally occurs with a low lightness color, the recording material is unintentionally recorded away and the ground color (usually white) of the recording medium is exposed. There are cases.

  Here, it is only necessary to print the patch with two different types of lightness, but it is preferable that the lightness of the two be separated from each other to some extent in order to make noise clearly appear for the colors of different lightness. For example, when the brightness value range is 0 to 100, it is possible to employ an example in which a patch having a brightness of 55 and a patch having a brightness of 35 are printed.

  Furthermore, a configuration may be adopted in which a plurality of patches are printed and these patches are arranged at substantially symmetric positions across the center position of main scanning in the printing apparatus. That is, the main scanning is normally in a direction perpendicular to the feeding direction of the printing medium, and it is ideal that the feeding amount is constant in the main scanning range, but the feeding amount is not constant in the main scanning range. This can cause noise. Therefore, in order to detect the occurrence of these noises, if the patches are printed at symmetrical positions with the center position of the main scan in between, and the noise occurrences of the two are compared, is the feed amount easy? Whether or not can be detected. As a matter of course, various causes other than the feeding amount of the recording medium are assumed as causes for generating different levels of noise at both ends in the main scanning direction.

  Here, in order to reliably compare whether or not the degree of noise generation differs depending on the position of main scanning, the colors of patches to be compared are the same color (that is, patches printed based on the same image data). It is preferable. In addition, the patch may be arranged at a substantially symmetrical position across the center position of the main scanning, but if it is considered that the difference in noise generation level occurs most at both ends of the main scanning, the patch is located as close to the end as possible. It is preferred to print the patch.

  Further, as a configuration example suitable for acquiring the distribution information of the recording material in the patch by the optical reading device, a configuration in which the light source emits light prior to the acquisition of the distribution information can be adopted to stabilize the light source. At this time, the light source may simply be emitted, or a preview scan (simple scan at a lower resolution than the main scan) normally provided in the optical reader may be performed. Of course, preview scanning and light emission from the light source may be performed a plurality of times. Furthermore, in order to reduce the influence of the reading variation of the optical reader as much as possible when acquiring the distribution information, averaging is performed by performing multiple readings, or the last or the last multiple reading results by performing multiple readings. It is possible to adopt a configuration for acquiring distribution information based on the above. Alternatively, reading may be performed a plurality of times, and a value for evaluating the noise may be calculated and averaged.

  In addition, as described above, the configuration for dividing the noise detection target data, comparing each divided region, extracting the singular region, and extracting the region including non-periodic noise in advance is other than the patch reading result. It is also possible to apply to. That is, image data including a plurality of pixels is acquired, divided, and a unique region is extracted by comparing each divided region. As a result, a region including non-periodic noise can be easily removed, and noise removal processing can be easily performed. Of course, it is also possible to employ a configuration in which noise (periodic noise or the like) included in the remaining regions is detected by further performing noise detection except for the divided regions extracted in this way.

  Of course, noise detection can be used during calibration. In other words, if the setting values of (variable) parameters that can be set in a plurality of states are changed in the printing apparatus, and the noise is detected by printing a patch using the plurality of parameters, the degree of noise generation in each patch is evaluated. can do. Therefore, calibration can be performed by setting the parameter with the least noise generation in the printing apparatus as the best parameter.

The present invention is a second pattern in which a plurality of first patterns including a plurality of patches arranged in the main scanning direction of the printing apparatus are arranged in the sub-scanning direction of the printing apparatus, and affects the generation of noise. Means for causing the printing apparatus to print a second pattern with different parameter settings for the first pattern on the recording medium, and the patches in each patch constituting the second pattern printed on the recording medium. Distribution information acquisition means for acquiring distribution information indicating the distribution of the recording material in a plurality of divided areas smaller than the patch, and a peculiar distribution compared to other areas based on the distribution information in each divided area Noise for detecting noise of each patch constituting the second pattern on the basis of distribution information in a divided area excluding the specific area and specific area extraction means for extracting the area Output means; and parameter setting means for summing up the noises of patches having the same parameters, identifying a parameter with the smallest summation result, and setting the printing apparatus to be driven by the identified parameter. As a calibration device.
Further, the present invention is a second pattern in which a plurality of first patterns including a plurality of patches arranged in the main scanning direction of the printing apparatus are arranged in the sub-scanning direction of the printing apparatus, and affects the generation of noise. Generated by scanning the second pattern printed on the recording medium, and means for causing the printing apparatus to print the second pattern with different parameter settings for each first pattern on the recording medium. Analysis target area extracting means for extracting, as an analysis target, an area for each patch constituting the second pattern from image data including a plurality of pixels, and area dividing means for dividing the analysis target area into a plurality of divided areas. Based on the image data for each divided region, a singular region extracting means for extracting a singular region having a singular distribution compared to other regions, and a divided region excluding the singular region Based on the distribution information of the image data, the noise detecting means for detecting the noise of each patch constituting the second pattern, and the noise of the patch having the same parameter are added together, and the parameter with the smallest sum result And a parameter setting means for setting the printing apparatus so as to be driven by the specified parameter.
By the way, such an apparatus may exist as a single apparatus, or may be used in a state of being incorporated in a certain device. However, the idea of the invention is not limited to this and includes various aspects. It is a waste. Therefore, it can be changed as appropriate, such as software or hardware. As embodied example of the spirit of the invention may be employed software and ing inventions of the device.

  Of course, the recording medium may be a magnetic recording medium, a magneto-optical recording medium, or any recording medium that will be developed in the future. The same is true for the replication stage of the primary replica and secondary replica. In addition, even when a part is software and a part is realized by hardware, the idea of the invention is not completely different, and a part is stored on a recording medium and is appropriately changed as necessary. It may be in the form of being read.

Further, in such an apparatus, when proceeding with the process according to the control procedure, it is natural that the invention exists at the basis of the procedure, and the method can also be applied. For this reason, the same effect is obtained in the method invention. Of course, it is possible to adopt a configuration corresponding to each claim in the subordinate format in the above program and method.

Here, embodiments of the present invention will be described in the following order.
(1) General configuration of the present invention:
(2) Configuration of calibration system:
(2-1) Computer configuration:
(2-2) Configuration of adjustment pattern:
(2-3) Configuration of calibration module:
(2-4) Printer configuration:
(3) Calibration process:
(3-1) Adjustment pattern printing process:
(3-2) Scan processing:
(3-3) Banding noise value acquisition processing:
(3-4) Best value extraction process:
(4) Other embodiments:

(1) General configuration of the present invention:
FIG. 1 is an explanatory diagram for explaining the outline of the present invention. In the figure, a printer 40 is an ink jet printer that can perform printing using a printing paper having a large size such as A2 size, and ejects ink droplets from a plurality of nozzles to record an image on the printing paper. In this embodiment, the feeding amount of printing paper and the nozzles used in the printer 40 are adjusted to suppress the occurrence of unintentional linear noise (banding noise) in the main scanning direction.

  For this reason, an adjustment system is formed by the printer 40, the computer 10 that controls the printer 40, and the scanner 30 at the time of adjustment. That is, the adjustment pattern including a plurality of patches is printed on the printing paper by the printer 40, the adjustment pattern is scanned by the scanner 30, and the banding noise is evaluated based on the scan result, thereby suppressing the banding noise. Get possible adjustment values.

  Here, in order to adjust to the optimum state with respect to the feeding amount of the printing paper and the selection of the nozzle to be used, which are factors that affect the banding noise, an adjustment pattern for each adjustment is used. That is, first, in order to adjust the feed amount, the feed amount is changed in the printer 40 and the same adjustment pattern is printed a plurality of times. If these adjustment patterns are captured by the scanner 30 and the banding noise states are compared between different feed amounts, the feed amount that can suppress the banding noise most can be acquired. Therefore, if the feed amount is set to the printer 40 so that the printing paper is fed with the optimum feed amount, the printer 40 can be driven with the feed amount that can suppress the banding noise most.

  Further, in order to grasp the optimum use nozzle in a state where the feed amount is set, the same adjustment pattern is printed a plurality of times by changing the nozzle used at the time of printing. If a similar analysis process is performed on the print result, a set of used nozzles that can suppress the banding noise most can be acquired. Therefore, if the printer 40 is set to execute printing using the obtained set of used nozzles, the printer 40 can be driven using the set of nozzles that can suppress the banding noise most. The present invention may be performed by the manufacturer of the printer 40 before shipping, or by the user of the printer 40.

  In the present embodiment, the scanner 30 reads the adjustment pattern, and the patch on the adjustment pattern is divided into a plurality of regions based on the read result. Then, the reading results are compared for each divided region, and when the ink distribution indicated by the reading result is a unique distribution compared to other divided regions, the region is excluded. Banding noise analysis is performed on the remaining divided areas after the exclusion. As described above, the non-periodic noise that is not included in the other areas is included in the divided areas having a unique distribution. Banding noise is noise generated with some periodicity. Therefore, by the above exclusion process, banding noise can be detected in a state in which non-periodic noise is removed in advance, and the influence of non-periodic noise can be eliminated and banding noise can be detected reliably. It becomes possible.

(2) Configuration of calibration system:
(2-1) Computer configuration:
Next, a configuration for realizing the above-described invention will be described in detail. FIG. 2 is a block diagram showing functions of the computer 10 by hardware and software. The computer 10 includes a program execution environment such as a CPU 11, a RAM 12, and a ROM 13, and can read various programs recorded in the ROM 13 and the HDD (hard disk drive) 14 into the RAM 12 and execute them. In this embodiment, the program includes an OS (not shown), a PRTDRV (printer driver) 20, a scanner DRV (driver) 30a, and the like, and various programs can be activated under the execution of the OS.

  The computer 10 is connected to the scanner 30 and the printer 40 via the USB I / F 15, and is connected to the display 17, the keyboard 16 a, the mouse 16 b, and the scanner 30 via an I / F (not shown). That is, a predetermined image can be output to the display 17 and various input operations can be accepted via the keyboard 16a and the mouse 16b. As described above, the computer 10 can implement the present invention by using hardware generally provided in a general-purpose computer and executing software described later.

  The PRTDRV 20 is a program that generates data for causing the printer 40 to print an image or characters to be printed. In the present embodiment, the PRTDRV 20 includes a module for calibration. That is, the PRTDRV 20 includes an image processing unit 21 and a calibration module 22. The image processing unit 21 acquires image data indicating an image to be printed, and performs predetermined image processing to generate print data.

  That is, the image data is subjected to resolution conversion processing, color conversion processing, halftone processing, and the like, and print data for causing the printer 40 to execute printing is generated. The generated print data is delivered to the printer 40 via the USB I / F 15. The printer 40 performs printing based on the print data as described later. At this time, the image processing unit 21 performs the above-described image processing according to the printing conditions. For example, resolution conversion processing is performed in order to execute printing at the instructed resolution, and color conversion is performed with reference to a color conversion table corresponding to the instructed printing paper.

  In the present embodiment, in order to print an adjustment pattern to be measured at the time of calibration, adjustment pattern data 14a indicating an image of the adjustment pattern and print condition data 14b indicating a print condition set at the time of adjustment are stored in the HDD 14 in advance. It is recorded. That is, the image processing unit 21 refers to the printing condition data 14b and performs image processing on the image data indicated by the adjustment pattern data 14a under the conditions. By outputting this result to the printer 40, the printer 40 prints the adjustment pattern.

  The print condition data 14b only needs to indicate the print condition to be adjusted, and various print conditions that can be set when applying the calibration result can be adopted as the print condition data 14b. For example, when applying a calibration result for each printing paper used at the time of printing, data indicating a plurality of printing paper types is included in the printing condition data 14b, and when a calibration result is applied for each resolution. The print condition data 14b includes data indicating a plurality of print resolutions.

(2-2) Configuration of adjustment pattern:
In FIG. 3, the adjustment pattern employed in this embodiment is shown in the upper part. This pattern includes a plurality of rectangular patterns P 1 to P 5 that are long along the longitudinal direction (main scanning direction) of the printing paper. In each of the rectangular patterns P 1 to P 5 , a set value (parameter value) to be calibrated is set. ) Are different (in FIG. 3, the parameters are different as # 1 to # 5). Further, the left and right ends of each of the rectangular patterns P 1 to P 5 are formed by gray (Gr) patches, and are printed by image data for uniformly printing gray of a predetermined brightness.

Violet (V) patches are arranged side by side for each of the gray patches. This patch is printed with image data for uniformly printing violet having a predetermined brightness. In the present embodiment, the gray patch and the violet patch have different brightness, the former having a higher brightness and the latter having a lower brightness. Each of the rectangular patterns P 1 to P 5 includes two gray patches and two violet patches, and the patches are arranged in the main scanning direction as shown in the upper part of FIG. However, in each of the rectangular patterns P 1 to P 5 , since the parameters to be calibrated are different as described above, the occurrence of banding noise may differ depending on the parameters.

  Further, gray is the easiest color change for human eyes, and in this sense, it is a preferred color for analyzing the occurrence of banding noise. Violet is a color in which banding noise is likely to be generated by the printer 40. In this sense, violet is a preferable color for analyzing the occurrence of banding noise. Of course, the color of the patch employed as the adjustment pattern may be any color that is preferable for detecting banding noise, and various other colors can be employed.

  Further, in the present embodiment, as described later, a plurality of colors of ink can be mounted on the printer 40, and all of the plurality of colors are used at least once by a gray patch and a violet patch. That is, any color ink is included in one or both of the gray patch and the violet patch. In this embodiment, patches used for banding noise analysis are not printed between violet patches, but other patches may be printed and used for banding noise detection. In addition, since the ink flying state can be stabilized by continuing to eject ink from the print head, some patches are printed throughout the main scanning direction even when not used for banding noise analysis. Is preferred.

(2-3) Configuration of calibration module:
The calibration module 22 mainly includes a module for analyzing the occurrence of banding noise, and includes a parameter setting unit 22a, a best value extraction unit 22b, a banding noise value acquisition unit 22c, a singular region removal unit 22d, and an FFT unit. 22e, a weighting processing unit 22f, a division processing unit 22g, and a patch cutout unit 22h. The parameter setting unit 22a is a module for setting various parameters for the printer 40 by outputting a predetermined control signal via the USB I / F 15.

  That is, in the present embodiment, an image based on the same adjustment pattern data 14a is printed while changing the parameter, and the best parameter value is set according to the banding noise state of the obtained printed matter. Therefore, the data indicating the parameter and the data indicating the changed value are recorded in the HDD 14 as the parameter data 14c. In the present embodiment, the calibration target is the feed amount and the used nozzle, and the feed amount is recorded with data indicating a plurality of feed amounts as changed values. The data indicating the used nozzle is data for designating a set of a plurality of nozzles.

  FIG. 4 is an explanatory diagram for explaining the parameter data for the nozzles used in the present embodiment. This figure schematically shows the arrangement of nozzles formed on the nozzle surface of a print head 47a mounted on the printer 40. FIG. The print head 47a shown in the figure is reciprocated in the main scanning direction shown in the figure by the main scanning during printing, and the printing paper relatively moves in the sub-scanning direction shown in the figure by paper feed. A plurality of (for example, 180) nozzles Nz for ejecting ink of the same color are arranged in the sub-scanning direction on the nozzle surface of the print head 47a, and this nozzle row corresponds to 7 ink colors. A line is formed.

  The parameter data in the present embodiment is data that specifies that the number of nozzles used in each column is fixed and nozzles at different positions in the sub-scanning direction are used. For example, as in the example shown in the figure, a set of 140 nozzles is used, and a set in which the number of nozzles is shifted by 10 in the sub-scanning direction is used, and nozzles such as parameters # 1 to # 5 are designated. . In any case, the parameter setting unit 22a specifies the calibration target and sets the parameter value. Thereby, the same image data is set to be printed with different parameters.

In this embodiment, the parameter setting unit 22a sets parameters, the image processing unit 21 prints an adjustment pattern, and the parameter setting unit 22a changes the parameters. The rectangular patterns P 1 to P 5 shown in FIG. 3 are printed by repeating the process. In addition, when the adjustment pattern is printed, there are two methods, that is, when only the parameter relating to the feed amount is changed and when only the parameter relating to the nozzle used is changed.

  The scanner DRV 30a is a driver that controls the capturing in the scanner 30. The scanner DRV 30a reads the reading target placed on the document table in accordance with the reading conditions such as the resolution set by the keyboard 16a and the like, and acquires the image data 14d. . In the present embodiment, the image data 14d is data in which an image is expressed by a plurality of images and the color of each image is expressed by gradation values for each color component of RGB (red, green, blue).

  In this embodiment, when the calibration of the printer 40 is performed by the calibration module 22, the scanner DRV 30a is activated, and scanning is performed by setting a reading condition corresponding to the printing condition for printing the adjustment pattern. For example, it is set to a reading resolution preferable for scanning adjustment pattern data printed under each printing condition. That is, in order to increase the accuracy of detecting banding noise, it is preferable to perform high-resolution reading. However, if the resolution is excessively high, the amount of data is excessively increased and the processing speed is decreased.

  If the reading resolution is close to the printing resolution (for example, within a range of ± 10 dpi), moire may occur. Therefore, it is preferable to set a preferable reading resolution in consideration of accuracy and processing speed, or to set the reading resolution to prevent the occurrence of moire. When reading is performed with a CCD, one pixel is often generated from the reading results of a plurality of CCDs. However, one pixel is generated from an integer number or (integer / 2) number of CCDs. Is preferred. For example, a configuration in which the reading resolution is set so that image data for two pixels is generated from five CCDs may be employed.

  In the above example, when analyzing the image data 14d, a configuration for accurately grasping the calibration target is adopted. That is, in the adjustment pattern shown in FIG. 3, information indicating a change target (feed amount or used nozzle in this example) when the adjustment pattern is printed is described at the upper left of the adjustment pattern. More specifically, these pieces of information are indicated by the position and number of the character “I”, and the scanner DRV 30a refers to the portion corresponding to these characters in the image data 14d to acquire the calibration target. Information indicating the calibration target is transferred to the calibration module 22 and is referred to when setting the best parameter value. Further, in the present embodiment, in order to eliminate reading variations by the scanner 30, a plurality of scans are performed as described later.

The patch cutout unit 22h performs processing for cutting out the patch for each color and converting the color system based on the image data 14d read as described above. That is, the image data 14d is acquired from the HDD 14, and data having the number of pixels corresponding to each patch is extracted. As a result, image data for each patch is extracted, for example, image data corresponding to the Gr patch in the rectangular pattern P 5 of parameter # 5 is extracted as shown in the center left of FIG.

Here, it suffices if patches of different colors can be cut out from the plurality of rectangular patterns P 1 to P 5, and in addition to a configuration in which patches are cut out with a predetermined number of pixels, gradation values for each color component of RGB are predetermined. Various configurations such as cutting out data within the range of can be adopted. Of course, it is possible to perform various incidental processes such as eliminating image data at the boundary between patches.

The patch cutout unit 22h converts the color system in the image data of each cut out patch and acquires the lightness (L * value) in each pixel. That is, typical banding noise is a linear part that occurs unintentionally in the main scanning direction (left-right direction in FIG. 3) in the image as shown in the center right of FIG. 3, and is different from the surrounding images. This occurs when colors (bright or dark) continue in the main scanning direction. Therefore, in this embodiment, banding noise is analyzed based on the brightness in the image.

  When the brightness data for each patch is acquired, the division processing unit 22g divides the brightness data. This division is performed by dividing each patch cut out for each color by a line parallel to the sub-scanning direction. For example, the patch for each color shown in the center left of FIG. 3 is implemented by grouping together a fixed number of pixels (for example, 16 pixels) in the main scanning direction. In the present embodiment, the area is divided into about 32 areas. The data obtained by the division corresponds to lightness data of a rectangular area that is long in the sub-scanning direction as shown in the center right of FIG. In this specification, each area is referred to as a divided area.

In the present embodiment, since processing is performed collectively for each divided area, the division processing unit 22g calculates the average brightness in the divided areas. Here, the purpose is to detect banding noise. Since the banding noise is substantially parallel to the main scanning direction, the brightness of the pixels arranged in the main scanning direction is averaged. That is, the brightness of 16 pixels shown in the center right of FIG. 3 is averaged, and the relationship of the average brightness with respect to the position in the sub-scanning direction is calculated for each divided region. The relationship of the L * value with respect to the position calculated in this way is a relationship like the graph shown in the lower part of FIG. In this graph, the horizontal axis is the position (unit: 0.1 mm), and the vertical axis is the L * value.

The weighting processing unit 22f is a module that performs weighting on the obtained brightness value. The weighting here is a process for further emphasizing a change in brightness that may occur due to banding noise. Specifically, various processes can be employed. For example, in a high-lightness patch (Gr in FIG. 3), a high-frequency component of lightness L * is obtained based on the insight that a low-lightness line is likely to be banding noise. Emphasizes lightness that is smaller than the overall change trend obtained. Detailed processing will be described later.

  In any case, when the lightness for each divided area is obtained, the FFT unit 22e (FFT: Fast Fourier Transformation) performs Fourier transform. That is, the FFT unit 22e is a module that performs Fourier transform on input data, acquires the weighted lightness for each divided region, and performs Fourier transform for each divided region.

  An example of the result of the Fourier transform is shown in the upper part of FIG. In the figure, the horizontal axis represents the spatial frequency, the vertical axis represents the power spectrum value, and the Fourier transform result is indicated by a broken line. The singular region removing unit 22d compares the Fourier transform results for each divided region, and a region (singular region) in which the Fourier transform result is significantly different from the Fourier transform results in other divided regions due to the influence of dust and noise attached to the recording medium during scanning. (Referred to as not to be used for analysis). Details of this processing will be described later.

  The banding noise value acquisition unit 22c is a module that acquires a value (referred to as a banding noise value) indicating the degree of occurrence of banding noise from the Fourier transform value for each divided region after the singular region is excluded. In the present embodiment, the background is removed from the power spectrum value shown in the upper part of FIG. 5 so that the influence of the banding noise is more clearly manifested to obtain the banding noise value. Details of the processing will be described later.

  As described above, the process of calculating the banding noise value is performed for each patch (in the example shown in FIG. 3, there are two gray and violet patches for each of the parameters # 1 to # 5, 20 in total). A banding noise value is calculated for each patch. The best value extraction unit 22b refers to these banding noise values and acquires the best parameters. In this embodiment, the calibration target is the feed amount and the used nozzle, and the calibration target parameter cannot be changed within one main scan when the printer 40 is driven.

Therefore, in the present embodiment, different parameters cannot be set on the left and right sides of the main scan, and parameters that suppress the banding noise as a whole should be set in view of the banding noise values on the left and right sides of the main scan. . Also, it is impossible to set different parameters for each color, and parameters for which banding noise is suppressed as a whole in each color patch should be set. For example, in the rectangular patterns P 1 to P 5 shown in FIG. 3, even if patches in different rectangles on the left and right of the main scan have the best banding noise value, parameter # 1 on the left of the main scan and the right of the main scan Thus, it is not possible to drive the printer 40 by setting the parameter # 2 or the like.

  Therefore, the best value extraction unit 22b refers to each banding noise value, compares the banding noise for each of the left and right of the main scanning and each color, and selects a parameter with the lowest banding noise. When the best value extraction unit 22b extracts the best parameter, the parameter value is transferred to the parameter setting unit 22a, and the parameter setting unit 22a sets the best parameter value to the printer 40 via the USB I / F 15. As a result, in the printer 40, printing can be executed using the best parameter value, and printing can be executed in a state where generation of banding noise is suppressed.

  In the present embodiment, the calibration targets are the feed amount and the used nozzle, and the adjustment pattern is printed for each. Therefore, the above-described processing is performed at least twice. Needless to say, when adjusting each calibration target, the parameter fluctuation range may be increased and coarsely adjusted for the first time, and after the coarse adjustment, the fluctuation range may be reduced and adjustment may be performed.

(2-4) Printer configuration:
FIG. 6 is a block diagram showing functions of the printer 40 by hardware and software. The printer 40 includes a program execution environment such as a CPU 41, a RAM 42, a ROM 43, and an EEPROM 44, and controls each unit via the I / Fs 45a to 45d according to a predetermined program. The USB I / F 45a is an I / F that acquires control data for setting the print data, the feed amount, and the nozzles used that are output from the computer 10.

  A carriage mechanism 46, a print head unit 47, and a paper feed mechanism 48 are connected to the I / Fs 45b to 45d, respectively. The print head unit 47 includes a mechanism for supplying the ink filled in the print head 47a and the ink cartridge to the ink chamber, and the like, and a piezo element for discharging ink to each ink nozzle in the print head 47a. Is provided. Each piezo element is controlled via the I / F 45c. As a result of this control, ink is recorded or not recorded for each pixel. Of course, this configuration is merely an example, and various other configurations such as a configuration for controlling ejection / non-ejection of ink by bubbles and a configuration for controlling the ejection amount of ink droplets in several stages can be employed. .

  The carriage mechanism 46 includes the print head unit 47 and includes a portion that can be driven in the main scanning direction (substantially perpendicular to the paper feed direction). The print head is synchronized with the ejection / non-ejection of the ink under the control of the CPU 41. The unit 47 is main-scanned. The paper feed mechanism 48 is provided with a mechanism for conveying the printing paper accumulated in an accumulation unit (not shown) one by one, and this mechanism conveys the printing paper to a site for recording ink. Then, sub-scanning is performed in synchronization with the ink ejection / non-ejection.

  In the printer 40 having such a configuration, in the present embodiment, processing by the print execution unit 41a and the parameter setting unit 41b can be executed under the control of the CPU 41 and the like. The print execution unit 41a acquires print data via the USB I / F 15, and executes printing by driving each unit based on the print data. That is, since the print data indicates ink recording / non-recording for each pixel, the print head unit 47, the carriage mechanism 46, and the paper feed mechanism 48 are controlled so that ink is recorded according to this data. To do. At this time, the paper feed mechanism 48 is controlled so that the print paper is fed at the set feed amount by referring to the feed amount set value and the used nozzle set value recorded in the EEPROM 44, The print head unit 47 and the carriage mechanism 46 are controlled so that ink droplets are ejected using the used nozzles.

The parameter setting unit 41 b is a module that records the feed amount setting value and the used nozzle setting value in the EEPROM 44. The parameter setting unit 41 b acquires the parameter setting value transmitted from the computer 10 via the USB I / F 15 and records it in the EEPROM 44. To do. Therefore, when the adjustment pattern data is printed as described above, the parameter set value is written to the EEPROM 44 each time the rectangular patterns P 1 to P 5 are printed, and each rectangular pattern P 1 to P 5 has a different parameter. Control to be printed. After the best parameter is extracted, the best parameter is written into the EEPROM 44, and thereafter, the printer 40 is controlled to print using this parameter.

(3) Calibration process:
7 to 13 are flowcharts showing the procedure of the calibration process in the present embodiment. A schematic flow of the calibration process is as shown in FIG. 7, and first, an adjustment pattern is printed by changing the paper feed amount in step S100.

(3-1) Adjustment pattern printing process:
FIG. 8 is a detailed flowchart showing the adjustment pattern printing process. In step S200, the counter n corresponding to the parameter number is initialized to “1”. In step S210, the parameter setting unit 22a refers to the parameter data 14c and performs setting for the printer 40. That is, referring to the parameter data 14c, the set value of the parameter number n is acquired, the control data is transmitted to the printer 40, and the set value is stored in the EEPROM 44. The flowchart shown in FIG. 8 is a common flow for changing the feed amount and changing the used nozzle. Of course, when changing the feed amount, the feed amount parameter n is set in the printer in step S210. If the used nozzle is changed, the parameter n of the used nozzle is set in the printer in step S210.

  In step S220, the image processing unit 21 acquires the adjustment pattern data 14a, and generates print data for printing under one of the printing conditions instructed in the printing condition data 14b in step S230. The generated print data is output to the printer 40 in step S240. As a result, the adjustment pattern (one of the rectangular patterns) is printed with the parameter to be calibrated set to a certain parameter n.

  In step S250, it is determined whether or not the counter n has exceeded a predetermined parameter number α (α = 5 in the example shown in FIG. 3). In step S250, it is determined that the counter n has exceeded the parameter number α. If not, the counter n is incremented at step S260, and the processing after step S210 is repeated. If it is determined in step S250 that the counter n has exceeded the parameter number α, printing of the adjustment pattern is completed while changing the parameter within a predetermined range, and the process returns to the process of FIG. .

(3-2) Scan processing:
Returning to the processing of FIG. 7, the printed adjustment pattern is scanned in step S <b> 110 and the image data is recorded in the HDD 14. FIG. 9 is a detailed flowchart showing the adjustment pattern scanning process. In step S300, the adjustment pattern printed in the flow shown in FIG. 8 is placed on the document table of the scanner 30. In step S310, the scanner DRV 30a controls the scanner 30 to perform a predetermined number of dummy scans in order to stabilize the light source that projects the document. Of course, it is only necessary to stabilize the amount of light from the light source here, so a standard preview may be implemented as a function of the scanner DRV 30a, or a predetermined time may be allowed to elapse while the light source is turned on. Various configurations such as turning on the light source can be employed.

  In step S320, the scanner DRV 30a sets a reading resolution suitable for the printing conditions when the adjustment pattern is printed. In step S330, scanning is performed twice based on the reading resolution. In step S340, image data based on the results of the two scans is recorded in the HDD 14, and the process returns to the process shown in FIG. In this scan, in order to reduce the variation in the measurement variation of the scan as much as possible, the average is calculated by performing the scan twice. Of course, it is only necessary to suppress variation in measurement here, and the number of scans is not limited to two. If the scan result is sufficiently stabilized by the dummy scan in step S310, the scan may be performed once. However, it is preferable that the scan is performed twice or more, for example, when the scan is performed by a so-called consumer scanner.

(3-3) Banding noise value acquisition processing:
After returning to the processing of FIG. 7, based on the image data 14d recorded on the HDD 14 in step S350, a banding noise value is acquired in step S120. FIG. 10 is a detailed flowchart showing the banding noise value acquisition process. In step S400, the patch cutout unit 22h acquires the image data 14d and cuts out a patch based on a predetermined number of pixels. That is, data for the number of pixels determined from the position corresponding to each patch is cut out from the image data 14d to generate individual patch data. For example, in the example shown in FIG. 3, 20 pieces of gray data located at both left and right ends in the main scanning direction and violet data located inside thereof are cut out.

In step S405, the patch cutout unit 22h acquires the L * value from the RGB data in each patch. Here, it is only necessary to be able to acquire the brightness of each patch. The L * value may be acquired by interpolation calculation or the like with reference to a color conversion table prepared in advance. The brightness may be simply calculated using an equation such as L = 0.3R + 0.59G + 0.11B.

  In step S410, the division processing unit 22g divides each patch in a direction orthogonal to the banding noise to generate a plurality of divided regions. In step S415, the lightness is averaged for the divided areas with the division processing unit 22g, and the relationship of the lightness to the position in the direction orthogonal to the banding noise is acquired for each divided area. The width of this division is preferably at least larger than the width of typical dust and noise (number of pixels in the main scanning direction).

  That is, in the present embodiment, the patch is divided in this way, and the specific divided area is excluded by the processing of the specific area removing unit 22d, thereby eliminating the scan result affected by the dust and noise. is doing. Therefore, assuming the size of typical dust and noise that can adhere to the adjustment pattern during scanning, the size of the divided area is determined so that the dust and noise can be contained in one divided area. For example, it has been found that dust and noise can be effectively removed when the width is about 16 pixels at a reading resolution of about 250 dpi.

  In step S420, the weighting processing unit 22f performs processing for conspicuous banding noise. Banding noise is a state in which a certain color is generated unintentionally, for example, a line of a different color is generated in a uniform image. Since this banding noise is particularly conspicuous in a dark part in a bright image or a bright part in a dark image, in this embodiment, in order to further emphasize the lightness value corresponding to such conspicuous banding noise, step S420. The weighting is performed for each color.

  FIG. 11 is a flowchart showing details of the weighting process. In this process, the weighting processing unit first calculates a change tendency value of the lightness L in step S500. That is, the change in the lightness L reflects the banding noise, but the banding noise is detected based on whether or not the lightness L has changed compared to the surrounding images. The change tendency is calculated. Since the patches in this embodiment are printed with print data for printing in a uniform color, the reference brightness should be almost constant, but in reality, a change tendency (low frequency undulation) is observed. It is done.

For example, in the graph shown in the lower part of FIG. 3, thin line brightness shows the change tendency by a thick line, as shown in the figure, the change trend changes slowly at about L * value 56. That is, the lightness L includes characteristics other than banding noise, such as bending of the print medium when printing a patch. Canceling this large change tendency makes banding noise in a uniform image manifest. It becomes easy. Therefore, in the present embodiment, instead of comparing the fixed lightness values, a change tendency value is calculated, and the change tendency value is canceled from the lightness L in step S505. Thereby, the low frequency undulation is canceled and the reference value becomes “0”.

  Here, the offset is set so that the reference is “0”, but even if the offset is performed, it can be said that the change in brightness is actually evaluated. Further, the above change tendency can be calculated by applying a low-pass filter to the change in lightness with respect to the position. For example, the change tendency value at a certain point is in the range of 2 mm before and after the point in the sub-scanning direction. The lightness value can be obtained and averaged by the corresponding pixel). As another example, a change tendency value of a point is obtained by obtaining a brightness value in a range of 2 mm in front and back in the sub-scanning direction around the point (corresponding to a pixel), and obtaining a median value by a median filter or the like. Can also be calculated by, for example, approximating. Of course, the change tendency may be a value indicating the tendency of the overall change in brightness, and the range for calculating the average is not limited to 2 mm, and can be calculated by various other methods.

  Of course, in the present embodiment, since the patch is printed with print data for printing in a uniform color, it is not essential to cancel the change tendency value when the lightness values are sufficiently uniform. . However, in this embodiment, banding noise is analyzed by Fourier transform, and it is possible to calculate the banding noise value more reliably by eliminating as much as possible the change in brightness caused by factors other than banding noise.

  After the change tendency value is canceled, the counter m indicating the position is initialized to “1” in step S510, and it is determined whether or not the value after the cancellation is larger than the reference “0” in step S515. Here, the value after cancellation is a value obtained by subtracting the change tendency value from the lightness L, and specifically, a value obtained by subtracting the change tendency value at the position m from the lightness L at the position m. When the value after the cancellation is larger than the reference “0”, the brightness at that point is brighter than the color of the patch and tends to cause white banding noise. On the contrary, when the value after cancellation is smaller than the reference “0”, the brightness at that point is darker than the color of the patch and tends to cause black banding noise.

  In this embodiment, the color of the patch is set to two colors, the Gr patch is set to high brightness, and the V patch is set to lower brightness. Therefore, in the Gr patch, the black banding noise is more noticeable than in the V patch, and in the V patch, the white banding noise is more noticeable than in the Gr patch. Therefore, in order to further enhance the lightness affected by the banding noise, a weighting coefficient depending on whether the value after cancellation is larger than the reference and the color of the patch is prepared in advance, and the weighting coefficient is used for emphasis.

That is, if it is not determined in step S515 that the value after cancellation is greater than the reference “0”, the color of the patch to be processed is determined in step S520. The color of the patch may be determined from the lightness L, or some flag may be set at the time of extraction in step S400, and various configurations can be employed. When it is determined in step S520 that the color of the patch is Gr, since the value after cancellation is smaller than the reference “0” in the high-intensity Gr patch, the value after cancellation is emphasized. The value after cancellation is multiplied by the weighting coefficient Cb 1 .

If the color patch is determined to be V at step S520, multiplied by the weighting factor Cb 2 in value after cancellation. In this case, since the value after cancellation is smaller than the reference “0” in the low-lightness V patch, the weighting coefficient Cb 2 is not a coefficient for emphasizing the value after cancellation. Here, since the former should be relatively emphasized between the Gr patch and the V patch, the weighting coefficient Cb 1 may be set to be larger than the weighting coefficient Cb 2 . For example, the weighting coefficient Cb 1 = 1.5 and the weighting coefficient Cb 2 = 1.0 may be set.

If it is determined in step S515 that the value after cancellation is greater than the reference “0”, the color of the patch to be processed is determined in step S535. If the color patch is determined to be in Gr at step S535, multiplied by the weighting coefficient Cw 1 to the value after cancellation. In this case, since the value after cancellation is larger than the reference “0” in the high brightness Gr patch, the weighting coefficient Cw 1 is not a coefficient for emphasizing the value after cancellation. If the color patch is determined to be V at step S535, multiplied by the weighting factor Cw 2 in value after cancellation. In this case, since the value after cancellation is greater than the reference “0” in the low brightness V patch, the weighting coefficient Cw 2 is a coefficient for emphasizing the value after cancellation. In this case as well, values may be set so that the former is relatively emphasized between the Gr patch and the V patch.

In terms of relative emphasis, it is only necessary to emphasize so as to make the influence of banding noise more obvious, and it is preferable to adjust the weighting coefficient for patches of the same color. That is, if the former size is increased by the weighting coefficient Cb 1 and the weighting coefficient Cw 1 in the Gr patch, it is possible to further emphasize the brightness change caused by the banding noise in the Gr patch in which black banding noise is conspicuous. it can. Further, if the former size is increased by the weighting coefficient Cb 2 and the weighting coefficient Cw 2 in the V patch, the brightness change caused by the banding noise can be more emphasized by the V patch in which white banding noise is conspicuous. it can. The weighting coefficients as described above may be determined in advance and recorded in the HDD 14.

  After the correction by the weighting coefficient, it is determined in step S550 whether or not the counter m is larger than a predetermined maximum value. If it is not determined that the counter m is larger than the maximum value, the counter m is incremented in step S560 and step S560 is performed. S515 and subsequent steps are repeated. That is, correction processing using weighting coefficients is performed for all one-dimensional positions.

  When the weighting is completed as described above, the reference is “0” and the brightness change due to the banding noise is further emphasized. In step S430 in FIG. 10, the FFT unit 22e performs Fourier transform on the data. Perform the conversion. When the Fourier transform result for each divided region is obtained as described above, in step S440, it is determined whether or not the Fourier transform processing has been completed for all the divided regions obtained for one patch. If it is not determined in step S440 that the Fourier transform processing has been completed for all the divided regions, the divided region to be processed is changed in step S445, and S415 and subsequent steps are repeated. That is, the same processing is performed for each of the divided areas.

  When it is determined in step S440 that the Fourier transform processing has been completed for all the divided areas, the singular area removing unit 22d excludes the singular divided areas in step S450. That is, a region in which the state of the Fourier transform result is different from other regions as compared between the divided regions is considered to be affected by dust and noise attached to the adjustment pattern, and is excluded from the analysis target.

  FIG. 12 is a detailed flowchart of the singular region removal process. In step S600, the singular region removing unit 22d first integrates the Fourier transform result for each divided region with respect to the spatial frequency. In step S610, the standard deviation σ is calculated on the assumption that the integral value for each divided region is a normal distribution. That is, the average value of each integral value is calculated, the average value of the square of the difference between the average value and the integral value is calculated, and the square root is taken. In step S620, twice the standard deviation σ is compared with each integral value, and it is determined whether or not there is a divided region having an integral value exceeding 2σ.

  If it is determined in step S620 that there is a divided region having an integral value exceeding 2σ, the data of the divided region is excluded from the analysis target in step S630. If it is not determined in step S620 that there is a divided region in which the deviation from the average is an integral value exceeding 2σ, step S630 is skipped. That is, in the present embodiment, the integrated values are compared when comparing the brightness changes of the divided areas. Further, the standard deviation of the integrated value eliminates the divided areas that are not included in the value range in which the majority of the integrated values are distributed.

  Further, the above processing excludes those whose deviation from the average exceeds 2σ. That is, the reason why the singular region is removed in the present embodiment is to remove the data affected by the dust and noise, and these influences act to increase the integrated value. More specifically, since the influence of dust or the like rarely occurs in a specific cycle, it is easy to contribute to the data after Fourier transform so that the power spectrum on the low frequency side as a whole increases. Therefore, in order to exclude the divided regions affected by these influences, it is determined whether or not the integral value exceeds + 2σ. Of course, it is assumed that a divided area different from the majority of integral values is a singular area, and for example, a divided area below −2σ may be excluded.

  Although the above comparison and exclusion are preferable in the sense that they can be easily realized by standard deviation, it is of course possible to remove specific regions by other configurations. For example, the average value of the integral values of each divided area may be calculated, and the divided areas having an integral value separated from the average value by a predetermined threshold or more may be excluded, or a value other than the integral value or a value before Fourier transform may be calculated. The singular region may be extracted and excluded using it.

  FIG. 14 is an explanatory diagram for explaining that the removal of the singular region as described above greatly contributes to the analysis of the banding noise value. In the figure, the banding noise value when the processing according to the present invention is performed on the image data when reading is performed in a state where a plurality of dust particles are attached to the print medium, and the above division is performed. It shows the banding noise value when not. In the figure, the horizontal axis represents the size of dust (one side of the square dust), and the vertical axis represents the ratio of banding noise values described in detail later.

  Here, a dust size “0” indicates that no dust is attached, and a banding noise value when the dust size is “0” is “1”. When no dust is attached, the banding noise value when the processing according to the present invention is performed is equal to the banding noise value when the above division is not performed. The target patch is an area of 512 pixels horizontally and 256 pixels vertically, the divided area is an area of 16 pixels horizontally and 256 pixels vertically, and the reading resolution is 300 dpi.

  As shown in the figure, the banding noise value in the case where the processing in the present embodiment is performed coincides with the state where no dust is attached until the size of the dust reaches 1.50 mm. That is, when the size of the dust is smaller than 1.50 mm, it can be said that the noise (dust) other than the banding noise can be surely removed and the banding noise value can be evaluated.

  On the other hand, when the division is not performed as in the present embodiment, the banding noise value is calculated by averaging the lightness in the lateral direction of the patch. In this case, the banding noise value increases as the dust becomes larger. The degree of the increase is large, and even if it is an integral multiple of the banding noise value when no dust is attached, it is easily reached. Since the degree of change of the banding noise value due to the influence of the banding noise is not so great, if the influence of the degree shown in FIG. 14 exists in the banding noise value, it means that the banding noise cannot actually be evaluated at all. Therefore, it can be said that the removal of the singular region in the present embodiment is very important in actually evaluating the banding noise.

  The size of the dust that can be reliably removed depends on the width of the divided area. That is, since 16 pixels at 300 dpi correspond to 1.35 mm, dust of this size can be removed very effectively. Therefore, if the typical size of dust is larger or smaller than 1.35 mm, the width of the divided area may be adjusted according to the size.

  In addition, since banding noise is generated substantially in parallel with a patch direction, it is preferable to evaluate using a one-dimensional Fourier transform that is substantially orthogonal to the direction. It can also be assumed that banding noise is evaluated by implementation. However, even when the banding noise value is calculated by performing the two-dimensional Fourier transform, the influence on the banding noise value is large unless dust is removed as described above. Therefore, the effect of removing the singular region as in the present embodiment appears remarkably.

  Further, by dividing the patch into a plurality of regions as described above, another remarkable effect can be obtained. That is, in the configuration in which the patches are averaged without being divided and the Fourier transform is performed, the effect is such that the analysis becomes impossible due to the sheet inclination at the time of scanning in the scanner 30. However, by performing the Fourier transform after dividing and averaging the patches, it is possible to prevent the analysis from becoming impossible due to the influence even if the sheet is inclined. More specifically, if the image data scanned with the printing paper on which the adjustment pattern is printed is averaged without being divided, the periodicity of banding noise (periodicity that occurs periodically in the sub-scanning direction) ) Substantially disappear, and even if Fourier transform is performed, the influence of banding noise is less likely to be reflected in the power spectrum value.

  FIG. 15 is an explanatory diagram for explaining this situation. In the figure, patch boundaries and banding noise are indicated by solid lines, and regions cut out for analysis are indicated by broken lines. In the figure, the boundary line of the large rectangle indicated by the broken line and the boundary line of the patch indicated by the solid line are inclined relatively, and this schematically shows that the printing paper is inclined when the scanner 30 scans. The technology that uses Fourier transform to analyze the situation of banding noise is based on the idea that banding noise has some periodicity, and aims to grasp the amount of periodic components included by Fourier transform.

  That is, in the printer, printing is performed by repeating the control of performing a main scan to record a recording material and performing a sub-scan to send a recording medium. Banding noise based on the same cause, such as the ejection direction of the ejected ink being different from the others, should appear repeatedly and have some periodicity. Therefore, in order to analyze banding noise, it is necessary to acquire a Fourier transform result in a state in which this periodicity is reliably reflected.

  In this banding noise analysis, assuming that the patches are averaged without being divided, the brightness value in each large rectangular pixel indicated by a broken line in FIG. (Direction) are added together and averaged. As a result, the average brightness with respect to the position in the vertical direction (scanning sub-scanning direction) is obtained. However, when the printing paper is tilted as shown in the figure, the banding noise also tilts. Affects across pixels. For example, on the right side of the rectangle indicated by the broken line in FIG. 15, a region obtained by projecting the banding noise in the sub-scanning direction at the time of scanning is indicated by an arrow, and an area in which the banding noise extends is shown. Therefore, banding noise that should be present in a narrow range is dispersed and affected over a wide range. Therefore, the brightness change after averaging becomes small and the spatial frequency becomes ambiguous (it can be detected only at a frequency much lower than the original frequency).

  On the other hand, FIG. 15 shows a state in which an area is divided under the same situation. The divided areas are indicated by long broken rectangles in the position direction shown in the figure. In the present embodiment, the lightness values at the respective positions are averaged for each divided area. Therefore, even if the solid banding noise is tilted within the broken rectangle as shown in the lower part of FIG. 15, the banding noises do not straddle in the position direction, and there is one banding noise as shown by the arrow. The range in the position direction (range obtained by projecting in the sub-scanning direction) is a very narrow range.

  Therefore, even if the brightness is averaged in the divided area, the influence is surely manifested in a certain period T. Even if the position affected by the banding noise differs in each divided region, if the period T of the banding noise is substantially constant, the power spectrum of a certain spatial frequency or its harmonic component becomes large after Fourier transform. As a result, a Fourier transform result including the influence of banding noise can be obtained with certainty, and the analysis can be performed without being affected by the inclination of the printing paper.

  As described above, if the analysis is performed without dividing the patch into a plurality of regions, a situation may occur in which no banding noise is detected due to the inclination of the printing paper. In addition, the degree of inclination greatly affects the Fourier transform result, and the analysis result becomes very unstable. However, if the region is divided as in the present invention, the reliability of analysis can be guaranteed very easily. The width of the divided region in the main scanning direction may be such that the spread of banding noise caused by a typical inclination does not overlap in the sub-scanning direction and the spread is within the smallest possible range.

  FIG. 16 is a diagram illustrating the influence of the inclination on the banding noise value. In the figure, the state where the main scanning direction at the time of scanning and the main scanning direction at the time of printing are parallel is an angle 0, and the banding noise value is shown when both are inclined relatively. is there. The vertical axis represents the ratio of banding noise values normalized by the value when the angle is zero. Here too, a comparison is made between the case where the processing in this embodiment is performed and the case where it is not performed at each angle. The scanning conditions and patches are the same as in the example in FIG. 14, but in the example shown in FIG. 16, processing is performed based on image data scanned in a state where no dust is attached.

  When the brightness is averaged for each divided region and Fourier transform is performed for each divided region as in the present embodiment, the angle is “0, 0.25, 0.5, 0.75, 1.0”. Even if it changes, the ratio of the banding noise value is almost 1. That is, there is almost no influence by rotation. On the other hand, when the Fourier transform is performed by averaging the whole without dividing as in the present embodiment, the ratio of the banding noise value greatly fluctuates with the change of the angle. Again, the fluctuation is large, and the degree to which the banding noise value changes due to the influence of the banding noise is not so great. Therefore, when averaged as a whole, the banding noise cannot be evaluated at all due to the inclination. Therefore, it can be said that the area division in this embodiment is very important in actually evaluating the banding noise.

  Although division is preferably performed as described above, it is not preferable to perform Fourier transform on each pixel in the main scanning direction without averaging for each divided region. That is, since printing by the printer 40 is realized by recording ink droplets, it is affected by the droplets of the ink droplets in a scanned pixel unit, and may cause an excessive lightness change beyond the lightness change that can be perceived by human eyes. In addition, if processing is performed for each pixel, the data tends to be excessively fine. Therefore, by averaging the brightness for each divided region as in this embodiment, it is possible to effectively analyze the banding noise without handling data with excessively changing brightness or an excessively large amount of data. .

  As described above, when the singular region is removed in step S450, a Fourier transform result for each divided region in which the influence of the brightness change due to the banding noise is reliably reflected in the power spectrum value remains. Therefore, in order to be able to compare the occurrence status of banding noise for each patch using these results, a banding noise value indicating the occurrence status of banding noise in each patch is calculated.

  In step S455, the banding noise value acquisition unit 22c removes the background from each Fourier transform result. The background here is a spectrum value that does not reflect periodicity due to banding noise. That is, since banding noise is periodically generated as described above, in the power spectrum as shown in the upper part of FIG. 5, it is considered that each peak is a spectrum generated by banding noise. Therefore, the background that exists over the entire spatial frequency is calculated and removed from the spectrum value. The lower part of FIG. 5 shows the peak after removing the background.

  In the present embodiment, the background is calculated by applying a median filter to the power spectrum value shown in the upper part of FIG. 5 and fitting the obtained result with a sixth-order function. Here, as the median filter, for example, a filter that replaces a power spectrum value of a certain spatial frequency with a median value of power spectrum values included within five frequencies before and after that can be adopted. The reason for fitting with the 6th order function is to make the background change as smooth as possible. Of course, it is only necessary that the background can be calculated and removed, and various other configurations can be calculated.

  When the background is removed for each Fourier transform result, in step S457, all the Fourier transform results are summed for each spatial frequency and averaged. That is, a value reflecting the banding noise in the entire patch is acquired from the Fourier transform result for each divided region. Here, the values corresponding to the spectra in one divided region are obtained by averaging, and the relative comparison of the same color can be performed by this processing. Therefore, as long as a relative comparison can be performed, various processes other than the averaging can be employed.

  Further, in step S460, processing for leaving only the power spectrum of the spatial frequency that can be perceived by humans based on human visual characteristics is performed on the obtained value. That is, since there exist spatial frequencies that are easy to recognize for human vision and spatial frequencies that cannot be recognized, filter processing based on visual characteristics is performed in order to more reliably analyze banding noise perceived by humans. As the filter based on visual characteristics, various filters reflecting human visual characteristics can be adopted. For example, a visual transfer function (VTF) may be used, or only a fixed spatial frequency (for example, 0). .2 mm to 4 mm) may be left.

  In step S465, the obtained data, that is, the singular region is removed, the background is removed, and the spatial frequency is integrated with respect to the data obtained by filtering the visual characteristic to the value reflecting the banding noise in the patch. . This integration result is the banding noise value. In steps S455 to S465, it is only necessary to calculate the banding noise value for each patch. Before removing the background, the Fourier transform results may be added or the average may be calculated. Processing procedures can be employed.

  As a result of the above processing, banding noise values for a certain patch have been acquired. In step S470, it is determined whether or not banding noise values have been calculated for all patches included in the adjustment pattern. If it is not determined in step S470 that banding noise values have been acquired for all patches, the patch to be processed is changed in step S475, and the processes in and after step S400 are repeated. If it is determined in step S470 that banding noise values have been acquired for all patches, it is determined in step S480 whether banding noise values have been calculated for scanned images. If it is not determined in step S480 that the banding noise value has been calculated for the scanned image, the target image data is changed in step S485, and the processes in and after step S400 are repeated. That is, in step S330, since the image data is acquired by performing the scan twice, the banding noise value is calculated based on each. In step S490, the banding noise values calculated for each image data are averaged and stored in the RAM 12. When the banding noise values of all patches are acquired and stored by the above processing, the processing returns to the processing in FIG.

(3-4) Best value extraction process:
Returning to the processing of FIG. 7, in step S130, the banding noise value is referred to, and the best parameter value is selected. Further, the selected parameter is set for the printer 40. FIG. 13 is a detailed flowchart showing the best parameter selection processing, and FIG. 17 is a diagram showing an example of the banding noise value together with the adjustment pattern. In step S700, the best value extraction unit 22b acquires the banding noise value of the same color patch in the adjustment pattern.

  For example, in the pattern shown in FIG. 17, patches of the same color arranged in the sub-scanning direction are acquired for each column to acquire a banding noise value. That is, five Gr patches arranged in the left end (column A), five V patches arranged in the second column from the left (column B), five Gr patches arranged in the right end (column C), and the second column from the right The banding noise values of the five V patches (column D) arranged in a row are acquired. In the following, an example of acquiring banding noise values for patches of the same color for each column will be described. However, here, it is only necessary to acquire patches of the same color, and even if the configuration is such that columns A and D are acquired simultaneously. good.

  When banding noise values for a plurality of patches are acquired, a minimum value is searched for in step S705, and each banding noise value is divided by the minimum value in step S710. In step S715, it is determined whether or not the above processing has been performed for all patches. If it is not determined that the processing has been performed for all patches, the processing target is changed and the processing in step S700 and subsequent steps is repeated. The above processing can be said to be processing in which normalization is performed for each column with the minimum banding noise value. In FIG. 17, the result of normalization is shown in parentheses. For example, in column A, the minimum banding noise value is the value 15 of parameter # 3, and the banding noise values “25, 21, 15, 22, 28” of parameters # 1 to # 5 are divided by “15”. , Value in parentheses.

  In steps S720 and S725, the normalized values are added for each parameter. That is, the values after normalization are summed for a certain parameter in step S720, and the processes in and after step S720 are repeated until it is determined in step S725 that all parameters have been processed. For example, in the example shown in FIG. 17, the standardized values for the parameter # 1 are “1.67, 1.47, 1.30, 1.58” for the columns A to D, respectively. In addition, the value “6.02” is calculated. The similarly calculated values for parameters # 2 to # 5 are “4.82, 4.20, 5.40, 6.13”.

  In step S730, the minimum value is extracted from the total values for each parameter calculated as described above. Since the banding noise value increases as the frequency and intensity of banding increase, it can be said that by extracting the value with the smallest total value, it is possible to extract the state where banding does not occur as much as possible when printing each patch. . That is, the occurrence of banding noise differs in each of the plurality of patches, and it can be said that the smallest banding noise is the smallest banding noise when the patches are compared for each column.

  However, since it is necessary to set a certain parameter for the printer 40 and perform printing, when a patch having a minimum banding noise value exists in different parameters (for example, parameter # 3 in column A and parameter # 3 in column B). Even if parameter # 2 is the minimum), one of the parameters must be selected. Therefore, in the present embodiment, the best parameter is selected by normalizing the banding noise value, adding up each parameter, and comparing the total value. As a result, the best parameter (the lowest frequency of banding noise as a whole) can be extracted.

  In extracting the parameters as described above, the banding noise value is standardized. This is a countermeasure for the fact that the correction using the above-described weighting coefficient differs depending on each patch. Since patches of the same color have the same weighting coefficient value, it is considered that correction using the weighting coefficient is performed in the same manner for patches of the same color. However, for the patches of different colors, the magnitudes of the weighting coefficients are not always the same, and the degree and frequency of emphasis by the correction may be different. Therefore, if the same color is first standardized, it is possible to provide an index indicating how much noise is present relative to the best (least) noise state in each color. The best parameter can be selected based on the value.

  Furthermore, in this embodiment, in order to perform the best setting more reliably, it is determined whether or not it is a defective aircraft. That is, even if the best parameter is selected as described above, in rare cases, there may be a defective aircraft that is insufficient to complete the calibration. For example, if the banding noise values for all the parameters exceed the allowable value, it cannot be said that the calibration is completed even if the best parameter is selected after the relative comparison. Further, even if the total value of the normalized values is the smallest, if there is a large difference in banding noise values at both ends of the main scanning, the main scanning direction and the sub-scanning direction are inclined at right angles. it is conceivable that. Therefore, even in this case, the calibration should not be completed.

Therefore, in the present embodiment, it is determined in step S735 whether or not the banding noise value at the best parameter is equal to or less than a predetermined threshold value. That is, the threshold value T 1 for the Gr patch and the threshold value 2 for the V patch are set in advance and recorded in the HDD 14, and the banding noise values of the two Gr patches at the best parameters are compared with the threshold value 1 . Compare the banding noise value with threshold 2 . Each of the threshold values 1 and 2 is a banding noise value corresponding to the limit value of the allowable banding noise.

If it is determined in step S735 that the banding noise value at the best parameter is equal to or less than the threshold value, the difference between the banding noise values of the patches at both ends is calculated in step S740. At this time, the same color, that is, the difference between the Gr patches and between the V patches is calculated. In step S745, it is determined whether the difference is equal to or less than a predetermined threshold value 3 . Here, the threshold 3 is an allowable value as a difference between banding noise values in the patches at both ends, and is determined in advance.

If it is determined in step S745 that the difference is equal to or less than the predetermined threshold 3 , banding noise can be suppressed by the best parameter extracted in step S730, and it may be appropriate as a setting value for the printer 40. It has been confirmed. Therefore, the parameter setting unit 22a sets the parameter in the printer 40 in step S747.

If it is not determined in step S735 that the banding noise value at the best parameter is less than or equal to a predetermined threshold value, and if the difference is not determined to be less than or equal to the predetermined threshold value 3 in step S745, the best in step S750. The value extraction unit 22b controls the display 17 and displays that the printer 40 is defective. When this display is made, manual adjustment such as carriage height adjustment, readjustment to change parameters, processing that adopts the threshold value or parameters that are allowed by visual observation, although not the best parameters, etc. To implement. On the other hand, when the process of step S747 is completed, the process corresponding to step S130 in FIG. 7 is terminated, and as a result of this process, the paper feed amount that can suppress the banding noise is set to the maximum in the printer 40.

  In the present embodiment, after the adjustment of the paper feed amount, the adjustment of the used nozzle is also performed, and the process for adjusting the used nozzle is performed in steps S140 to S170 of FIG. carry out. This process is the same as the process shown in FIGS. 8 to 13 described above, except that the processing target is the used nozzle. That is, the parameter value is changed as shown in FIG. 4, the adjustment pattern is printed, the banding noise value is analyzed, and the setting is made so as to be the best parameter. Of course, values such as the above-described threshold values 1 to 3 can be changed by adjusting the paper feed amount and the nozzles used. As a result of this processing, printing is set to be performed using a set of nozzles in which banding noise is most suppressed.

  In the present embodiment, the adjustment pattern is printed by individually changing the generation factor of banding noise (setting of the feed amount and setting of the nozzle to be used). Even if there are multiple factors that generate banding noise, the best parameter is set individually while focusing on only one of the factors, and then the best parameter is set while focusing on other parameters. By adjusting each factor, banding noise can be suppressed reliably.

  Further, when the change in the feed amount is compared with the change in the used nozzle, the change in the feed amount has a greater influence on the degree of occurrence of banding noise. Therefore, in the present embodiment, the factors that have a large influence on the degree of occurrence of banding noise are adjusted first. Since the influence of the feed amount is large, it is preferable to perform readjustment when the degree of occurrence of banding noise is large because the feed amount is not adjusted. However, since the influence of changing the nozzle used is small, there is a high possibility that calibration can be completed by readjusting the nozzle used without adjusting the feed amount again or resetting the parameters. Therefore, the calibration can be completed easily and reliably as compared with the case where adjustment is performed from a factor that has a small influence on the degree of occurrence of banding noise.

(4) Other embodiments:
The above embodiment is an example for realizing the present invention, and it is possible to adopt other configurations and processing procedures. For example, the present invention may be applied to a laser printer instead of an ink jet printer. In this case, as factors for generating banding noise, attention can be paid to various factors such as laser output as well as the paper feed amount. Also, the arrangement of the nozzles is not limited to the arrangement shown in FIG. 4, and two rows of nozzles may be formed for one color of ink, and the number of ink colors is not limited to seven.

  Furthermore, when changing the use nozzle, the structure which fixes the number of use nozzles as shown in FIG. 4 is not necessarily required. In this case, a set of a plurality of nozzles obtained by changing the number of used nozzles is specified by the parameter data 14c, and these are appropriately changed by the setting in the parameter setting unit 22a to print the adjustment pattern. Regardless of how the number of nozzles used and the position of the nozzles used are changed, a parameter that reliably suppresses the banding noise can be obtained by calculating the banding noise value in the adjustment pattern described above.

  Furthermore, the adjustment pattern is not limited to the above-described pattern. For example, patches composed only of ink droplets ejected from each of a plurality of ink colors may be arranged in the main scanning direction. As a result, it is possible to detect whether banding noise can be generated individually for each color. Of course, for example, a gray patch may be added at a symmetrical position at the end in the main scanning direction with respect to this pattern, and various configurations can be adopted.

  Furthermore, the image data to which the present invention is applied is not limited to a patch scan image. Even in the case of image data indicating an image other than a patch, if the region is divided, averaged in one direction for each divided region, and the one-dimensional FFT result is integrated and compared as in the present invention, the singular region, that is, the non-patent region An area including periodic noise can be easily extracted. Therefore, aperiodic noise can be easily found. In addition, by comparing the integrated values of the one-dimensional FFT results, it is possible to grasp the degree of occurrence of periodic noise, and it is easy to find periodic noise. Of course, the application target of the present invention is preferably an image in which noise is conspicuous due to the uniform image, for example, a dark part of the night sky.

It is explanatory drawing for demonstrating the outline of this invention. It is a block diagram which shows the structure of a computer. It is a figure which shows the pattern for adjustment. It is explanatory drawing explaining the parameter data about a use nozzle. It is a figure which shows the result of a Fourier-transform, and its subsequent process. FIG. 2 is a block diagram illustrating a configuration of a printer. It is a schematic flowchart which shows the procedure of a calibration process. It is a flowchart which shows the printing process of the pattern for adjustment. It is a flowchart which shows a scanning process. It is a flowchart which shows the process at the time of acquiring a banding noise value. It is a flowchart which shows a weighting process. It is a flowchart which shows the removal process of a specific area | region. It is a flowchart which shows the selection process of the best parameter. It is explanatory drawing explaining the effect by removing a specific area | region. It is explanatory drawing explaining the effect | action by dividing | segmenting a patch. It is a figure which shows the influence which inclination has on a banding noise value. It is a figure which shows an example of a banding noise value, and a pattern for adjustment.

Explanation of symbols

  DESCRIPTION OF SYMBOLS 10 ... Computer, 14a ... Adjustment pattern data, 14b ... Printing condition data, 14c ... Parameter data, 14d ... Image data, 20 ... PRTDRV, 21 ... Image processing part, 22 ... Calibration module, 22a ... Parameter setting part, 22b ... best value extraction unit, 22c ... banding noise value acquisition unit, 22d ... singular region removal unit, 22e ... FFT unit, 22f ... weighting processing unit, 22g ... division processing unit, 22h ... patch cutout unit, 30 ... scanner, 40 ... Printer, 41a ... Print execution unit, 41b ... Parameter setting unit, 44 ... EEPROM, 46 ... Carriage mechanism, 47 ... Print head unit, 47a ... Print head, 48 ... Paper feed mechanism

Claims (20)

  1. A first pattern including a plurality of patches arranged in the main scanning direction of the printing apparatus is a second pattern configured by arranging a plurality of patches in the sub-scanning direction of the printing apparatus, and affects the noise generation in the printing apparatus Means for causing the printing apparatus to print a second pattern with different settings for each first pattern on a recording medium;
    Distribution information acquisition means for acquiring distribution information indicating the distribution of the recording material in a plurality of divided regions smaller than the patch in each patch constituting the second pattern printed on the recording medium ;
    Based on the distribution information in each divided region, a singular region extracting means for extracting a singular region having a singular distribution compared to other regions,
    Noise detection means for detecting noise of each patch constituting the second pattern based on distribution information in the divided areas excluding the singular area ;
    Parameter setting means for adding the noise of patches having the same parameters together, specifying a parameter with the minimum combined result, and setting the printing apparatus to be driven by the specified parameter A calibration device .
  2. The calibration apparatus according to claim 1, wherein the division information acquisition unit averages the unidirectional distribution of the recording material for each divided region and acquires distribution information in each divided region.
  3. 3. The division information acquisition unit according to claim 1, wherein the division information acquisition unit acquires the division region by cutting the patch along a line substantially perpendicular to a banding noise generation direction when the patch is printed. The calibration device according to any one of the above.
  4. The length of the divided area in the short direction specifies in advance the magnitude of either or both of typical dust that can adhere to the recording medium and typical noise that occurs when the distribution information is acquired. The calibration device according to any one of claims 1 to 3, wherein the size is determined to be smaller than the length.
  5. The distribution information acquisition means is a scanner that reads a recording medium in a two-dimensional manner, and the length of the divided area in the short direction is a state in which a typical inclination occurs in the recording medium during reading by the scanner. The calibration apparatus according to claim 1, wherein the range of the banding noise existing in each divided region is determined so as not to overlap.
  6. The specific area extracting means calculates a distribution characteristic value indicating a characteristic of the recording material to be distributed for each divided area, and extracts the specific area based on the distribution characteristic value. 6. The calibration device according to any one of items 5.
  7. The calibration apparatus according to claim 6, wherein the distribution characteristic value is calculated based on a value obtained by converting the distribution information into a spectrum with respect to a spatial frequency.
  8. 8. The calibration apparatus according to claim 7, wherein the distribution characteristic value is a value obtained by integrating the spectrum with respect to a spatial frequency.
  9. The said singular area extraction means calculates the standard deviation of the said distribution characteristic value, and discriminate | determines whether it is a singular area | region using the constant multiple of the said standard deviation as a threshold value. The calibration device according to any one of the above.
  10. 10. The calibration apparatus according to claim 1, wherein the patch includes a color in which the noise is easily recognized by human eyes.
  11. The calibration apparatus according to claim 1, wherein the patch includes a color in which the noise is frequently generated in a printing apparatus that prints the patch.
  12. 12. The method according to claim 1, wherein the first pattern includes patches of a plurality of colors, and the patches include recording materials of all colors usable in the printing apparatus. A calibration apparatus according to claim 1 .
  13. The calibration according to any one of claims 1 to 12, wherein the first pattern includes two colors of a patch having a predetermined lightness color and a patch having a lightness color lower than the patch . Equipment .
  14. The said 1st pattern contains the some patch arrange | positioned in the substantially symmetrical position on both sides of the center position of the main scanning in a printing apparatus, The said Claim 1 characterized by the above-mentioned. Calibration device .
  15. The distribution information acquisition means is an optical reader, and performs a preview scan or a process of emitting light to stabilize the light source prior to acquisition of distribution information. A calibration apparatus according to claim 1 .
  16. A first pattern including a plurality of patches arranged in the main scanning direction of the printing apparatus is a second pattern configured by arranging a plurality of patches in the sub-scanning direction of the printing apparatus, and affects the noise generation in the printing apparatus Means for causing the printing apparatus to print a second pattern with different settings for each first pattern on a recording medium;
    Analysis target region extraction means for extracting, as an analysis target, a region for each patch constituting the second pattern from image data including a plurality of pixels generated by scanning the second pattern printed on a recording medium ;
    Area dividing means for dividing the area to be analyzed into a plurality of divided areas;
    Based on the image data for each divided region, a singular region extracting means for extracting a singular region having a singular distribution compared to other regions ;
    Noise detection means for detecting noise of each patch constituting the second pattern based on distribution information of the image data in the divided area excluding the singular area;
    Parameter setting means for adding the noise of patches having the same parameters together, specifying a parameter with the minimum combined result, and setting the printing apparatus to be driven by the specified parameter A calibration device .
  17. A first pattern including a plurality of patches arranged in the main scanning direction of the printing apparatus is a second pattern configured by arranging a plurality of patches in the sub-scanning direction of the printing apparatus, and affects the noise generation in the printing apparatus A function for causing the printing apparatus to print a second pattern, which is different for each first pattern, on a recording medium;
    A distribution information obtaining function for obtaining the distribution information indicating the distribution of the recording material in a small plurality of divided regions from the patch definitive each patch constituting the printed on a recording medium said second pattern,
    Based on the distribution information in each divided region, a singular region extraction function that extracts a singular region that has a unique distribution compared to other regions,
    A noise detection function for detecting noise of each patch constituting the second pattern based on distribution information in the divided areas excluding the singular area ;
    The computer implements a parameter setting function that adds the noise of patches having the same parameters together, identifies a parameter with the smallest summation result, and sets the printing apparatus to be driven by the identified parameter A calibration program characterized by
  18. A first pattern including a plurality of patches arranged in the main scanning direction of the printing apparatus is a second pattern configured by arranging a plurality of patches in the sub-scanning direction of the printing apparatus, and affects the noise generation in the printing apparatus A function for causing the printing apparatus to print a second pattern, which is different for each first pattern, on a recording medium;
    An analysis target region extraction function for extracting, as an analysis target, a region for each patch constituting the second pattern from image data including a plurality of pixels generated by scanning the second pattern printed on a recording medium ;
    A region dividing function for dividing the region to be analyzed into a plurality of divided regions;
    Based on the image data for each divided region, a singular region extraction function that extracts a singular region that has a unique distribution compared to other regions ,
    A noise detection function for detecting noise of each patch constituting the second pattern, based on the distribution information of the image data in the divided area excluding the singular area;
    The computer implements a parameter setting function that adds the noise of patches having the same parameters together, identifies a parameter with the smallest summation result, and sets the printing apparatus to be driven by the identified parameter A calibration program characterized by
  19. A first pattern including a plurality of patches arranged in the main scanning direction of the printing apparatus is a second pattern configured by arranging a plurality of patches in the sub-scanning direction of the printing apparatus, and affects the noise generation in the printing apparatus A step of causing the printing apparatus to print a second pattern with different settings for each first pattern on a recording medium;
    A distribution acquisition step of acquiring distribution information indicating a distribution of the recording material in a small plurality of divided regions from the patch definitive each patch constituting the printed on a recording medium said second pattern,
    Based on the distribution information in each divided region, a singular region extraction step for extracting a singular region having a unique distribution compared to other regions,
    A noise detection step of detecting noise of each patch constituting the second pattern based on distribution information in the divided areas excluding the singular area ;
    A parameter setting step of adding the noises of the patches having the same parameters together, specifying a parameter having the minimum combined result, and setting the printing apparatus to be driven by the specified parameter. Calibration method .
  20. A first pattern including a plurality of patches arranged in the main scanning direction of the printing apparatus is a second pattern configured by arranging a plurality of patches in the sub-scanning direction of the printing apparatus, and affects the noise generation in the printing apparatus A step of causing the printing apparatus to print a second pattern with different settings for each first pattern on a recording medium;
    An analysis target region extraction step of extracting, as an analysis target, a region for each patch constituting the second pattern from image data including a plurality of pixels generated by scanning the second pattern printed on a recording medium ;
    A region dividing step of dividing the region to be analyzed into a plurality of divided regions;
    Based on the image data for each divided region, a singular region extraction step for extracting a singular region having a singular distribution compared to other regions ,
    A noise detection step of detecting noise of each patch constituting the second pattern, based on distribution information of image data in a divided region excluding a singular region;
    A parameter setting step of adding the noises of the patches having the same parameters together, specifying a parameter having the minimum combined result, and setting the printing apparatus to be driven by the specified parameter. Calibration method .
JP2004205068A 2004-07-12 2004-07-12 Calibration apparatus, calibration program, and calibration method Expired - Fee Related JP4543796B2 (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
JP2004205068A JP4543796B2 (en) 2004-07-12 2004-07-12 Calibration apparatus, calibration program, and calibration method

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
JP2004205068A JP4543796B2 (en) 2004-07-12 2004-07-12 Calibration apparatus, calibration program, and calibration method

Publications (2)

Publication Number Publication Date
JP2006026943A JP2006026943A (en) 2006-02-02
JP4543796B2 true JP4543796B2 (en) 2010-09-15

Family

ID=35893861

Family Applications (1)

Application Number Title Priority Date Filing Date
JP2004205068A Expired - Fee Related JP4543796B2 (en) 2004-07-12 2004-07-12 Calibration apparatus, calibration program, and calibration method

Country Status (1)

Country Link
JP (1) JP4543796B2 (en)

Families Citing this family (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP4902456B2 (en) * 2007-07-31 2012-03-21 シャープ株式会社 Unevenness evaluation apparatus, Unevenness evaluation method, Unevenness evaluation program, recording medium, and color filter manufacturing method
JP5813610B2 (en) 2012-09-28 2015-11-17 富士フイルム株式会社 Image evaluation apparatus, image evaluation method, and program
JP5813611B2 (en) * 2012-09-28 2015-11-17 富士フイルム株式会社 Image evaluation apparatus, image evaluation method, and program
JP6555580B2 (en) * 2015-07-07 2019-08-07 株式会社リコー Image quality change detection apparatus, image quality change detection method, and image quality change detection program

Citations (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JPH1096696A (en) * 1996-09-25 1998-04-14 Mitsubishi Heavy Ind Ltd Method and apparatus for inspecting irregularity in object
JPH11232462A (en) * 1998-02-12 1999-08-27 Canon Inc Printing inspection device
JP2000148987A (en) * 1998-11-17 2000-05-30 Olympus Optical Co Ltd Image processor
JP2001074602A (en) * 1999-09-09 2001-03-23 Fuji Xerox Co Ltd Apparatus and method for evaluating image
JP2001292331A (en) * 2000-04-07 2001-10-19 Canon Inc Image processing method and device, image processing system and recording medium
JP2003067723A (en) * 2001-06-12 2003-03-07 Eastman Kodak Co Evaluation method of noise appearance in image
JP2003219158A (en) * 2002-01-17 2003-07-31 Ricoh Co Ltd Image forming apparatus
JP2004053944A (en) * 2001-11-22 2004-02-19 Ricoh Co Ltd Image quality detecting device, image forming apparatus, image quality control device and image quality control method

Patent Citations (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JPH1096696A (en) * 1996-09-25 1998-04-14 Mitsubishi Heavy Ind Ltd Method and apparatus for inspecting irregularity in object
JPH11232462A (en) * 1998-02-12 1999-08-27 Canon Inc Printing inspection device
JP2000148987A (en) * 1998-11-17 2000-05-30 Olympus Optical Co Ltd Image processor
JP2001074602A (en) * 1999-09-09 2001-03-23 Fuji Xerox Co Ltd Apparatus and method for evaluating image
JP2001292331A (en) * 2000-04-07 2001-10-19 Canon Inc Image processing method and device, image processing system and recording medium
JP2003067723A (en) * 2001-06-12 2003-03-07 Eastman Kodak Co Evaluation method of noise appearance in image
JP2004053944A (en) * 2001-11-22 2004-02-19 Ricoh Co Ltd Image quality detecting device, image forming apparatus, image quality control device and image quality control method
JP2003219158A (en) * 2002-01-17 2003-07-31 Ricoh Co Ltd Image forming apparatus

Also Published As

Publication number Publication date
JP2006026943A (en) 2006-02-02

Similar Documents

Publication Publication Date Title
US20050094871A1 (en) Production of color conversion profile for printing
US20110012949A1 (en) Printing method for reducing stitch error between overlapping jetting modules
US6970606B2 (en) Automatic image quality evaluation and correction technique for digitized and thresholded document images
US8432582B2 (en) Uniformity compensation in halftoned images
US6266439B1 (en) Image processing apparatus and methods
AU2002300864B2 (en) Image Processing Apparatus for Print Process of Photographed Image
US6764155B2 (en) System and method for compensating for non-functional ink cartridge ink jet nozzles
US7118188B2 (en) Hardcopy apparatus and method
TWI467515B (en) Multi-color dropout for scanned document
CN101491075B (en) Image processing device, and image processing method
JP2006172417A (en) Image evaluating method, image evaluating device, and printer
JP6111730B2 (en) Color measuring device, image forming apparatus, color measuring system and color measuring method
US7450268B2 (en) Image reproduction
US20080137146A1 (en) Image data generating apparatus, image printing apparatus, and image data generating method
US20040223025A1 (en) System and method for compensating for non-functional ink cartridge ink jet nozzles
DE60122276T2 (en) Determination of the adjustment value for changing the recording position when printing by two types of inspection patterns
KR100542365B1 (en) Appratus and method of improving image
JP5457307B2 (en) Defective recording element compensation parameter selection chart, defective recording element compensation parameter determination method and apparatus, and image forming apparatus
US20060165285A1 (en) Image processing apparatus, image forming apparatus, image reading process apparatus, image processing method, image processing program, and computer-readable storage medium
JP2006297691A (en) Printing apparatus for printing image on skin and printing method
JP5067499B2 (en) Image processing device
US20020080375A1 (en) Reduction of artefacts in reproduced images
EP1195981B1 (en) Image processing apparatus, print control apparatus, image processing method, and recording medium
US7289248B2 (en) Image recording apparatus and image correction method
US7440123B2 (en) Adaptive printing

Legal Events

Date Code Title Description
A621 Written request for application examination

Free format text: JAPANESE INTERMEDIATE CODE: A621

Effective date: 20070119

A131 Notification of reasons for refusal

Free format text: JAPANESE INTERMEDIATE CODE: A131

Effective date: 20091216

A521 Written amendment

Free format text: JAPANESE INTERMEDIATE CODE: A523

Effective date: 20100209

TRDD Decision of grant or rejection written
A01 Written decision to grant a patent or to grant a registration (utility model)

Free format text: JAPANESE INTERMEDIATE CODE: A01

Effective date: 20100608

A01 Written decision to grant a patent or to grant a registration (utility model)

Free format text: JAPANESE INTERMEDIATE CODE: A01

A61 First payment of annual fees (during grant procedure)

Free format text: JAPANESE INTERMEDIATE CODE: A61

Effective date: 20100621

R150 Certificate of patent or registration of utility model

Free format text: JAPANESE INTERMEDIATE CODE: R150

FPAY Renewal fee payment (event date is renewal date of database)

Free format text: PAYMENT UNTIL: 20130709

Year of fee payment: 3

LAPS Cancellation because of no payment of annual fees