JP2008160604A - Image processor, chart for proofreading used in the same processor and image processing method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an image processor eliminating an influence of characteristic variation of a reading means in a minute area of image data. <P>SOLUTION: The image processor is provided with the reading means 100 for reading an original, a recognition means 210 for recognizing a predetermined pattern of the image data according to whether or not the image data of the read original satisfies a predetermined recognition condition and a correction means 210 for correcting the recognition condition by the recognition means based on the size of a pattern to be recognized by the recognition means. <P>COPYRIGHT: (C)2008,JPO&INPIT

Description

  The present invention relates to an image processing apparatus such as a color scanner or an MFP (Multi Function Peripherals) equipped with a color scanner, a calibration chart used for examining characteristics relating to image input of the image processing apparatus, and an image processing method. About.

  Generally, when examining the color characteristics of a scanner, a calibration chart (color chart) called IT8 is used. In the IT8 calibration chart, a plurality of patches having a uniform size with sides of about 8 mm to 10 mm are arranged side by side.

  The color characteristics of the scanner are estimated based on the R (red), G (green), and B (blue) values of the image data obtained by scanning the calibration color chart with the scanner, and image recognition conditions (parameters) Or changing the color of the image data.

Conventionally, a reference chart provided with a character area of a predetermined number of characters and a halftone dot area of a predetermined number of pixels is read by a scanner unit, and the number of read character pixels and the number of halftone dots are compared with preset determination values. A technique for correcting the region discrimination parameter for each pixel based on the comparison result is disclosed (for example, see Patent Document 1).
JP 2005-210492 A

  Some scanners have stable characteristics when a patch of a certain area or more is read. However, when the area of the patch is as small as about 1 to 5 square millimeters, the characteristics may vary.

  However, when using a calibration chart in which patches are configured with a uniform size with a certain area or more, such as IT8, it is impossible to detect such characteristic fluctuations. When the patch size is smaller than that of the IT8 calibration chart, correction based on the IT8 calibration chart is insufficient, and a correct result cannot be obtained.

  This problem occurs even when the reference chart described in Patent Document 1 is used.

  The present invention has been made in view of the above circumstances, and an image processing apparatus capable of eliminating the influence of fluctuations in characteristics of a reading unit in a minute region of image data, a calibration color chart used in the apparatus, and an image It is an object to provide a processing method.

The above problem is solved by the following means.
(1) Reading means for reading an original, recognizing means for recognizing a predetermined pattern of image data depending on whether image data of the original read by the reading means meets a predetermined recognition condition, and the recognition means An image processing apparatus comprising: correction means for correcting a recognition condition by the recognition means based on a size of a pattern to be recognized by the recognition means.
(2) The image processing apparatus according to (1), wherein the correction unit changes a correction amount based on a reading method of the reading unit.
(3) The image processing apparatus according to (1) or (2), wherein the correction unit changes a correction amount according to a density of a pattern to be recognized.
(4) The image processing apparatus according to any one of Items 1 to 3, wherein the correction of the recognition condition by the correction unit is performed when the size of the pattern to be recognized is small.
(5) Image processing comprising: reading means for reading image data; and correction means for correcting the image data based on the size of the processing target area of the image data read by the reading means. apparatus.
(6) The image processing apparatus according to (1), wherein the correction unit changes a correction amount based on a reading method of the reading unit.
(7) The image processing apparatus according to (5) or (6), wherein the correction unit changes a correction amount according to a density of image data.
(8) The image processing apparatus according to any one of Items 5 to 7, wherein the correction of the image data by the correction unit is performed when the size of the processing target area is small.
(9) A calibration chart used in the image processing apparatus according to any one of items 1 to 8, wherein a patch having a plurality of sizes is formed.
(10) The calibration chart according to item 9, wherein the plurality of size patches are equal to or higher than a predetermined density and have a plurality of densities.
(11) The plurality of sizes of patches having a predetermined size or less are formed more than the others, or the patches having a predetermined density or more are formed more than the others. The calibration chart according to 11.
(12) A step of reading a document, a step of recognizing a predetermined pattern of image data depending on whether or not the image data of the read document matches a predetermined recognition condition, and a size of the pattern to be recognized. And correcting the recognition condition based on the image processing method.
(13) An image processing method comprising: reading image data; and correcting the image data based on a size of a processing target area of the read image data.

  According to the invention described in item (1) above, since the recognition condition by the reading unit is corrected based on the size of the pattern to be recognized, the characteristic variation of the reading unit with respect to the minute region can be absorbed. Recognition accuracy can be improved.

  According to the invention described in item (2) above, the influence of characteristic fluctuations caused by the reading method of the reading means, for example, a method of reading an image by directly placing a document on a document table or a method of using an automatic paper feeder is taken into account. In addition, it can be eliminated.

  According to the invention described in item (3) above, it is possible to perform optimal correction according to the density.

  According to the invention described in the preceding item (4), the correction can be performed efficiently only in the case of a minute size in which the characteristic variation occurs.

  According to the invention described in item (5), since the image data is corrected based on the size of the processing target area of the image data, it is possible to absorb the characteristic variation of the reading unit with respect to the minute area.

  According to the invention described in the preceding item (6), it is possible to eliminate the influence of the characteristic variation caused by the difference in the type of reading method of the reading means.

  According to the invention described in item (7), the optimum correction according to the density is possible.

  According to the invention described in item (8) above, it is possible to perform correction efficiently only in the case of a minute size in which characteristic fluctuation occurs.

  According to the invention described in item (9) above, since a patch having a plurality of sizes is formed on the calibration chart, the use of this calibration chart can eliminate the influence of the characteristic fluctuation of the reading means of the image processing apparatus. it can.

  According to the invention described in item (10) above, since the patches of a plurality of sizes are composed of a plurality of densities that are equal to or higher than a predetermined density, the use of this calibration chart eliminates the influence of characteristic fluctuations particularly on the density. be able to.

  According to the invention described in the preceding item (11), since a plurality of patches of a plurality of sizes are formed to have a predetermined size or less, or more than a predetermined concentration than those having a predetermined concentration, the calibration chart is used. Thus, it is possible to eliminate the influence of the characteristic variation on the density.

  According to the invention described in item (12) above, since the recognition condition is corrected based on the size of the pattern to be recognized, characteristic fluctuations with respect to the micro area can be absorbed, and the recognition accuracy of the micro area is improved. be able to.

  According to the invention described in item (13), since the image data is corrected based on the size of the processing target area of the image data, it is possible to absorb the characteristic variation of the reading unit with respect to the minute area.

  Hereinafter, an embodiment of the present invention will be described with reference to the drawings.

  FIG. 1 is a block diagram showing an electrical configuration of an MFP to which an image processing apparatus according to an embodiment of the present invention is applied.

  1, the MFP includes an image reading unit 100, an image processing unit 200, an image forming unit (print engine) 300, and an operation panel 400.

  The image reading unit 100 is a scanner unit provided with a CCD or the like. The image reading unit 100 reads an image of a document and outputs read data of R, G, and B.

  The image processing unit 200 converts data read by the image reading unit 100 into digital image data, performs various correction processes, and then performs cyan (C), magenta (M), yellow (Y), and black (K). Convert to print data.

  The image forming unit 300 forms an image on a sheet based on print data.

  The image processing unit 200 is connected to a CPU 202 that controls the whole, a memory 204 that stores various programs, an image memory 206 that stores image data, a print data creation unit 208, and a specific pattern detection unit 210. Yes.

  In the print data creation 208, print data is created from the input read data, temporarily stored in the image memory 206, and then sent to the image forming unit 300. The processing in the print data creation unit 208 is actually hardware and software processing, and is a well-known technique, so a description thereof is omitted.

  The specific pattern detection unit 210 determines whether or not there is a specific pattern (such as a mark) present in the image data input from the image reading unit 100, and prohibits copying in the MFP when the specific pattern is detected. Take measures such as

  The operation panel unit 400 has a display unit (not shown) for inputting various settings for operations and displaying messages.

  FIG. 2 is a configuration diagram that also serves as an operation sequence of image recognition processing by the specific pattern detection unit 210.

  The image recognition processing is described here as software processing by a CPU (not shown) of the specific pattern detection unit 210, but at least a part of the image recognition processing may be configured as a hardware circuit.

  The image input unit 10 receives three color images of R (red), G (green), and B (blue) with a density of 8 bits (256 gradations).

  Next, in order to extract a specific color from the input multi-valued image, the binarization processing unit 12 performs color extraction processing by determining whether or not the density of the input image is within a predetermined reference density range. Thereby, a binarized image is obtained. In the following description, an image within the reference range is defined as “black pixel”, and an image outside the range is defined as “white pixel”.

  This binarization is performed at the resolution of the multi-value image density data (first resolution), and the color extraction processing result (binarization data) is stored in the memory 14. The memory 14 also stores in advance binarized data within a detection range (16 pixels × 16 pixels) of a plurality of types of marks (circular patterns).

  Next, in order to increase the speed of pattern detection, the resolution is lowered within a range where pattern detection is possible. Therefore, the resolution conversion unit 16 performs resolution conversion from the first resolution binarized image stored in the memory 14 to the second resolution binarized image for recognition processing. The binarized image of the second resolution after the resolution conversion is stored in the memory 18.

  Next, with respect to the binarized data whose resolution has been reduced, the mark area candidate detection unit 20 detects a predetermined mark area candidate from the binarized image in the memory 18, and the pattern matching unit 24 detects the detected mark area. Are compared with a predetermined criterion pattern, and the degree of similarity (compatibility) is determined.

  The detection result determination unit 26 comprehensively determines the results determined by the pattern matching unit 24, and outputs a result of whether or not a mark area has been detected.

  Examples of the mark area candidate include a circular pattern and a character pattern of a specific color. In the case of a circular pattern, the center and radius of the detected circular pattern are calculated, the roundness of the circular pattern is determined, and the degree of roundness is determined.

  Next, an outline of pattern recognition by the image reading unit (scanner unit) 100 in this MFP will be described.

<About calculation of characteristics>
Here, the relationship between the size of the region to be recognized and the characteristics of the scanner is calculated, and based on this, the image recognition parameter is corrected or the image is corrected. A specific example of the characteristic calculation method will be described below. Note that the color of the area (density, RGB value) is also considered as one of the variation factors.

  FIG. 3 shows changes in scan values when the area size is changed for a specific color patch.

  FIG. 4 shows changes in scan value when the density is changed for patches of the same size.

When the target color is fixed, only the relationship between the area size and the changed color is extracted as shown in FIG. The relationship is expressed by an approximate expression method or a unified LUT (Look-Up Table). The unit of the area size is “mm ² ”.

  When the size of the target patch is fixed, only the relationship between the color density and the changed color is extracted as shown in FIG. If the target color or size is not fixed, it is necessary to express the characteristics according to two variables.

  An example of expressing the characteristics is shown below.

  First, based on an approximate expression representing the relationship between the region size and the correction amount or a one-dimensional LUT, a variation relationship at the darkest density is calculated. Next, a coefficient is applied to the variation relation depending on the input color density. At the darkest density, the coefficient is “1”, and the coefficient value decreases as the density increases.

  The characteristics for the two variable elements may be expressed by this equation, or the characteristics may be expressed using a two-dimensional LUT as shown in FIG.

  In the LUT of FIG. 5, when the color (read value) and the size of the area after the change are input, the LUT is interpolated to calculate the correction value of the read value.

  Also, as shown in FIG. 3, when the area size is increased or the density of the area is below a certain level (brighter) as shown in FIG. Therefore, when these non-fluctuating sizes and reading values are input, the correction process may not be performed or may be omitted from the LUT.

  Further, when the reliability of the minute patch is extremely low due to the low resolution of the scanner, the correction in the minute region may be omitted. Alternatively, the characteristic value of the minute region may be estimated from the characteristic tendency of the reliable size.

<About correction using characteristics>
There are two methods for correction using the above characteristics. One is a method of changing a parameter (recognition condition) for recognizing an image, and the other is a method of correcting the image itself.
(I) Method for Changing Recognition Parameters Parameters used for image recognition are changed according to the calculated characteristics so that a highly accurate recognition result can be obtained.

  FIG. 6 is a sequence showing a flow of processing for changing the recognition parameter.

  In FIG. 6, an image is input in step S1, and the input image data is binarized in step S2. At this time, based on the characteristic extracted (calculated) in step S5, the binarization threshold value as a parameter is changed in step S6.

In step S3, the binarized data is recognized by pattern matching, and in step S4, the recognition result is output.
(1) Recognition parameter change (size variation)
When the color to be detected is fixed and the size of the parameter (character or mark) to be recognized is different, the color parameter used for recognition is changed according to the size of the mark to be recognized.

  First, the relationship between the region size and the read value is examined for the color of the mark to be detected, and a one-dimensional LUT is created as shown in FIG. A one-dimensional LUT is created for each channel (R, G, B) of the input image.

  Next, the rough size of the character size to be recognized is checked. For example, binarization may be performed using recognition parameters that are not corrected, and the character size may be estimated from the distribution.

  Then, the recognition parameter is changed according to the mark size to be recognized. After performing binarization with higher accuracy, recognition processing is performed. A one-dimensional LUT is used for parameter correction.

  When the character size to be recognized is large, a parameter correction amount corresponding to a large region size is calculated. When trying to recognize a small character, a parameter correction amount corresponding to a small region size is calculated. Finally, recognition processing is performed.

By performing such a correction and performing the recognition process, the image recognition accuracy can be increased.
(2) Change recognition parameters (multiple sizes, multiple colors)
When the size of the mark to be detected is fixed and there are a plurality of colors, the area of the size to be recognized and the characteristics when the input color is changed by the size are checked, and the parameters are corrected.

  First, the relationship between the input color and the read value is examined for the size of the mark to be detected, and a one-dimensional LUT is created as shown in FIG. A one-dimensional LUT is created for each channel (R, G, B) of the input screen.

Next, color parameters (Rmax, Rmin, Gmax, Gmin, etc.) for recognizing the mark color to be detected are input to the one-dimensional LUT, and the color parameters are corrected. Finally, recognition processing is performed based on the corrected color parameter.
(3) Change of recognition parameter (change by reading method)
When there are a plurality of reading methods of the image reading unit 100 (for example, when a document is placed directly on the platen glass and when an automatic paper feeder is used), as shown in FIG. The characteristics relating to the density and the correction amount are examined, and the parameters are corrected according to the size and density of the mark to be detected.

  First, in the color to be detected, the relationship between the mark size and the read value is examined to create a one-dimensional LUT. A one-dimensional LUT is created for each channel (R, G, B) of the input image.

  Then, it is determined whether or not the characteristic result is different between the case where the document is placed directly and the case where the automatic paper feeder is used. If the result is different, the correction is performed as follows.

  (A) Of the plurality of reading methods, a reference reading method (for example, a method of placing a document directly on a document table) is determined.

  (B) A recognition parameter is determined in accordance with a reference reading method. Subtract other one-dimensional LUTs to create a differential LUT.

  (C) When image data is read by a method other than the reference reading method, the parameter is corrected using the differential LUT according to the mark size, and recognition is performed.

  In the above embodiment, the method of dynamically changing the parameter when the color is fixed and the size is variable has been described. However, the same processing is performed when the size is fixed, the color is variable, or all are variable. can do.

If the color and size to be recognized are fixed, the recognition rate may be increased by converting the recognition parameter by the above method and setting a parameter that further includes the recognition parameter.
(II) Method of correcting the image itself The RGB value of the input image is corrected based on the correction table according to the detected characteristics.

  FIG. 10 is a sequence showing image data correction processing at this time.

  In FIG. 10, in step S11, an image is input (read), and in step S12, a region to be processed is extracted from the input image. Specifically, the input image is analyzed and divided into regions composed of the same color. At this time, when regions of similar colors are adjacent to each other, they may be integrated into one region.

  On the other hand, in step S15, the relationship between the region size and the color of the input image and the read value is checked, and in step S16, a correction table is created. It is assumed that the correction table is composed of a two-dimensional LUT as shown in FIG. When the characteristic is linear, an approximate expression may be used instead of the correction table.

  The correction table is created for each channel (R, G, B) of the input image. When the characteristics of the RGB channels are similar, processing may be performed with one correction table.

  In step S13, the color and area of each region are input to the correction table created earlier, and RGB correction values corresponding to the color and area of each region are acquired. The print data creation unit 208 of the MFP corrects the color of each area according to the output correction value. Finally, in step S14, the corrected image is output.

  In this embodiment, the processing may be applied only when the area size is a certain value or less. The reason is that as the region size increases, the variation in the detected characteristic decreases, and the significance of the correction decreases.

  Further, as described above, the correction table may be changed depending on the reading method of the image reading unit 100 (when a document is placed directly on the glass surface of the document table or when an automatic paper feeder is used). .

  Next, a calibration chart for examining characteristics will be described.

  In this embodiment, in order to detect a change in the reading characteristic of the reading unit 100 due to the size variation, as shown in FIG. 11, a plurality of patches P with different sizes are provided on a calibration color chart T for confirming the color characteristic. It is arranged.

  The patch P having a different size has a circular shape and is changed from a diameter of 10 mm to a diameter of 1 mm. For the smallest diameter of 1 mm, a donut-shaped patch with a hollow portion is also prepared.

  These patches are scanned by the image reading unit 100, and the relationship between the size and the read value is calculated. Further, as described above, the image recognition parameter built in the MFP is changed or the input image is corrected using the calculated characteristics.

  By using such a calibration chart T, it is possible to increase the accuracy of image recognition for a minute area or improve the precision of a minute area in a scanned image.

  In addition, even when the reading value of a minute region varies depending on the type of reading method, stable image recognition and image output can be obtained by this correction regardless of the reading method.

Since this characteristic variation is likely to occur in a dark color and a small size region, a patch (preferably an area of 50 mm ² or less) with a dark color (for example, a color darker than at least the gradation 128) and a small size The number of circular patches may be larger than that of other gradations and areas. In addition, it is preferable to include a patch having an area of 5 mm ² or less. Further, the size variation direction may be a double scanning direction so as not to pick up a difference in characteristics of the CCD in the main scanning direction in the image reading unit 100.

  FIG. 11 shows an example of a calibration chart T for A4, and shows a case where the target is a single color. FIG. 12 shows an example of a calibration chart T for A4, where the target is a plurality of colors. The calibration chart T in FIGS. 11 and 12 is provided with a plurality of circular patches P arranged in a darker color than the halftone and having different sizes.

  The patch P having the smallest diameter (diameter 1 mm) has a donut shape. The calibration chart T is read, for example, by placing it on the glass of the document table so that the CCD scans from top to bottom.

  In this embodiment, the example in which the image processing apparatus is an MFP including the image reading unit 100 as a scanner unit has been described. However, the present invention is not limited to the MFP and can be applied to a single scanner apparatus.

1 is a block diagram showing an electrical configuration of an MFP to which an image processing apparatus according to an embodiment of the present invention is applied. FIG. 5 is a configuration diagram also serving as an image recognition processing operation sequence by a specific pattern detection unit in an MFP. FIG. 10 is a characteristic diagram showing fluctuations in reading values according to the region size in a dark color region. FIG. 6 is a characteristic diagram illustrating fluctuations in reading color according to density in a minute region. It is a figure which shows an example of the two-dimensional LUT which correct | amends the characteristic fluctuation | variation of an image reading part. It is a flowchart which shows the process which changes a recognition parameter. It is a conceptual diagram regarding preparation of the one-dimensional LUT which is a correction table for every color. It is a conceptual diagram regarding creation of a one-dimensional LUT that is a correction table for each mark size. It is a conceptual diagram regarding the recognition parameter correction | amendment by the difference in a reading system. It is a flowchart which shows the correction process of an input image. FIG. 2 is a diagram illustrating an example of a calibration chart used in the MFP of FIG. 1. FIG. 7 is a diagram showing another example of a calibration chart used in the MFP of FIG. 1.

Explanation of symbols

DESCRIPTION OF SYMBOLS 100 Image reading part 200 Image processing part (correction means)
202 Control unit 208 Print data creation unit (correction unit)
210 Specific pattern detection unit (recognition means)
P Patch T Calibration chart

Claims (13)

Reading means for reading a document;
Recognizing means for recognizing a predetermined pattern of image data depending on whether image data of a document read by the reading means matches a predetermined recognition condition;
Correction means for correcting the recognition condition by the recognition means based on the size of the pattern to be recognized by the recognition means;
An image processing apparatus comprising:   The image processing apparatus according to claim 1, wherein the correction unit changes a correction amount based on a reading method of the reading unit.   The image processing apparatus according to claim 1, wherein the correction unit changes a correction amount according to a density of a pattern to be recognized.   The image processing apparatus according to claim 1, wherein the correction of the recognition condition by the correction unit is performed when the size of the pattern to be recognized is small. Reading means for reading image data;
Correction means for correcting the image data based on the size of the processing target area of the image data read by the reading means;
An image processing apparatus comprising:   The image processing apparatus according to claim 1, wherein the correction unit changes a correction amount based on a reading method of the reading unit.   The image processing apparatus according to claim 5, wherein the correction unit changes a correction amount according to a density of image data.   The image processing apparatus according to claim 5, wherein the correction of the image data by the correction unit is performed when the size of the processing target area is small. A calibration chart used in the image processing apparatus according to claim 1,
A calibration chart in which patches of a plurality of sizes are formed.   The calibration chart according to claim 9, wherein the plurality of sizes of patches have a predetermined density or more and a plurality of densities.   The patch of the plurality of sizes according to claim 9 or 11, wherein a plurality of patches having a predetermined size or less are formed more than other patches, or a patch having a predetermined density or more is formed more than other patches. The calibration chart described. Scanning the document,
Recognizing a predetermined pattern of image data depending on whether or not the image data of the read original meets a predetermined recognition condition;
Correcting the recognition condition based on the size of the pattern to be recognized;
An image processing method comprising: Reading image data;
Correcting the image data based on the size of the processing target area of the read image data;
An image processing method comprising:

Images