JPH06160298A - Automatic color tone deciding method - Google Patents

Abstract

PURPOSE:To decide defect due to different color tone on a scale corresponding to the human sense of color difference through a simple constitution by determining a quantitative value for multi-stage evaluation using the weight of color region obtained through multi-stage evaluation of a visual inspector and measurement of color difference. CONSTITUTION:Color tone decision result of an inspector, chromaticity values measured 130 at a plurality of specified points of a printed matter 101, and a watching point in one color screen of a printed matter 140 to be inspected are inputted as three- dimensional address, for M sheets of printed matter (sample) provided for determining decision reference, to a quantitative decision reference calculating section 111 where quantitative values, color tones, and quantitative evaluation values of defective color tone are calculated for multi-stage evaluation. Two-dimensional image of the printed matter 140 picked up by means of a color camera 150 is then inputted to a deciding/ processing section 112 where difference from chromaticity value of reference image and an evaluation quantitative value as a print image are calculated at a watching point on the printing face and the value thus calculated is compared with a threshold value predetermined based on a quantitative value for multi-stage evaluation thus deciding pass/fail of the printed matter 140.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a technique for quantifying a color tone judgment in a color difference inspection of a color printed matter. In particular, in the color tone judgment visually conducted by an inspector, the judgment is automated. To achieve this, the present invention relates to an automatic color tone determination method that quantifies human color tone determination and enables appropriate determination based on that.

[0002]

2. Description of the Related Art Generally, in a printing company, the quality of a product is judged by a human visual inspection. Judgment is made by comparison with a normal product as a reference. In the case of inspection regarding color tone, as a result of comparison with the standard, if the difference in color is visually small by the inspector, it is judged to be normal, and if the difference is large, it is judged to be defective (defective).

In the case of automatic inspection, a printed matter is input as an image by a color camera, and RGB values are obtained as color signals for each pixel. In order to make a judgment close to human sense, RGB values are usually converted into chromaticity values in the CIE L ^* a ^* b ^* system recommended by CIE (International Commission on Illumination) as a uniform color space, and CIE L values are converted. ^The color difference is evaluated using the color difference value in the ^* a ^* b ^* color space.
As the conversion, for example, the RGB value is converted into a chromaticity value in the XYZ color system which is the standard value of the color by a matrix conversion of 3 × 3, and then the conversion formula of the CIE presentation (for example, “Color Society of Japan” Color Science Handbook ", published by The University of Tokyo Press, pp.141
-142) to convert to L ^* , a ^* , b ^* values.

In the inspection judgment, the CI for each local area of the image is determined.
Calculate the color difference value ΔEab * in the EL ^* a ^* b ^* color system,
If the color difference is smaller than a preset threshold value indicating the boundary between normal and defective, it is determined to be normal, and if the color difference is large, it is determined to be defective.

[0005]

However, in the conventional color tone determination method, the threshold value for normality and defect is determined without considering the relationship between the stepwise color difference value and the evaluation result of visual inspection. Therefore, there is a problem in that it is not possible to quantitatively express a stepwise difference in color and to make a judgment based on a standard that takes the quantitatively expressed value into consideration.

Further, in the actual color tone inspection, there are a part that is considered important and a part that is not considered important depending on the pattern on the printed surface. Since the conventional color tone determination method does not consider the difference in the degree of important attention due to the difference in area, it is not possible to make a determination that matches the human sense of the color tone inspection when determining the color tone of the entire print surface. There are also problems.

The present invention is intended to solve these problems and is capable of quantitatively evaluating a stepwise difference in color tone, and is regarded as important in the inspection of local areas on the printing surface. It is an object of the present invention to provide a means capable of making a judgment close to the visual judgment by a human inspector by setting a certain degree.

[0008]

Therefore, according to the present invention,
For a color area in a print image obtained by inputting a color image, a color difference between a reference image and an inspection image is calculated by a signal value obtained by converting a color signal value, and a quality determination is performed by a predetermined threshold value. The process shown in is performed. In FIG. 1, 101 is a printed matter to be used as a sample for obtaining judgment reference data, 111 is a quantitative judgment reference calculation unit for calculating quantitative judgment reference data, and 112 is a judgment process in an automatic color tone inspection device for determining the quality of the printed matter. Section, 13
Reference numeral 0 denotes a color tone measuring device for measuring the color tone of the printed matter 101 for each color region, 140 denotes a printed matter to be inspected, and 150 denotes a color camera for obtaining a color image of the printed matter 140.

First, in order to obtain quantitative judgment reference data, a process 201 allows a visual inspector to set a color tone for each of a plurality of samples of the printed matter 101 whose color tone is changed stepwise from good to bad. Input the multi-level comprehensive evaluation judgment result judged by inspection. On the other hand, in the process 202, the color tones of the samples of the plurality of prints 101 are measured by the color tone measuring device 130 or the like in the plurality of color areas on the print surface, and the color difference measurement value of the prints of the reference print is obtained.

Next, in processing 203, the weight for each color region when the score for each multi-level evaluation of the visual inspector is calculated as the weighted sum of the color difference measurement values, the eigenvector of the principal component analysis is calculated. Calculate as Then, by the processing 204, using the obtained weights of the color areas and the color difference measurement values, a quantitative value that integrates a plurality of color areas is calculated and divided by the sum of the weights of the color areas. Obtain a quantitative value for the evaluation. Thus, for example, quantitative value of determination A is 2.0, determined A ^- Determination value 2.8, ...,
The quantitative value of the judgment B ⁻ is 7.0, ... By using this data as a reference of a threshold value used when the automatic color tone inspection device determines quality, it is possible to make a determination close to a human visual determination.

In the pass / fail judgment of the actual printed matter 140, the color camera 1 is processed by the process 211 by using the information about the point of importance in the judgment and the degree of the importance, which is presented in advance by the visual inspector for the pattern to be inspected.
The evaluation value of the color tone of the printed matter 140 input from 50 is quantitatively calculated. Then, in the process 212, the quality of the printed matter 140 to be inspected is compared by comparing the threshold value determined based on the quantitative value for each of the multi-level evaluations with the evaluation value calculated for the inspected printed matter. To judge.

According to the second aspect of the present invention, a new color difference conversion is further calculated in which the color difference is calculated as a value obtained by adding a weight to the square of the difference between each of the criteria of lightness, saturation and hue and adding the weights. After obtaining the formula, the color difference is calculated, the quantitative evaluation value of the color tone defect of the printed matter is calculated, and the inspection judgment is performed.

[0013]

[Function] In the automatic color tone inspection of the color printed matter, the color difference between the reference image and the inspection image is calculated by the signal value obtained by converting the color signal value with respect to the color area in the printed image obtained by inputting the color image, and by the threshold value. In the inspection method for determining the quality, the result of the color tone judgment by the visual inspector is obtained for the printed matter samples from the non-defective product to the defective product, and the degree of importance in the inspection of the multiple color regions on the printed surface is determined by the eigenvector of the principal component analysis. Then, using the obtained weights and color difference measurement values, quantitative values for each multi-level evaluation are obtained, and the important points and their degrees presented by the visual inspector for the inspected pattern are shown. Quantitatively calculate the evaluation value of the color tone of the printed matter by using the color difference, and add the color difference to the square of the difference between each of the criteria of lightness, saturation, and hue with weighting. A new color difference conversion formula that is calculated as a calculated value is obtained, color difference calculation is performed, a quantitative evaluation value of the color tone defect of the printed matter is calculated, and inspection judgment is performed. Defect judgment based on the difference of can be shown in multiple stages on a scale that is similar to human color difference perception. Further, it is possible to perform the color tone determination in consideration of the color tone differences of many local areas on the printing surface.

[0014]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An automatic color tone determination method as an embodiment of the present invention will be described with reference to the drawings.

As shown in FIG. 2, the reference print S and the inspection pair
There is an elephant print T. Image of standard printed matter with color camera
Input the image and store it in the memory. The printed matter to be inspected
Input the image and store it in the memory. For color inspection
Is the average value of the RGB values of the corresponding area on the image memory.
Calculated, RGB value for each of the reference image and inspection image
Is a uniform color system CIEL^*a^*b^*Chromaticity value of color system
Converted to CIEL^*a ^*b^*Calculate the distance between the two in space
Color difference is obtained by taking out. The calculated color difference value,
By correlating with the visual inspection evaluation, automatic judgment can be performed.
To do. Below is a method to associate color difference values with visual evaluation.
Show.

First, the amount of ink is adjusted at the time of printing to prepare a plurality of samples having different color tones from good to bad. Then, have all the inspectors at the printing site make a visual judgment. For example, A, A ⁻ , B ⁺ , B, B ⁻ , C ⁺ , C, C ⁻ , which is obtained by adding + and − corrections to the evaluation from A to D,
The determination may be made in 10 steps of D ⁺ and D. At the time of judgment, it is instructed to clearly indicate the judgment result as a printed matter and the area on the printing surface that has an influence on the judgment for all the samples. As a result, as shown in FIG. 3, a table showing the judgment for each printed matter and the influence area in the judgment can be made. By inputting this result, a quantitative evaluation value corresponding to 10-step evaluation judgment is calculated.

The difference in color with respect to the reference printed matter to be inspected is locally different for each pattern area. For example, as shown in FIG.
There are 2, ... The color difference for each color region can be accurately measured by a commercially available colorimeter without depending on conversion from RGB signals. In order to calculate the quantitative evaluation for the visual judgment, it is necessary to use the unknown weight as the color difference at what point in the region that the inspector emphasizes in the visual judgment and to what extent. is there. For example, the weight for each local color area x on the printing surface is set as l _x .

The measured color difference value of each local area is weighted by l _x , and a numerical evaluation value for each multi-level evaluation of each inspector is calculated by the following principal component analysis. A plurality (J) of local color areas within the important visual area indicated by the inspector are selected and l _i (i = 1, 1) that minimizes the evaluation function E according to equation (2) by the Lagrange's undetermined constant method 2, ..., J).

Let e _k (n) be the ΔEab * value at the measurement point k of the nth sample, and P (n) be the score of the nth printed sample,

[0020]

[Equation 1]

Then, the evaluation function is as follows.

[0022]

[Equation 2]

However, M is the number of printed matter samples. Further, μ _{a * (m)} is the average of the evaluation values of all samples having the same evaluation as the evaluation of the m-th sample, and is calculated by the equation (3).

[0024]

[Equation 3]

Here, N _{a * (m)} is 1 of the m-th sample
It is the number of samples having the same evaluation as the 0-level evaluation a ^* (m).

[0026]

[Equation 4]

However, S _ij is obtained by the equation (6).

[0028]

[Equation 5]

[0029]

[Equation 6]

Λ is obtained from the equation (4), and all the inspectors have eigenvectors with positive values.
Let J be a weighting amount. The evaluation value for each multi-level evaluation of the inspector can be calculated by the formula (3).

For example, as shown in FIG. 4, for the inspector # 1, a quantitative evaluation value for 10-step evaluation from A to D is calculated. In this example, the judgments A, A ⁻ , B ⁺ ,
The quantitative values of ... Are calculated as 2.0, 2.5, 3.2 ,.

Next, the evaluation values are set so that the evaluation values from A to D monotonically increase from the good judgment to the bad judgment. The 10-point evaluation score of each inspector is approximated by the equation (7) using the least squares method.

[0033]

[Equation 7]

However, x corresponds to a 10-step evaluation from A to D, has a value of 1.0 to 10.0, and y is an evaluation point for each evaluation. By the approximation by this quadratic equation,
For example, a 10-point evaluation point can be approximated as shown by a curve indicated by a dotted line in FIG.

The logarithmic function as shown in equation (8) may be used instead of the quadratic equation as shown in equation (7).

[0036]

[Equation 8]

If the equation (7) or the equation (8) monotonically increases in the range of x from 1.0 to 10.0, this is appropriate as an approximate function. Now, the evaluation value obtained here is
Since it is a weighted sum of J measurement points, the evaluation value for each measurement point is the sum of the weights l _i (i = 1, 2, ..., J) for each J measurement point. It is obtained by dividing the value.

As described above, the quantitative value for the multi-level evaluation by the inspector is calculated. In the 10-point evaluation, A and B are normal, and C and D are defects. It can also be said that the boundary between B and C is a range where the determination is uncertain. It is only necessary to use the obtained evaluation values for the 10-step evaluation to perform a good / bad or indeterminate automatic determination.

As described above, the color difference value ΔEab * and the visual judgment can be associated with each other by performing the calculation considering the weight for the judgment of each color area from the color difference values of the plurality of color areas.

The actual visual judgment is made based on the color difference of the entire printing surface including a plurality of areas. The area to be emphasized in the inspection can be determined in advance by the visual inspector for each pattern of the printed matter. The contribution rate when calculating the judgment evaluation value as a printed matter may be set as the weighting difference between the important area and the other area.

For example, as shown in FIG. 2, the inspector himself designates a plurality of color areas 1, 2, ... Which the inspector considers important when judging the color tone, and presents the degree of importance for each. Ask. Let the importance of the presented i region be l _i .

According to this, according to the equation (1), e
_k (n) is the color difference calculation value at the measurement point k of the nth sample,
Using P (n) as the quantitative evaluation value of the n-th printed matter sample, P (n) is calculated by the following equation.

[0043]

[Equation 9]

By making a judgment with a threshold value for P (n), it is possible to make a judgment in consideration of the difference in local color tone of the printing surface. As l _{k in the} equation (9), the importance presented by the inspector may not be used as it is, but the importance shown may be further weighted.

In the above method, CIEL ^* is used as the color difference value ^.
Although the color difference value in the a ^* b ^* color system is adopted, it does not necessarily match the human sense of color difference. Therefore, the measured and calculated L ^* a ^* b ^* values are represented by lightness L, saturation C, and hue H.
In addition, the method of calculating the weight of each element of the LCH using the visual determination result is effective. That is,

[0046]

[Equation 10]

Then, ΔL, ΔC, and ΔH are calculated respectively. Regarding the lightness, ΔL = ΔL ^* . Standard a
^{The *} and b ^* values are a ^* and b ^* . The angle formed by the hues of the reference color and the inspection color on the a ^* b ^* plane is θ. Further, if the saturations of the reference color and the inspection color are C ₁ and C ₂ , respectively,

[0048]

[Equation 11]

[0049]

[Equation 12]

[0050]

[Equation 13]

[0051]

[Equation 14]

Can be converted into a difference in LCH. In the actual human sense, it is considered that the difference in lightness, the difference in saturation, and the difference in hue are not equivalent, and the respective weightings are different. Therefore,

[0053]

[Equation 15]

The weighting factors l, c, and h are calculated as follows. J
The weight for each measurement point and the quantitative value of the 10-step evaluation are the values of the eigenvector calculated as described above, and (1
It is calculated by minimizing the sum of squares of the difference between (ΔE) ² according to the equation (5) and the square of the quantitative value of the 10-step evaluation of the color sample.

From this result, a new color difference calculation formula is obtained. A value obtained by dividing each weighting coefficient by a weight l of ΔL, that is, an equation normalized by setting 1 to 1 may be used. The calculation of the evaluation points in the case of automatic evaluation using the color difference value by the new color difference calculation formula is performed by the same method as described above.

The conversion from the RGB value obtained by the color camera to the XYZ chromaticity value which is the standard value of the color is performed in advance in XYZ.
Input RGB images of many color samples with known values to obtain RGB values,
The conversion formula may be determined by the least squares method. As a result, for example,

[0057]

[Equation 16]

A conversion formula like the following is obtained. The conversion from XYZ values to the CIE L ^* a ^* b ^* color system is C as shown above.
It is performed by a predetermined formula of IE presentation. A flow of a method of inputting an image of a printed matter by a camera and performing a determination process by image processing will be described with reference to FIG. An image of the printing surface is picked up by a color camera, and RGB video signals V are obtained. The image data obtained from the printing surface of the reference print is the reference image memory 11
The image data obtained from the printing surface of the print object to be inspected is stored in the inspection image memory 12. The position of the local area designated by the visual inspector and the difference in weight with respect to the local area are stored in the local area data memory 20 in advance.

At the time of color tone inspection, RG is set for each local area.
The B signal value is converted to obtain the chromaticity values 21 and 22, and the color difference 30 is calculated based on the color difference calculation formula. A quantitative evaluation value 31 is calculated using the color difference 30 and the weight amount for the local area. The determination 32 is performed by performing threshold processing on the calculated evaluation value 31.

FIG. 6 shows a system configuration example of an automatic color tone inspection apparatus to which the present invention is applied. An inspector's judgment check table 102 is created for M printed matter 101. The determination is made by, for example, a 10-step evaluation of A, B, C and the like. The result of this determination is input as data to the quantitative determination criterion calculation unit 111 in the processor 110. In addition, a color tone measuring device 1 such as a color micro-analyzer for a plurality of predesignated locations on the printing surface of the M printed matter 101.
The colorimetric chromaticity value is measured by 30. The measurement result is
CIEL ^* a ^* b ^* is converted into chromaticity values in the color system,
It is input to the quantitative determination criterion calculation unit 111.

In order to input the gazing point of the inspector at the time of inspection, for example, one of the printed materials 140 is input by the color camera 150, displayed on the monitor screen 120, and the cursor 122 operated by the mouse 123 is displayed. The designated location is used to indicate the gaze point as a rectangular area on the screen. The obtained gazing point is the 3D address of the image.
It is input to the quantitative determination criterion calculation unit 111.

As described above, after the inspector's judgment check data, color tone measurement data, and gazing point address are input to the quantitative judgment reference calculation unit 111, A, B are calculated by the above-described quantitative value calculation method by principal component analysis. , C, etc. are calculated for the multi-level evaluation. This value is used for automatic evaluation.

In the case of the automatic inspection, first, the image of the printed matter to be the inspection reference is input. At this time, the printed matter 141 is conveyed by a belt conveyor 161 driven by rollers 160, and an image is input using a color camera 150 such as a CCD line sensor. When a two-dimensional image is obtained using a line sensor, the belt conveyor 161
The movement of the printed material due to the conveyance of the above can be used as the sub-scan for image capturing.

The reference image obtained as the RGB signal is input to the determination processing section 112 of the processor 110. In the determination processing unit 112, the RGB values are converted into the chromaticity values of the CIEL ^* a ^* b ^* color specification system by signal conversion processing such as matrix calculation for the gazing point on the printing surface to which the address is input as described above. The obtained chromaticity value is stored in the memory as a reference value.

When the inspection is performed, the printed matter 140 is sequentially conveyed by the belt conveyor 161, a two-dimensional image is obtained as described above, and the inspection image is input to the determination processing unit 112. Similar to the case of the reference image, after converting to the chromaticity value in the CIEL ^* a ^* b ^* colorimetric system, the difference between the gazing point on the printing surface and the chromaticity value of the reference image (color space name) Euclidean distance in) is calculated. Furthermore, the quantitative determination standard calculation unit 1
Using the weight of each gazing point calculated in 11, the evaluation quantitative value as a print image is calculated. Then, the quality determination is performed using the determination standard data calculated by the quantitative determination standard calculation unit 111, and the belt conveyor 16 is determined based on the determination result.
The sorting mechanism 162 provided on the first unit is operated to sort the printed matter into the non-defective stocker 164 and the defective stocker 163.

[0066] quality determination, for example when using a quantitative criterion inspector # 1, when a quantitative evaluation value as in the example shown in FIG. 4 were obtained, determination B ^- quantitative value of about 7.0,
Since the judgment C ⁺ is 9.0, with the intermediate value of 8.0 as the boundary, if the evaluation quantitative value calculated for the inspection image is less than 8.0, it is a non-defective product, and if it is 8.0 or more. If so, it may be set so that it is determined to be defective.

[0067]

As described above, according to the present invention,
This is useful for automatically determining the difference in color tone on the printing surface using a color camera, and the following excellent effects can be obtained with an extremely simple configuration.

By associating the color difference value calculated using the color signal value with the multi-level evaluation scale of the visual inspector, the defect judgment due to the difference in color tone is multi-staged in a scale that is similar to the human sense of color difference. Can be shown.

Color tone determination can be performed in consideration of the color tone differences of a large number of local areas on the printing surface.

[Brief description of drawings]

FIG. 1 is a diagram illustrating the principle of the present invention.

FIG. 2 is a diagram showing color regions in a reference printed matter and an inspection target printed matter.

FIG. 3 is a diagram showing a region affected in a multi-level evaluation and determination by a visual inspector performed on a plurality of printed materials.

FIG. 4 is a diagram showing a result of calculating quantitative values for multi-level evaluation in an example of the present invention.

FIG. 5 is a diagram showing a flow until automatic color tone determination using a color image according to an embodiment of the present invention.

FIG. 6 is a diagram showing a system configuration example of an automatic color tone inspection device to which the present invention is applied.

[Explanation of symbols]

 101 Printed Material (Sample) 111 Quantitative Judgment Criteria Calculation Section 112 Judgment Processing Section 130 Color Tone Measuring Instrument 140 Printed Material to Be Inspected 150 Color Camera 201-204 Processing in Quantitative Judgment Criteria Calculation Section 211-212 Processing in Judgment Processing Section

Claims (2)

[Claims] 1. A method for inspecting a color tone of a color printed matter, wherein a color difference between a reference image and an inspected image is calculated by a signal value obtained by converting a color signal value in a color area in a print image obtained by inputting a color image. Then, in the automatic color tone determination method using a computer that determines quality by comparing with a predetermined threshold value, a visual inspector is used for each of a plurality of printed matter samples whose color tone is gradually changed from good to bad. Inputting the multi-level comprehensive evaluation judgment result judged by the color tone inspection, and the process of obtaining the color difference measurement value of the printed matter sample with respect to the reference printed matter for a plurality of color areas on the printing surface,
As a weighted sum of the color difference measurement values, a process of calculating a weight for each color region when a score for each multi-level evaluation of the visual inspector is calculated as an eigenvector of the principal component analysis, A process of calculating a quantitative value that integrates a plurality of color regions using the weight of the color region and the color difference measurement value, and obtaining a quantitative value for each multi-level evaluation based on the quantitative value and the weight of the color region, Using the information about the place to be emphasized in the judgment and the degree of importance, which is presented by the visual inspector for the pattern to be inspected,
A process of quantitatively calculating an evaluation value of a color tone as an inspection target printed matter, a threshold value determined based on the quantitative value for each of the multi-level evaluations, and an evaluation value calculated for the inspection target printed matter. And a process of judging whether the printed matter to be inspected is good or bad by comparing the above. 2. The automatic color tone determination method according to claim 1, wherein in calculating the color difference, the chromaticity value in the CIEL ^* a ^* b ^* color system is converted into a value corresponding to lightness, saturation and hue, Color difference is the square of the difference from the standard of lightness, saturation, and hue,
The weighted value is added up, and the calculated quantitative value for each multi-level evaluation is converted into the weighting coefficient that is most approximated by the linear sum of squares of differences in lightness, saturation, and hue. An automatic color tone determination method, characterized in that a new color difference conversion formula is obtained by calculating according to, and the color difference is calculated using the obtained conversion formula.