JP6015606B2 - Inspection method and inspection apparatus - Google Patents

Description

  The present disclosure relates to an inspection method for inspecting an object to be inspected based on an image obtained by imaging the object to be inspected, and an inspection apparatus for inspecting an object to be inspected by such an inspection method.

  When the inspection object is inspected, the quality of the inspection object is often determined by performing image processing based on an image obtained by photographing the inspection object. Thereby, compared with the case where an inspector inspects the inspected object visually, for example, an objective and efficient inspection becomes possible.

  Various techniques are disclosed for such an inspection apparatus for an object to be inspected. For example, Patent Document 1 discloses an image processing apparatus that captures a display image of a display device in an inspection of the display device, and determines the quality of the display image using the Mahalanobis distance based on the captured image. Has been.

JP 2004-101338 A

  Thus, it is desirable that the inspection of the object to be inspected is performed efficiently, and further improvement in efficiency is expected.

  The present disclosure has been made in view of such problems, and an object thereof is to provide an inspection method and an inspection apparatus capable of efficiently inspecting an object to be inspected.

The inspection method of the present disclosure is obtained by photographing an object to be inspected, obtains a photographed image having a brightness change region in which the value of brightness information continuously changes in a predetermined direction, and brightness information for each pixel in the photographed image Based on the luminance information for each pixel in the luminance change region, to determine the maximum value and the minimum value of the difference in luminance information value between adjacent pixels in a predetermined direction in the luminance change region, the histogram , The square distance of Mahalanobis is obtained as an inspection value based on the maximum value and the minimum value, and the object to be inspected is inspected based on the inspection value .

The inspection apparatus according to the present disclosure includes an image processing unit and an inspection unit. The image processing unit obtains a histogram based on luminance information for each pixel in a captured image having a luminance change region in which the value of luminance information is continuously changed in a predetermined direction, obtained by imaging the object to be inspected , Based on the luminance information for each pixel in the luminance change region, the maximum value and the minimum value of the difference in luminance information values between adjacent pixels in a predetermined direction in the luminance change region are obtained. The inspection unit obtains the Mahalanobis square distance as an inspection value based on the histogram , the maximum value, and the minimum value, and inspects the object to be inspected based on the inspection value .

  In the inspection method and the inspection apparatus of the present disclosure, the object to be inspected is inspected based on a captured image obtained by photographing the object to be inspected. At that time, a histogram is obtained based on the luminance information for each pixel in the captured image, and the object to be inspected is inspected based on the histogram.

  According to the inspection method and the inspection apparatus of the present disclosure, since the inspection object is inspected based on the histogram, the inspection object can be efficiently inspected. In addition, the effect described here is not necessarily limited, and there may be any effect described in the present disclosure.

It is a block diagram showing the example of 1 composition of the inspection device concerning an embodiment of this indication. It is explanatory drawing showing an example of the image for a test | inspection. It is explanatory drawing showing an example of a histogram. It is explanatory drawing which shows an example of the luminance information in the luminance change part shown in FIG. It is explanatory drawing which shows an example of the difference in the brightness | luminance change part shown in FIG. It is explanatory drawing showing the other example of a histogram. It is explanatory drawing which shows the other example of the luminance information in the luminance change part shown in FIG. It is explanatory drawing which shows the other example of the difference in the brightness | luminance change part shown in FIG. It is a flowchart showing the example of 1 operation of an inspection device. 8 is a flowchart illustrating an operation example of the statistical data calculation process illustrated in FIG. 7. FIG. 8 is a flowchart illustrating an operation example of the threshold value determination process illustrated in FIG. 7. 8 is a flowchart illustrating an operation example of the inspection process illustrated in FIG. 7.

Hereinafter, embodiments of the present disclosure will be described in detail with reference to the drawings. The description will be given in the following order.
1. Embodiment 2. FIG. Application examples

<1. Embodiment>
[Configuration example]
FIG. 1 illustrates a configuration example of an inspection apparatus according to an embodiment. The inspection device 1 inspects the image display device 9 by determining whether the display image displayed on the image display device 9 such as a liquid crystal display or an organic EL (Electro-Luminescence) display is acceptable. The inspection method according to the embodiment of the present disclosure is embodied by the present embodiment, and will be described together. The inspection apparatus 1 includes an imaging unit 10, an image processing unit 20, and a determination unit 30.

  The imaging unit 10 is configured using, for example, a camera or the like, and images a predetermined inspection image Q displayed on the image display device 9.

  FIG. 2 shows an example of the inspection image Q displayed on the image display device 9. The inspection image Q has a plurality of luminance change portions Q1. The luminance change portion Q1 is a portion where the luminance continuously changes in the horizontal direction (the horizontal direction in the figure). In this example, eight luminance change portions Q1 are provided side by side in the vertical direction (vertical direction in the figure). In each of the luminance change portions Q1, the luminance is continuously changed for various colors. Specifically, in this example, the eight luminance changing portions Q1 continuously change the luminance for each of yellow, cyan, green, white, white, magenta, red, and blue from the top. . At that time, in the adjacent luminance change portion Q1, the direction in which the luminance changes is set to be opposite to each other. Specifically, for example, in the luminance change portion Q1 for the top yellow, the luminance increases toward the right, and in the luminance change portion Q1 for the second cyan color from the top, the luminance increases toward the left. It has become. The same applies to the other luminance change portions Q1. As described above, the inspection image Q includes various colors and includes a change in luminance for each color. By displaying such an inspection image Q on the image display device 9 that is the inspection object, the image display device 9 makes it easy to judge whether the image is good or not. It is possible to make it easier to check whether such a problem exists.

  The imaging unit 10 captures such an inspection image Q displayed on the image display device 9 and acquires a captured image P. The captured image P has red (R) luminance information IR, green (G) luminance information IG, and blue (B) luminance information IB for N pixels. At that time, the number of pixels of the captured image P and the number of pixels of the inspection image Q do not necessarily need to match. Specifically, one pixel of the captured image P may correspond to one pixel of the inspection image Q, and one pixel of the captured image P corresponds to a plurality of pixels of the inspection image Q. It may be. The photographing unit 10 supplies the photographed image P to the image processing unit 20.

  The image processing unit 20 obtains the histogram information IH based on the captured image P supplied from the imaging unit 10. Specifically, the image processing unit 20 generates a histogram HR for red luminance information based on N pieces of red (R) luminance information included in the captured image P, and includes N included in the captured image P. A histogram HG for the green luminance information is generated based on the green (G) luminance information, and the blue luminance information is determined based on the N blue (B) luminance information included in the captured image P. The histogram HB is generated. At that time, the image processing unit 20 divides the entire range of the luminance information value into k luminance classes CC for each of the red, green, and blue luminance information, and obtains the frequency for each luminance class CC to obtain the histogram HR. , HG, HB are generated.

  FIG. 3 shows an example of the histogram HR. Although an example of the histogram HR is shown here, the same applies to the histograms HG and HB. In this example, 16 (k = 16) luminance classes CC are set. The histograms HR, HG, and HB change depending on the display characteristics of the image display device 9 and the presence or absence of defective pixels. That is, the histograms HR, HG, and HB are so-called feature amounts that represent the features of the image display device 9 that is the inspection target.

  The image processing unit 20 generates such histograms HR, HG, and HB, and supplies these to the determination unit 30 as histogram information IH. That is, the image processing unit 20 supplies 48 (= 16 × 3) parameters to the determination unit 30 as the histogram information IH.

  The image processing unit 20 also has a function of obtaining the difference information ΔI based on the luminance change portion Q1 included in the captured image P. Specifically, the image processing unit 20 determines, for example, between adjacent pixels in the horizontal direction based on the luminance change portion Q1 for the fifth white color from the top among the eight luminance change portions Q1 illustrated in FIG. The difference ΔIR between the values of the red luminance information IR is obtained, and the maximum value MAXR and the minimum value MINR are obtained. Similarly, the image processing unit 20 obtains the difference ΔIG in the value of the green luminance information IG between adjacent pixels in the horizontal direction based on the luminance change portion Q1, and obtains the maximum value MAXG and the minimum value MING. The difference ΔIB between the values of blue luminance information IB between adjacent pixels in the horizontal direction is obtained, and the maximum value MAXB and the minimum value MINB are acquired.

  FIG. 4A shows the value of the luminance information IR in the horizontal direction of the luminance change portion Q1, and FIG. 4B shows the difference ΔIR in the value of the red luminance information IR between adjacent pixels. Here, an example of the luminance information IR and the difference ΔIR is shown, but the same applies to the luminance information IG and the difference ΔIG, and the luminance information IB and the difference ΔIB. In this example, as shown in FIG. 4A, the luminance information IR changes linearly according to the pixel coordinates in the horizontal direction. As a result, the difference ΔIR between the values of the luminance information IR becomes uniform as shown in FIG. 4B. The image processing unit 20 acquires the maximum value MAXR and the minimum value MINR out of the difference ΔIR. The maximum values MAXR, MAXG, MAXB and the minimum values MINR, MING, MINB vary depending on the display characteristics of the image display device 9 and the presence / absence of defective pixels. That is, the maximum values MAXR, MAXG, MAXB and the minimum values MINR, MING, MINB are feature quantities of the image display device 9 to be inspected.

  In this way, the image processing unit 20 acquires the maximum values MAXR, MAXG, MAXB and the minimum values MINR, MING, MINB, and supplies them to the determination unit 30 as difference information ΔI. That is, the image processing unit 20 supplies six (= 2 × 3) parameters as difference information ΔI to the determination unit.

  In this manner, the image processing unit 20 supplies the histograms HR, HG, and HB (48 parameters in this example) as the histogram information IH to the determination unit 30, and also includes the maximum values MAXR, MAXG, MAXB, and the minimum value MINR. , MING, MINB (six parameters in this example) are supplied to the determination unit 30 as difference information ΔI. That is, the image processing unit 20 supplies two types of feature amounts for the image display device 9 to the determination unit 30.

  The image processing unit 20 may be configured by hardware or software. That is, such an operation performed in the image processing unit 20 is a simple operation as compared with a general image processing operation, and has a small operation load. Can be configured.

  The determination unit 30 obtains the Mahalanobis square distance based on the histogram information IH and the difference information ΔI using the MTS (Mahalanobis Taguchi System) method, and determines whether the captured image P is good or bad based on the Mahalanobis square distance ( The image display device 9 related to the captured image P is inspected by classifying it into a good image or a defective image.

  Further, the determination unit 30 also performs processing (statistical data calculation processing and threshold value determination processing, which will be described later) for generating information necessary for the inspection as a preliminary stage of the inspection of the image display device 9. It has become.

[Operation and Action]
Then, operation | movement and an effect | action of the test | inspection apparatus 1 of this Embodiment are demonstrated.

(Overview of overall operation)
First, an overall operation outline of the inspection apparatus 1 will be described with reference to FIG. The inspection image Q displayed on the image display device 9 is captured to acquire a captured image P, and the captured image P is supplied to the image processing unit 20. The image processing unit 20 obtains the histogram information IH and the difference information ΔI based on the captured image P, and supplies these to the determination unit 30. The determination unit 30 obtains the Mahalanobis square distance based on the histogram information IH and the difference information ΔI, and determines whether the captured image P is good or bad (classifies as a good image or a bad image) based on the Mahalanobis square distance. The image display device 9 related to the captured image P is inspected.

(Histogram information IH and difference information ΔI)
Based on the captured image P, the image processing unit 20 obtains histogram information IH and difference information ΔI. Several examples of the operation will be described below.

  FIG. 5 shows an example of the histogram HR, where (A) shows a case where the captured image P is a good image, and (B) shows a case where the captured image P is a defective image. Thus, differences appear in the histograms HR, HG, and HB depending on the display characteristics of the image display device 9 and the presence or absence of defective pixels. In that case, as shown in FIG. 5, a difference arises in the frequency in a plurality of luminance classes CC.

  In the inspection apparatus 1, since the quality of the captured image P is determined based on the histograms HR, HG, and HB, the difference in the captured image P can be detected efficiently. That is, the total value of the frequencies in all luminance classes CC in the histograms HR, HG, and HB matches the number N of pixels of the captured image P and is a fixed value. Therefore, for example, when the luminance information of a certain pixel in the captured image P deviates from the desired luminance information, the frequency in the luminance class CC to which the desired luminance information belongs decreases and deviates from the desired luminance information. The frequency in the luminance class CC to which the actual luminance information belongs increases. That is, a defect of one pixel appears as a change in the frequency of the two luminance classes CC. Thereby, the detection sensitivity at the time of detecting the difference of the picked-up image P can be raised, and the difference of the picked-up image P can be detected efficiently.

  The detection sensitivity at the time of detecting the difference in the captured image P can be adjusted by the number of luminance classes CC. Specifically, increasing the luminance class CC can increase the detection sensitivity, and decreasing the luminance class CC can decrease the detection sensitivity. This is because, as the luminance class CC is larger, the range of the luminance information value in each luminance class CC becomes narrower, so that the increase and decrease in frequency are less likely to be offset in each luminance class CC. .

When the detection sensitivity is increased by increasing the luminance class CC, the width CW of the luminance information value in each luminance class CC is reduced. In this case, the width CW needs to satisfy the following expression.
CW> σ (1)
Here, σ is a variation (standard deviation) in luminance information of each pixel in the captured image P when the same display image is repeatedly captured on the video display device 9. By setting the width CW of the value of the luminance information in each luminance class CC so as to satisfy this equation, it is possible to reduce the possibility that the frequency in each luminance class CC is affected by measurement variations. In addition, when the luminance class CC is increased, the processing amount in the determination unit 30 increases. Therefore, it is desirable to determine the number of luminance classes CC in consideration of this processing amount.

  FIG. 6A shows the value of the luminance information IR in the horizontal direction of the luminance change portion Q1 when the captured image P is a defective image, and FIG. 6B shows the difference ΔIR. In this example, as shown in FIG. 6A, the luminance information IR is lower than a desired value in a certain pixel (part W1), so that the difference ΔIR has a maximum value near this pixel. MAXR and minimum MINR appear. The maximum value MAXR is larger than that when the captured image P is a good image (FIG. 4B), and the minimum value MINR is greater than that when the captured image P is a defective image (FIG. 4B). Small value. Thus, differences appear in the maximum values MAXR, MAXG, MAXB and the minimum values MINR, MING, MINB depending on the display characteristics of the image display device 9 and the presence or absence of defective pixels.

  In the inspection apparatus 1, the quality of the captured image P is determined based on the maximum values MAXR, MAXG, MAXB and the minimum values MINR, MING, MINB. It can be detected with high sensitivity. That is, when the defect shown in FIG. 6A occurs, as shown in FIGS. 4B and 6B, there is a great difference between the two parameters (maximum value MAXR and minimum value MINR). In the histogram HR, for example, the frequency in one luminance class CC is decreased by one, and the frequency in another luminance class CC is only increased by one, so that the difference in the histogram HR is small. . As described above, the quality of the captured image P is determined based on the maximum values MAXR, MAXG, MAXB and the minimum values MINR, MING, MINB, so that a difference in luminance information in some pixels is detected with high sensitivity. Can do.

  In this way, the inspection apparatus 1 detects the presence or absence of defects in the entire captured image P by performing pass / fail judgment of the captured image P based on the histogram information IH (histograms HR, HG, HB), and detects the difference information ΔI ( Based on the maximum values MAXR, MAXG, MAXB and the minimum values MINR, MING, MINB), the presence or absence of a defect in some pixels is detected by making a pass / fail judgment on the captured image P. Accordingly, it is possible to efficiently check the presence or absence of defects from the entire captured image P to the details. Specifically, for example, when an inspector inspects, for example, the inspector visually inspects the captured image P while looking at a so-called limit sample. Such visual inspection requires time for inspection and may increase the inspection cost. In addition, it is difficult to make an objective judgment, and there is a possibility that a subjective judgment of an inspector may be entered. Further, a defective product may be judged as a non-defective product due to an inspection mistake. Furthermore, when a defect that is not assumed in advance occurs, there is a possibility that it cannot be determined appropriately. On the other hand, in the inspection apparatus 1 according to the present embodiment, the quality of the captured image P is determined based on the histogram information IH and the difference information ΔI. This makes it possible to make objective judgments based on the quantified information, make it possible to make a judgment in a short time and at a low cost, and determine that a defective product may be judged as a good product. Can be reduced. Further, even when a defect that is not assumed in advance occurs, the defect is quantitatively represented by the histogram information IH and the difference information ΔI, so that the possibility of determining such a defective product as a non-defective product can be reduced. it can.

(Detailed operation of the detection device 1)
FIG. 7 illustrates an operation example of the inspection apparatus 1. In the inspection apparatus 1, before performing the inspection process (step S3) of the image display apparatus 9, the statistical data calculation process (step S1) and the threshold value determination process (step S2) are performed in advance, and the threshold value MD _th ² Determined later. Then, in the inspection process (step S3), and using this threshold MD _th ^2, inspecting the image display device 9. Hereinafter, each process will be described in detail.

  FIG. 8 shows statistical data calculation processing. In the statistical data calculation process, an average vector AVE and a covariance inverse matrix R are obtained based on a plurality of video display devices (non-defective samples) determined to be non-defective by the inspector. The statistical data calculation process will be described in detail below.

  First, the imaging unit 10 acquires the captured image P by capturing the inspection image Q displayed on the non-defective sample (step S11). Then, the photographing unit 10 supplies the photographed image P to the image processing unit 20.

  Next, the image processing unit 20 calculates histogram information IH and difference information ΔI based on the captured image P (step S12). Then, the image processing unit 20 supplies the histogram information IH and the difference information ΔI to the determination unit 30.

  Next, the determination unit 30 stores a total of 54 parameters included in the histogram information IH (48 parameters in this example) and the difference information ΔI (6 parameters in this example) in the memory in the determination unit 30. Store (step S13).

  The inspection apparatus 1 repeats steps S11 to S13 for M (for example, about 100) non-defective samples. At that time, the process is repeatedly performed so that sampling is performed at the same density in the non-defective sample space (step S14).

Through these steps S11 to S14, the parameters obtained from the M non-defective samples are stored in the memory of the determination unit 30. The determination unit 30 generates a feature vector X based on this parameter (step S15). This feature vector X is expressed by the following equation.
X = (X ₁ X ₂ X ₃ ... X ₅₄ ) (2)
Xi = {x _{i, 1} x _{i, 2} x _{i, 3} ... X _{i, M} } (3)
This Xi (i = 1 to 54) represents a set of data for M non-defective samples for one parameter out of 54 parameters.

Next, the determination unit 30 calculates an average vector AVE based on the feature vector X (step S16). This average vector AVE is expressed by the following equation.
AVE = (A ₁ A ₂ A ₃ ... A ₅₄ ) (4)
Here, Ai (i = 1 to 54) is an average value of M pieces of data included in the set Xi shown in Expression (3).

この共分散行列Ｃに基づいて、共分散逆行列Ｒを求める。この共分散逆行列Ｒは、次式で表されるものである。
Ｒ＝Ｃ^-1／５４ ・・・（７）
ここで、Ｃ^-1は、共分散行列Ｃの逆行列であり、次式で表されるものである。

式（７）で示したように、共分散行列Ｃの逆行列を、パラメータの数（５４）で割るようにしたので、５４次元における距離が１となる単位空間で評価することができるようになるため、計算を容易にすることができる。 Next, the determination unit 30 calculates a covariance matrix C based on the feature vector X and the average vector AVE to obtain a covariance inverse matrix R (step S17). This covariance matrix C is expressed by the following equation.

Based on this covariance matrix C, a covariance inverse matrix R is obtained. This covariance inverse matrix R is expressed by the following equation.
R = C ⁻¹ / 54 (7)
Here, C ⁻¹ is an inverse matrix of the covariance matrix C, and is expressed by the following equation.

As shown in Equation (7), the inverse matrix of the covariance matrix C is divided by the number of parameters (54), so that it can be evaluated in a unit space with a distance of 1 in 54 dimensions. Therefore, calculation can be facilitated.

  Thus, this flow ends.

FIG. 9 shows threshold value determination processing. In the threshold value determination process, an image is displayed based on one video display device (defective product sample) determined to be defective by the inspector, and the average vector AVE and covariance inverse matrix R obtained by the statistical data calculation process. The Mahalanobis square distance MD _th ² used as a threshold when the display device 9 is inspected is determined. Hereinafter, the threshold value determination process will be described in detail.

  First, in the determination unit 30, the imaging unit 10 acquires the captured image P by capturing the inspection image Q displayed on the defective specimen (step S21). Then, the photographing unit 10 supplies the photographed image P to the image processing unit 20.

  Next, the image processing unit 20 obtains histogram information IH and difference information ΔI based on the captured image P (step S22). Then, the image processing unit 20 supplies the histogram information IH and the difference information ΔI to the determination unit 30.

Next, the determination unit 30 calculates the feature vector X _NG based on a total of 54 parameters included in the histogram information IH (48 parameters in this example) and the difference information ΔI (6 parameters in this example). Generate (step S23).

Next, the determination unit 30, a feature vector X _NG obtained in step S23, determined by the step S16, S17 statistics calculation average based on the vector AVE and covariance inverse matrix R, and the feature vector X _NG The Mahalanobis square distance MD _NG ² with the average vector AVE is calculated (step S24). This Mahalanobis square distance MD _NG ² is represented by the following equation.
MD _NG ² = (X _NG −AVE) R (X _NG −AVE) ^T (9)

Next, the determination unit 30 determines the Mahalanobis square distance MD _th ² used as a threshold value (step S25). This threshold value MD _th ² is determined so as to satisfy the following expression.
MD _th ² <MD _NG ² (10)

  Thus, this flow ends.

In this example, one defective sample is used. However, when a plurality of defective samples can be prepared, the Mahalanobis square distance MD _NG ² is obtained for each using Equation (9). The smallest of them can be selected.

The inspection apparatus 1 thus determines the threshold value MD _th ² by performing the statistical data calculation process (FIG. 8) and the threshold value determination process (FIG. 9). Then, the inspection apparatus 1 inspects the image display apparatus 9 using the threshold value MD _th ² .

FIG. 10 shows an inspection process of the image display device 9. In this test process, it based on the captured image P of the image display device 9 to be inspected (specimen), obtains the Mahalanobis squared distance MD ^2, comparing the squared distance MD ² and threshold MD _th ² of the Mahalanobis Thus, the image display device 9 is inspected. Hereinafter, this inspection process will be described in detail.

  First, in the determination unit 30, the imaging unit 10 acquires the captured image P by capturing the inspection image Q displayed on the specimen (step S31). Then, the photographing unit 10 supplies the photographed image P to the image processing unit 20.

  Next, the image processing unit 20 obtains histogram information IH and difference information ΔI based on the captured image P (step S32). Then, the image processing unit 20 supplies the histogram information IH and the difference information ΔI to the determination unit 30.

Next, the determination unit 30 calculates the feature vector X _S based on a total of 54 parameters included in the histogram information IH (48 parameters in this example) and the difference information ΔI (6 parameters in this example). Generate (step S33).

Next, the determination unit 30, a feature vector X _S obtained in step S33, based on the mean vector AVE and covariance inverse matrix R calculated by the statistical data calculating process step S16, S17 of the feature vectors X _S calculating the Mahalanobis squared distance MD ² between the average vector AVE (step S34). Squared distance MD ² of this Mahalanobis is represented by the following formula.
MD ² = (X _S -AVE) R (X _S -AVE) ^T (11)

Next, the determination unit 30, Mahalanobis square distance MD ² obtained in step S34, is compared with a threshold value MD _th ² obtained in step S25 in the threshold determination process (step S35). If squared distance MD ² Mahalanobis is smaller than the threshold value _{^{MD th 2 (MD 2 <MD}} th 2) determines that the sample is good (step S36), Mahalanobis squared distance MD ² is sill If the value is MD _th ² or more (MD ² ≧ MD _th ² ), the sample is determined to be defective (step S37).

  Thus, this flow ends.

[effect]
As described above, in the present embodiment, the quality of the captured image is determined based on the histogram information, so that the difference between the captured images can be detected efficiently.

  In the present embodiment, since the quality of the captured image is determined based on the difference information, it is possible to detect a difference in luminance information in some pixels in the captured image with high sensitivity.

  In the present embodiment, the quality of the captured image is determined based on the histogram information and the difference information. Therefore, the entire captured image can be inspected in a short time and at a low cost. The risk that a non-defective product is determined to be a non-defective product can be reduced. Moreover, even when a defect that is not assumed in advance occurs, the risk of being judged as a non-defective product can be reduced.

[Modification 1]
In the above embodiment, the quality of the captured image is determined based on the histogram information and the difference information. However, the present invention is not limited to this, and instead, for example, the histogram information is obtained without acquiring the difference information. The quality of the captured image may be determined based on the above. Even in this case, it is possible to detect the presence or absence of defects in the entire captured image P.

[Modification 2]
In the above embodiment, the entire range of luminance information values is divided into 16 luminance classes CC. However, the present invention is not limited to this. Instead, it is divided into 15 or less luminance classes CC or more. May be. Similarly, in the above embodiment, the histogram information and difference information have a total of 54 parameters. However, the present invention is not limited to this. Instead, 53 or less parameters or 55 or more parameters are used. You may make it have.

<2. Application example>
In the above-described embodiment and the like, the display image displayed on the image processing device 9 is inspected. However, the present invention is not limited to this, and can be used for various inspections. Specifically, for example, it can be used for appearance inspection of articles. In this case, for example, it is possible to inspect whether or not the color of the article is uneven and whether or not a part of the pattern is missing when the article is patterned. It is also possible to inspect whether the size of the article is appropriate.

  As described above, the present technology has been described with the embodiment and the modified examples, and specific application examples thereof. However, the present technology is not limited to the embodiments and the like, and various modifications can be made.

  For example, in the above-described embodiment and the like, the inspection image Q shown in FIG. 2 is used. However, the present invention is not limited to this, and various colors are included, and the luminance change for each color. Any image may be used as long as it contains an image.

  Further, for example, the image processing unit 20 performs the image processing using the luminance information of red (R), green (G), and blue (B), but is not limited to this. Instead, for example, image processing may be performed using luminance information of cyan, magenta, and yellow.

  Further, for example, in the above-described embodiment and the like, the image processing unit 20 performs image processing using luminance information of red (R), green (G), and blue (B). For example, the image processing may be performed using luminance information of cyan, magenta, and yellow instead.

  The square distance of Mahalanobis can also be applied to sound quality inspection, for example. In this reference example, specifically, for example, a speaker is an inspection target, and a sound for inspection is emitted from the speaker. As a sound source for inspection, for example, white noise or pink noise can be used. Then, the sound emitted from the speaker is subjected to a fast Fourier transform (FFT) to obtain a spectrum. This spectrum corresponds to the histogram according to the present embodiment. The sound quality can be inspected by obtaining the Mahalanobis square distance using each value of this spectrum as a feature amount.

  In addition, the effect described in this specification is an illustration to the last, and is not limited, Moreover, there may exist another effect.

  In addition, this technique can be set as the following structures.

(1) Obtaining a photographed image of the inspected object,
Obtain a histogram based on luminance information for each pixel in the captured image,
An inspection method for inspecting the object to be inspected based on the histogram.

(2) The inspection method according to (1), wherein a square distance of Mahalanobis is obtained as an inspection value based on the histogram, and the inspection object is inspected based on the inspection value.

(3) The captured image has a luminance change region in which the value of luminance information continuously changes in a predetermined direction;
Based on the luminance information for each pixel in the luminance change region, to determine the maximum value and the minimum value of the difference in luminance information value between adjacent pixels in the predetermined direction in the luminance change region,
The inspection method according to (2), wherein a Mahalanobis square distance is obtained as the inspection value based on the maximum value and the minimum value in addition to the histogram.

(4) The captured image has luminance information for each of a plurality of colors,
The inspection method according to (3), wherein the histogram, the maximum value, and the minimum value are obtained for each of the plurality of colors.

(5) Acquire reference photographed images of a plurality of reference inspected objects,
Obtain the square distance of Mahalanobis as a reference value based on the reference photographed image,
The inspection method according to any one of (1) to (4), wherein the inspection object is inspected by comparing the inspection value with the reference value.

(6) The inspection method according to (5), wherein the plurality of reference inspection objects include a plurality of non-defective products and one defective product.

(7) The inspection method according to any one of (1) to (6), wherein the object to be inspected is captured to acquire the captured image.

(8) The inspected object is a display device;
The inspection method according to any one of (1) to (7), wherein the captured image is an image obtained by capturing a display screen of the display device.

(9) The inspection method according to any one of (1) to (8), wherein the inspection object is classified into a plurality of groups based on an inspection result of the inspection object.

(10) An image processing unit that obtains a histogram based on luminance information for each pixel in a captured image obtained by capturing an inspected object;
An inspection apparatus comprising: an inspection unit that inspects the object to be inspected based on the histogram.

(11) The inspection apparatus according to (10), wherein the inspection unit determines a Mahalanobis square distance as an inspection value based on the histogram, and inspects the inspection object based on the inspection value.

DESCRIPTION OF SYMBOLS 1 ... Inspection apparatus, 9 ... Image display apparatus, 10 ... Imaging part, 20 ... Image processing part, 30 ... Determination part, AVE ... Average vector, CC ... Luminance class, IH ... Histogram information, [Delta] I ... Difference information, IR, IG , IB ... Luminance information, MAXR, MAXG, MAXB ... Maximum value, MINR, MING, MINB ... Minimum value, P ... Photographed image, Q ... Inspection image, Q1 ... Luminance change part, R ... Covariance inverse matrix, X, X _NG , X _S ... feature vector, MD ² , MD _NG ² , MD _th ² ... square distance of Mahalanobis.

Claims (8)

Obtaining a photographed image having a luminance change region obtained by photographing the object to be inspected, and the value of luminance information continuously changing in a predetermined direction ,
Obtain a histogram based on luminance information for each pixel in the captured image,
Based on the luminance information for each pixel in the luminance change region, to determine the maximum value and the minimum value of the difference in luminance information value between adjacent pixels in the predetermined direction in the luminance change region,
An inspection method for obtaining a square distance of Mahalanobis as an inspection value based on the histogram , the maximum value, and the minimum value, and inspecting the object to be inspected based on the inspection value . The captured image has luminance information for each of a plurality of colors,
Obtain the histogram, the maximum value, and the minimum value for each of the plurality of colors.
The inspection method according to claim 1 . Acquire reference images of multiple reference objects,
Obtain the square distance of Mahalanobis as a reference value based on the reference photographed image,
The inspection object is inspected by comparing the inspection value with the reference value.
The inspection method according to claim 1 or 2 . The plurality of reference inspected objects include a plurality of non-defective products and one defective product.
The inspection method according to claim 3 . The inspection method according to any one of claims 1 to 4 , wherein the object to be inspected is acquired to acquire the captured image. The object to be inspected is a display device,
The inspection method according to claim 1, wherein the captured image is an image obtained by capturing a display screen of the display device. The inspection method according to any one of claims 1 to 6 , wherein the inspection object is classified into a plurality of groups based on an inspection result of the inspection object. Obtaining a histogram based on luminance information for each pixel in a captured image having a luminance change region in which the value of luminance information is continuously changed in a predetermined direction, obtained by photographing the object to be inspected, and in the luminance change region An image processing unit for obtaining a maximum value and a minimum value of a difference in value of luminance information between adjacent pixels in the predetermined direction in the luminance change region based on luminance information for each pixel ;
An inspection apparatus comprising: an inspection unit that obtains a Mahalanobis square distance as an inspection value based on the histogram , the maximum value, and the minimum value, and inspects the object to be inspected based on the inspection value .

Images