JP4962763B2 - Defect inspection apparatus and defect inspection method - Google Patents

Description

  The present invention relates to a defect inspection apparatus and a defect inspection method for extracting a defect in a region to be inspected based on a feature amount obtained using an imaging optical device.

  As an inspection apparatus for detecting a pixel defect of a liquid crystal display device, an inspection apparatus that captures a display image by a camera and detects the defect by image processing is known. As shown in FIG. 8, the inspection apparatus captures the object 2 to be inspected, and image processing for capturing the image from the camera 2 and numerically processing the luminance information of the entire inspection area of the object 10 to be inspected. Device 103.

  FIG. 9 shows a processing procedure in the image processing apparatus 103. After capturing an image (step S11), as a preprocessing, correction of a camera defect or the like is performed, and image reduction is performed as necessary (step S12). In addition, when the purpose is to detect a bright spot defect (display defect in which a pixel shines during black display), a differential enhancement filter is applied to the captured image (step S13). Next, the image is binarized to extract defect candidates (step S14), and each is numbered (step S15). Next, a feature amount is calculated for each defect candidate (step S16). The feature amount is a numerical value for determining a defect, and is, for example, peak luminance, luminance volume, contrast, and the like. This feature amount is compared with a preset threshold value and identified (step S17), and output as a defect (step S18).

  Patent Document 1 describes a defect detection apparatus that images an inspection pattern displayed on an inspection object such as a flat panel display and detects defects based on image data obtained by the imaging.

JP 2005-156396 A

  However, there is a problem that a uniform image cannot be obtained over the entire inspection area due to the optical characteristics of the lens 21 provided in the camera 2. For example, the brightness of the image decreases around the area to be inspected. Further, the resolution of the lens 21 also decreases around the inspection area. For this reason, it is necessary to change the threshold value (step S17) for determining the defect between the vicinity of the center of the inspection area and the vicinity of the periphery. In addition, the resolution of the lens varies depending on the individual lens, and also varies greatly depending on the focus adjustment state of the lens. For this reason, an operation for readjusting and checking the threshold value of the entire inspection area is required for each lens or each time focus adjustment is performed. Furthermore, when a plurality of lenses are used in one inspection apparatus, it is necessary to set a threshold value for each lens. As described above, setting and managing threshold values cause an increase in maintenance time and cost, and also causes a reduction in the operation rate of the line.

  An object of the present invention is to provide a defect inspection apparatus and the like that can eliminate the influence of the optical characteristics of the apparatus on the inspection without requiring complicated work.

Defect inspection apparatus of the present invention is based on the feature amount obtained by using the imaging optical device, in the defect inspection apparatus for extracting elements individual defects constituting the pattern in the region to be inspected with a regular pattern, wherein An imaging unit that acquires an inspection image by imaging the inspected area with an imaging optical device, and a feature that acquires a feature amount distribution in the imaging area of the imaging optical device based on the inspection image obtained by the imaging unit The photographing of the imaging optical device based on a variation range of luminance between pixels based on the pattern appearing in the inspection image, which is a target for acquiring the distribution of the feature amount by the feature amount distribution obtaining unit Characteristic distribution acquisition means for acquiring a contrast distribution in the region, and based on the contrast distribution acquired by the characteristic distribution acquisition means. A feature amount correction unit that corrects the feature amount acquired by the feature amount distribution acquisition unit for each individual region in the imaging region, and the inspection target based on the feature amount corrected by the feature amount correction unit. And a detecting means for detecting a defect in the region.
According to this defect inspection apparatus, since the feature amount obtained via the imaging optical apparatus is corrected based on the distribution of the optical characteristics, the influence on the inspection of the optical characteristics of the apparatus can be eliminated without requiring a complicated operation. .

The imaging optical device may use a two-dimensional area sensor as an imaging element.

The imaging optical device may use a one-dimensional line sensor as an imaging element.

The inspection area may be a screen of a display device.

Defect inspection method of the present invention, based on the feature amount obtained by using the imaging optical device, in the defect inspecting method for extracting elements individual defects constituting the pattern in the region to be inspected with a regular pattern, wherein An imaging step of obtaining an inspection image by imaging the inspection area with an imaging optical device, and a feature amount distribution in the imaging area of the imaging optical device based on the inspection image obtained by the imaging step The imaging optical device based on a change range of luminance between pixels based on the pattern appearing in the inspection image, which is a target for acquiring a distribution of the feature amount in the feature distribution acquisition step; A characteristic distribution acquisition step for acquiring a contrast distribution in the region, and a characteristic distribution acquisition step before the characteristic distribution acquisition step; Based on the distribution of contrast, the feature amount acquired by the feature amount distribution acquisition step is corrected for each individual region in the imaging region, and the feature amount corrected by the feature amount correction step. And a detecting step for detecting a defect in the inspection region based on the detection step.

  According to the defect inspection apparatus of the present invention, since the feature amount obtained through the imaging optical device is corrected based on the distribution of the optical characteristics, the influence of the optical characteristics of the apparatus on the inspection can be reduced without requiring complicated work. Can be eliminated.

  According to the defect inspection apparatus of the present invention, since the threshold value of the feature amount for detecting the defect is corrected by the detection unit based on the distribution of the optical characteristic, the optical characteristic of the apparatus is not required to be complicated. The influence on inspection can be eliminated.

  According to the defect inspection method of the present invention, since the feature amount obtained through the imaging optical device is corrected based on the distribution of the optical properties, the influence of the optical properties of the device on the inspection can be reduced without requiring complicated work. Can be eliminated.

  According to the defect inspection method of the present invention, since the threshold value of the feature amount for detecting a defect is corrected based on the distribution of optical characteristics, it is possible to inspect the optical characteristics of the apparatus without requiring complicated work. The influence can be eliminated.

  Hereinafter, an embodiment of a defect inspection apparatus according to the present invention will be described with reference to FIGS.

  FIG. 1 is a block diagram showing the configuration of the defect inspection apparatus of this embodiment.

  As shown in FIG. 1, the defect inspection apparatus of the present embodiment includes a camera 2 that images a screen of a matrix display type liquid crystal display device 1 as an object to be inspected, and a liquid crystal display device 1 based on an imaging signal of the camera 2. And an image processing device 3 for extracting the defective pixels. The image processing apparatus 3 can be configured by a personal computer or the like.

  The image processing apparatus 3 includes a characteristic distribution acquisition unit 31 that acquires a distribution of optical characteristics of the camera 2 in a shooting region based on an image captured by the camera 2, and a distribution of the optical characteristics acquired by the characteristic distribution acquisition unit 31. Based on the feature amount correction unit 32 that corrects the feature amount obtained via the camera 2 based on the feature amount corrected by the feature amount correction unit 32, the detection unit detects a screen defect of the liquid crystal display device 1. 33.

  FIG. 2 is a flowchart showing the operation of the image processing apparatus 3.

  In step S1 of FIG. 2, the screen of the liquid crystal display device 1 that is lit black is photographed by the camera 2, and an image (imaging signal) from the camera 2 is captured.

  Next, in step S2, processing such as image correction for a defect or the like of the camera 2 or reduction of the captured image is executed. Next, in step S3, the captured image is subjected to a differential enhancement filter.

  Next, as a process for extracting defect candidates, in step S4, the image is binarized by setting the level higher than the preset threshold to “1” and the lower level to “0”, and the higher level part is selected as a defect candidate. Extract as Next, in step S5, each extracted defect candidate is numbered (labeled).

  Next, steps S6 to S9 are executed as processing for determining whether or not the defect is present.

  In step S6, a feature amount is calculated for each of the extracted defect candidates. The feature amount is a numerical value for finally determining the presence / absence of a defect, such as peak luminance, luminance volume, and contrast. The feature amount is appropriately defined according to the defect to be detected.

  Next, in step S7, contrast calculation is executed based on the captured image. This process corresponds to the function of the characteristic distribution acquisition unit 31.

  The contrast calculation method will be described below.

  FIG. 3A is a diagram illustrating an arrangement of display pixels of the liquid crystal display device 1. As shown in FIG. 3A, red (R), green (G), and blue (B) unit pixels (sub-pixels) are periodically arranged in the inspection area 1A (FIG. 1) of the liquid crystal display device 1. The size of the bright spot defect to be detected is approximately the same size as one unit pixel. For example, as shown in FIG. 3A, when the unit pixel 11 is abnormal, a bright spot defect occurs in that region.

  FIG. 3B is a diagram illustrating a part of an output signal (imaging signal) from the area sensor of the camera 2. As shown in FIG. 3B, the luminance detected by the area sensor changes periodically. This is because the luminance is not completely zero even when the liquid crystal display device 1 is lit black, and the backlight light leaks through the liquid crystal substrate. Further, since the leakage light has a higher luminance in the order of green (G), red (R), and blue (B), a light and shade pattern having a period corresponding to a pixel (pixel) is formed. This period has a spatial frequency of 1/3 of a sub-pixel which is a detection unit of a bright spot defect, and the amplitude (contrast) is proportional to the resolution of the lens.

  In step S7, the contrast of the camera 2 is calculated based on the output signal from the area sensor.

For example, assuming that the lens resolution at the spatial frequency of the sub-pixel and the resolution of the lens at the spatial frequency of 1/3 of the sub-pixel have a certain proportional relationship, the luminance amplitude of the output signal from the area sensor is By measuring, it is possible to know the resolution of the lens at that location (the resolution at the spatial frequency of the subpixel). As shown in FIG. 3B, when the luminance peak value is “L1” and the bottom value is “L2”, the contrast is
(L1-L2) / (L1 + L2)
Is required.

  Note that the contrast calculation can be performed by Fourier transforming the luminance data of FIG. 3B regardless of the luminance at the spatial frequency of 1/3, and the spatial frequency contrast of the subpixel can be obtained. Accurate value can be calculated.

  FIG. 4A is a diagram showing a positional relationship among the inspection region 1A, the lens 21 and the area sensor 22. The principal ray of the image at the center of the inspection area 1 </ b> A passes through the optical axis 21 a of the lens 21 and forms an image near the center of the area sensor 22. On the other hand, the principal ray of the image near the periphery of the inspection area 1A enters the lens 21 with the optical axis 21a of the lens 21 and the inclination θ, and forms an image near the periphery of the area sensor 22. At this time, with a general lens, even if the luminance of the inspection area 1A is uniform over the entire area, the brightness of the image of the area sensor 22 decreases in the periphery. This is known as a law (cos θ fourth power law) in which the brightness of the image is proportional to the fourth power of cos θ with respect to the inclination θ between the optical axis and the principal ray.

  Further, the resolution of the lens varies depending on the inclination θ of the principal ray with respect to the optical axis. The resolution is generally higher on the optical axis, and generally decreases as the aberration increases as it goes to the periphery (θ increases).

  FIG. 4B is a diagram illustrating the resolution of the lens as a graph. The horizontal axis represents the distance from the optical axis on the image plane, and the vertical axis represents the contrast when a black and white striped pattern is imaged. ing. Three sets of lines consisting of a solid line and a dotted line on the graph indicate the resolution when the fineness (spatial frequency) of the black and white stripe pattern is different. There are three types of spatial frequencies: 10 lines / mm (lp · mm), 20 lines / mm (lp · mm), and 40 lines / mm (lp · mm). This value indicates how many pairs of black and white striped patterns are arranged within 1 mm. The solid line indicates the contrast in the circumferential direction of the lens, and the dotted line indicates the contrast in the radial direction of the lens. For example, focusing on the contrast at 40 lines / mm having the highest spatial frequency in this graph, it can be seen that the contrast on the image plane 20 mm away from the optical axis is reduced to half or less on the optical axis.

  FIG. 5A illustrates the state of the output signal of the area sensor 22 when the inspection area 1A is imaged using such a lens. In this example, the peak 41 and the peak 44 are peaks corresponding to bright spot defects near the periphery of the image, and the peak 43 is a peak corresponding to bright spot defects near the center of the image. The peak 42 is noise near the center of the image and should not be detected as a bright spot defect. As described above, since the resolution of the lens is reduced at the periphery, the output value of the area sensor varies depending on the region to be inspected even at the same level of luminance peak (peak indicating a bright spot defect). Here, for the sake of explanation, assuming that the defect is detected based on the output signal (luminance) itself of the linear sensor 22, if the output signal of the linear sensor 22 is not corrected, for example, in the vicinity of the center of the image, a threshold value is set. It is necessary to detect a bright spot defect using a threshold value 52 near the periphery of the image 51.

  As described above, generally, the brightness and contrast of imaging change in a region to be inspected 1A instead of being constant. In the defect inspection apparatus of this embodiment, the area sensor 22 is divided into 100 regions (R00, R01,... R99) as shown in FIG. 5B, for example, and in step S7, the division is performed. The contrast is calculated for each region. In step S7, the calculated contrast is stored in the memory or the like of the image processing apparatus 3 in association with the region (R00, R01,... R99).

  Next, in step S8, the feature amount calculated in step S6 is corrected based on the contrast calculated in step S7 and stored in the memory or the like. This process corresponds to the function of the feature amount correction unit 32.

  In step S8, for example, the normalized feature value can be obtained by correcting the feature value according to the resolution of the lens by dividing the feature value by the contrast. In this case, the standardized feature value can be considered as a feature value when the resolution of the lens is ideal and the contrast is “1”. Therefore, the feature that eliminates the influence of individual lens differences and the focus adjustment of the camera 2 is excluded. It can be regarded as a quantity.

  The feature amount is corrected using a different contrast for each of the divided regions (R00, R01,... R99). Therefore, the correction of the feature amount in each region correctly reflects the contrast of the region, and the normalized feature amount can be obtained for the entire region of the linear sensor 22.

  Next, in step S9, a bright spot defect is detected by comparing a predetermined threshold value with the corrected feature amount. This process corresponds to the function of the detection means 33. As described above, since the feature value is standardized in the entire region of the linear sensor 22, the bright spot defect can be detected with high accuracy by using a common threshold value for the entire region.

  In step S10, the bright spot defect detection result is output, and the process ends.

  As described above, according to the defect inspection apparatus of the present embodiment, the uniformity of luminance defect detection conditions in the inspected area is ensured by correcting the feature amount. There is no need to change the threshold value. In addition, since the influence of individual differences among lenses is corrected, an operation for determining a threshold value for each lens is unnecessary. There is no need to change the threshold value in accordance with a change in the lens timing. Furthermore, since the influence of the lens focus adjustment is also canceled by the correction, there is no need to reset the threshold value every time focus adjustment is performed.

  Further, since the feature amount is standardized by the correction, even if the size of the sub-pixel changes, such as when the size of the liquid crystal panel changes, it is possible to use basically the same threshold value.

  In the above embodiment, an example using an area sensor in which photoelectric conversion elements are two-dimensionally arranged has been shown. However, the inspection sensor is not limited to an area sensor, and photoelectric conversion elements are arranged one-dimensionally (on a straight line). The present invention can also be applied to an inspection apparatus using a line sensor.

  In the above embodiment, the inspection of the matrix display type liquid crystal display device has been described. However, the present invention is widely applied to the inspection of the inspection object on which the regular pattern having the same spatial frequency as the defect is formed. For example, the present invention can be applied to inspection of a liquid crystal panel substrate on which TFT elements are arranged, a color filter for liquid crystal, a PDP (plasma display panel) and its substrate, a silicon wafer on which an IC pattern is formed, and the like.

  In the above embodiment, the feature amount is corrected by dividing the feature amount by the contrast of the image pattern of the corresponding region. However, the correction method is not limited. For example, the correction coefficient can be defined by a polynomial having contrast as a variable, and can be corrected by any calculation. For example, in the case where the contrast in the vertical and horizontal directions affects, appropriate correction may be possible by dividing by the square of the contrast.

  Further, in the above embodiment, the imaging state of the lens is specified and the feature amount is corrected using the contrast when the periodic pattern formed on the inspection target is photographed. Any lens can be used as long as it can detect the image formation state of the lens numerically, and is not limited to the contrast.

  For example, the spread of the histogram of the luminance of the image can be used for detecting the imaging state. As shown in FIG. 6A, a broad luminance distribution 61 is obtained when the imaging state is good, whereas a luminance distribution 62 concentrated on a specific luminance due to an abnormality in the imaging state is obtained. In this case, the imaging state can be determined based on the luminance distribution. For example, the brightness spread can be obtained as a standard deviation and used for correction.

  In the above-described embodiment, the contrast is obtained using the image used for the inspection of the bright spot defect. However, an image different from the image at the time of inspection may be used as the image for calculating the contrast.

  For example, as shown in FIG. 6B, when the brightness of the bright spot defect 63 is high, the exposure time for capturing the inspection image may be shortened. There is a possibility that the contrast cannot be calculated sufficiently using 64. In such a case, the contrast amount may be calculated using the signal 65 of another image with a longer exposure time to correct the feature amount of the inspection image. In this case, if two images (an image for contrast calculation and an inspection image) are captured while the focus state of the object to be inspected and the camera 2 is fixed, fluctuations in optical characteristics due to changes in the focus state are avoided. And high correction accuracy can be secured.

  In the above embodiment, the feature amount is corrected. However, instead of correcting the feature amount, a threshold for detecting a defect may be corrected.

  FIG. 7 is a block diagram showing the configuration of such a defect inspection apparatus.

  As illustrated in FIG. 7, the image processing apparatus 3 </ b> A includes a detection unit 36 that detects a defect in the inspected region based on a feature amount obtained via the camera 2, and a camera 2 based on an image captured by the camera 2. Based on the distribution of optical characteristics acquired by the characteristic distribution acquisition means 34 and the characteristic distribution acquisition means 34 for acquiring the distribution of optical characteristics in the inspected region, the threshold value of the feature amount for detecting defects in the detection means 36. And threshold correction means 35 for correcting.

  In the defect inspection apparatus shown in FIG. 7, instead of correcting the feature amount, the threshold value used in the detection unit 36 is corrected for each region of the linear sensor 22 to compensate for the optical characteristics of the camera 2. For example, the linear sensor 22 is divided into 100 regions, and a threshold value that matches the optical characteristics of each region is set, so that uniform bright spot defects can be detected in the entire region to be inspected. For this reason, substantially the same effect as the case of correcting the feature amount can be obtained.

  As described above, according to the defect inspection apparatus of the present invention, the feature amount obtained through the imaging optical device is corrected based on the optical characteristic distribution, or the defect is detected based on the optical characteristic distribution. Since the threshold value of the feature amount for correction is corrected, it is possible to eliminate the influence on the inspection of the optical characteristics of the apparatus without requiring complicated work.

  The scope of application of the present invention is not limited to the above embodiment. The present invention can be widely applied to a defect inspection apparatus and a defect inspection method for extracting a defect in a region to be inspected based on a feature amount obtained using an imaging optical device.

The block diagram which shows the structure of the defect inspection apparatus of one Embodiment. 6 is a flowchart showing the operation of the image processing apparatus. 4A and 4B are diagrams illustrating output signals of display pixels and area sensors, where FIG. 5A is a diagram illustrating an array of display pixels of a liquid crystal display device, and FIG. 5B is a diagram illustrating output signals (imaging signals) from an area sensor of a camera. The figure which illustrates a part. It is a figure which shows the optical characteristic of a lens, (a) is a figure which shows the positional relationship of a to-be-inspected area | region, a lens, and an area sensor, (b) is a figure which illustrates the resolution of a lens. It is a figure which shows the output signal etc. of an area sensor, (a) is a figure which illustrates the mode of the output signal of an area sensor, (b) is a figure which shows the example which divides | segments an area sensor into a some area | region. It is a figure which shows the luminance distribution of an area sensor, an output signal, etc., (a) is a figure which shows the luminance distribution obtained with a camera, (b) is a figure which shows an output signal when the brightness | luminance of a luminescent spot defect is high. The block diagram which shows the structure of the defect inspection apparatus which correct | amends the threshold value at the time of detecting a defect. The block diagram which shows the structure of the conventional defect inspection apparatus. The flowchart which shows the procedure of the conventional defect inspection.

31 characteristic distribution acquisition means 32 feature quantity correction means 33 detection means 34 characteristic distribution acquisition means 35 threshold correction means 36 detection means

Claims (5)

In the defect inspection apparatus for extracting defects of individual elements constituting the pattern in the inspection region having a regular pattern based on the feature amount obtained by using the imaging optical apparatus,
An imaging means for obtaining an inspection image by imaging the inspection area by the imaging optical device;
Feature quantity distribution acquisition means for acquiring a distribution of feature quantities in the imaging region of the imaging optical device based on the inspection image obtained by the imaging means;
Contrast distribution in the imaging region of the imaging optical device based on a change width of luminance between pixels based on the pattern that appears in the inspection image that is a target for acquiring a distribution of feature amounts by the feature amount distribution acquisition unit Characteristic distribution acquisition means for acquiring
Based on the contrast distribution acquired by the characteristic distribution acquisition unit, a feature amount correction unit that corrects the feature amount acquired by the feature amount distribution acquisition unit for each individual region in the imaging region;
Detecting means for detecting a defect in the inspected region based on the feature amount corrected by the feature amount correcting unit;
A defect inspection apparatus comprising: The defect inspection apparatus according to claim 1, wherein the imaging optical apparatus uses a two-dimensional area sensor as an imaging element. The defect inspection apparatus according to claim 1, wherein the imaging optical apparatus uses a one-dimensional line sensor as an imaging element. The defect inspection apparatus according to claim 1, wherein the inspection area is a screen of a display device. In the defect inspection method for extracting individual defects of the elements constituting the pattern in the inspection region having a regular pattern based on the feature amount obtained using the imaging optical device,
An imaging step of obtaining an inspection image by imaging the inspection area by the imaging optical device;
A feature amount distribution obtaining step for obtaining a feature amount distribution in a photographing region of the imaging optical device based on the inspection image obtained by the photographing step ;
Contrast distribution in the imaging region of the imaging optical device based on a change width of luminance between pixels based on the pattern that appears in the inspection image that is a target for acquiring the distribution of the feature amount in the feature amount distribution acquisition step A characteristic distribution acquisition step of acquiring
Based on the distribution of the contrast acquired by the characteristic distribution acquisition step, a feature amount correction step of correcting the feature amount acquired by the feature amount distribution acquisition step for each individual region in the imaging region;
A detection step of detecting a defect in the inspection region based on the feature amount corrected by the feature amount correction step;
A defect inspection method comprising:

Images