JP2009229197A - Linear defect detecting method and device - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a display defect evaluation method capable of accurately evaluating a linear defect of a display device. <P>SOLUTION: The linear defect detecting method includes: a taken image acquisition processing (step S1) for acquiring an taken image of a display device; an inspection area setting processing (step S2) for setting a to-be-inspected area on the taken image; a differential processing (step S3) for forming a differential image by removing a high frequency component and a linear defect component from the taken image and subtracting the differential image from the taken image; a smoothing processing (step S4) for removing the high frequency component form the differential processed image; a one-dimensional addition processing (step S5) for adding brightness data of each pixel of the smoothing processed image in the vertical direction and horizontal direction and forming profile data in each direction; a moving average processing (step S6) for performing moving average processing on each profile data; and a defect detection processing (step S7) for determining a differential value of data of each pair of neighboring pixels from the profile data after the moving average processing and detecting a linear defect by extracting a part with a large change by comparing the differential value and a threshold. <P>COPYRIGHT: (C)2010,JPO&INPIT

Description

  The present invention relates to a linear defect detection method and a linear defect detection apparatus for detecting a linear defect in a display device such as a liquid crystal panel and a display device such as a projector which is an application product thereof.

  A display device typified by a liquid crystal panel is driven mainly by TFTs. If a defect occurs in a part of this element, a line-shaped display defect of the minimum pixel is caused in the X direction or the Y direction. This is called a linear defect. When the entire screen is displayed in gray gradation, brighter or darker lines appear.

As a method for detecting such a linear defect, a method has been proposed in which a smoothing process is performed on a captured image to remove noise and integration is performed in the horizontal and vertical directions to detect the linear defect (patent). Reference 1).
Further, as a method for detecting a linear defect, a method of detecting a linear defect by applying a horizontal and vertical detection filter, dividing it into a plurality of rectangular areas, and integrating them in the area is proposed (Patent Document 2). reference).
Further, a method has been proposed in which a linear derivative waveform is obtained in order to detect a linear defect, and a linear defect is detected by counting peak positions (see Patent Document 3).

JP 2003-168103 A JP 2005-172559 A JP 2006-201523 A

However, the detection method of Patent Document 1 has a problem that when detecting a linear defect such as a liquid crystal panel, the influence of the black matrix is large, and even if a profile is created, the amplitude becomes large and it is difficult to detect a threshold value. is there.
In addition, in the detection method of Patent Document 2, when detecting a linear defect such as a liquid crystal panel, components other than the defect appear on the image. There is a problem such as.

Furthermore, the detection method of Patent Document 3 has a problem in that components such as a black matrix are counted, noise increases, and linear defects cannot be detected satisfactorily.
The above problems are not limited to detecting linear defects in a liquid crystal panel, but detecting linear defects in various display devices in which each pixel is divided by a black matrix such as a plasma display or an organic EL display. It was a common problem.

  An object of the present invention is to provide a display defect evaluation method and a display defect evaluation apparatus that can accurately evaluate a linear defect of a display device.

  The display defect evaluation method of the present invention is a linear defect detection method for detecting a linear defect of the display device based on a captured image acquired by capturing the display device, and images the captured image by capturing the display device. A picked-up image acquisition processing step of acquiring a picked-up image, an inspection region setting processing step of setting an inspection target region in the picked-up image, and creating a differential image by removing high-frequency components and linear defect components from the picked-up image A difference processing step for obtaining a difference processed image by subtracting the difference image from the image, a smoothing step for removing high frequency components from the difference processed image, and brightness of each pixel in the smoothed image A one-dimensional integration processing step of integrating the data in the vertical direction and the horizontal direction to create profile data in the vertical direction and the horizontal direction; From the moving average processing step that performs the moving average process to remove the noise component and emphasize the linear defect component, and the profile data subjected to the moving average process, the difference value of the data of each adjacent pixel is obtained, And a defect detection processing step of comparing the difference value with a preset threshold value to extract a portion having a large change in profile data and detecting it as a linear defect.

In the present invention, the display device to be inspected is imaged in the captured image acquisition processing step to acquire an image, and the non-inspection region is excluded from the captured image in the inspection region setting processing step so that only the image of the inspection target region is obtained. To do.
In the captured image acquisition processing step, for example, one pixel of a display device such as a liquid crystal panel is captured by six pixels of a CCD camera so as to obtain a sufficient resolution for detecting a linear defect. Then, it is preferable to set so that one pixel of the display device is displayed with 6 pixels. In general, in the present invention, a pixel of a display device to be inspected is referred to as a pixel, and a pixel of an imaging unit such as a CCD camera, that is, each pixel of a captured image is a pixel. Is written.

  In the difference processing step, since a difference processed image between the captured image and the difference image of the captured image is obtained, low-frequency components such as shading by illumination on the image and peripheral light reduction by the lens Are removed by differential processing. For example, when a low-pass filter such as a Gaussian filter of a predetermined size is applied, a high-frequency component typified by a black matrix arranged around each pixel of the display device and linear defects are removed, and a differential image of only the low-frequency component is obtained. Created. Therefore, when the difference image is subtracted from the captured image, the low-frequency component included in each image is removed, and the difference-processed image becomes an image having only high-frequency components such as linear defects and black matrix included in the captured image.

Next, in the smoothing processing step, for example, a low pass filter such as a Gaussian filter of a predetermined size is applied to the difference processed image, so that the difference processed image is converted into a black matrix without erasing the linear defect. The representative high frequency component is removed.
Then, in the one-dimensional integration processing step, when the luminance data of each pixel is integrated in the vertical direction and the horizontal direction, the luminance values of the linear defect portions along the vertical direction and the horizontal direction are integrated. Since it differs greatly from the integrated value of the non-existing portion, the linear defect portion can be emphasized. The vertical direction and the horizontal direction are directions along data lines and scanning lines of the TFT circuit, for example, when each pixel of the display device is driven by the TFT circuit.
Furthermore, in the present invention, the noise component can be removed by performing the moving average processing step, and the difference value, that is, the amount of change of the data of each adjacent pixel is obtained in the defect detection processing step, and the portion where the amount of change is larger than the threshold If there is, it can be detected that a linear defect exists.

  Therefore, according to the present invention, in particular, by including the difference processing step and the smoothing processing step, the influence of low-frequency components such as shading due to illumination and peripheral dimming by a lens, and high-frequency components represented by a black matrix. And a linear defect to be inspected can be extracted. Further, since the linear defect can be emphasized by the one-dimensional integration processing step, even a light linear defect can be emphasized. Further, by the moving average processing step and the defect detection processing step, it is possible to reliably detect the linear defect portion by further removing noise and comparing the amount of change of each pixel data with a threshold value. For this reason, according to this invention, a linear defect can be evaluated accurately.

In the present invention, the difference processing step creates a difference image by applying a Gaussian filter having a size larger than a pixel size of a display device in the captured image to an image obtained by duplicating the captured image. The difference image is preferably obtained by subtracting the difference image from the image and adding a preset setting value.
Here, the size of the Gaussian filter is larger than one pixel of the display device, for example, when one pixel of the display device is displayed with 6 pixels of the captured image, the vertical and horizontal sizes are 31 pixels. Any size is acceptable. The size of the Gaussian filter may be set to a size that can remove a high-frequency component typified by a black matrix of a display device and a linear defect component.
Further, the set value may be a value that can be converted to a positive value when the difference result becomes a negative value in the difference processing step. For example, the gradation of the captured image when the inspection target is imaged What is necessary is just to set to the intermediate value of a value. That is, when the gradation value of the captured image is represented by 12 bits (4096 gradations), the setting value may be set to 2049, which is an almost intermediate value.

In the present invention, by applying a large-size Gaussian filter to the replicated captured image, high-frequency components and linear defect components typified by the black matrix are removed, and shading by illumination and peripheral reduction by lenses are reduced. A difference image of only low frequency components such as light can be obtained. Therefore, if this difference image is subtracted from the captured image, the low frequency component included in each image is deleted, and a difference processed image including the high frequency component and the linear defect component can be obtained.
Further, since the difference processing image adds a setting value to the difference processing, when the difference image is subtracted from the captured image, that is, the data of each pixel of the difference image is obtained from the data (luminance value) of each pixel of the captured image. Even if there is a negative pixel when subtracting, all the pixel values can be set to a positive value by adding a setting value (for example, a gradation intermediate value) to each pixel of the difference result. The subsequent image processing can be easily performed.

In the present invention, it is preferable that the smoothing process step creates a smoothed image by applying a Gaussian filter having a size smaller than the pixel size of the display device in the captured image to the difference processed image.
Here, the size of the Gaussian filter may be smaller than the size of the pixel of the display device, as long as it can remove a high-frequency component typified by a black matrix and does not remove a linear defect component. Good. For example, when one pixel of the display device is displayed with 6 pixels of the captured image, a filter having a size of 3 × 1 pixel, which is about half that size, may be applied.

  In the present invention, since a Gaussian filter having a small size is applied to the difference processed image, a high frequency component typified by a black matrix can be removed from the difference processed image.

  In the present invention, the defect detection processing step obtains a difference value between data of adjacent pixels, performs a moving average process on the absolute value of the difference value, compares the processing result with the threshold value, and generates profile data. It is preferable to extract a portion having a large change in and detect it as a linear defect.

In the present invention, the amount of change can be calculated by obtaining the difference value between the data of adjacent pixels.
Further, since this change amount may be a positive value and a negative value, when comparing the threshold values as they are, it is necessary to provide positive and negative threshold values, respectively. On the other hand, in the present invention, since the absolute value of the difference value is compared with the threshold value, a single threshold value may be set.
Further, since the moving average processing of the absolute value of the difference value is performed, the noise can be removed so that only the linear defect portion has a large amount of change, and the linear defect can be detected with high accuracy.

In the present invention, it is preferable that the inspection region setting processing step detects a change in luminance at an outer peripheral portion of the captured image, sets the inspection region by determining a position where the luminance has changed as a boundary between the inspection region and the region to be inspected. .
In the present invention, since the inspection area can be set only by detecting the outer peripheral luminance change position of the captured image, the inspection area can be easily detected.

  The linear defect detection apparatus of the present invention is a linear defect detection apparatus that detects a linear defect of the display device based on a captured image acquired by imaging the display device, and images and images the display device. A captured image acquisition unit that acquires an image, an inspection region setting unit that sets an inspection target region in the captured image, and a differential image in which a high-frequency component and a linear defect component are removed from the captured image, and the captured image Difference processing means for obtaining a difference processed image by subtracting the difference image, smoothing processing means for removing high frequency components from the difference processed image, and luminance data of each pixel in the smoothed image Are integrated in the vertical and horizontal directions, respectively, and one-dimensional integration processing means for creating vertical and horizontal profile data, and for each profile data The difference value of the data of each adjacent pixel is obtained from the moving average processing means that performs the moving average process to remove the noise component and emphasize the linear defect component, and the profile data subjected to the moving average process. And defect detection processing means for extracting a portion having a large change in the profile data and detecting it as a linear defect.

  According to the present invention, the same effects as the linear defect detection method can be obtained. Moreover, in the linear defect detection apparatus of this invention, you may limit the structure of each said invention of Claim 2-5, In this case, the effect similar to each said invention is acquired.

Hereinafter, embodiments of the present invention will be described with reference to the drawings.
[Configuration of linear defect detection device for display panel]
As shown in FIG. 1, the linear defect detection apparatus 1 of this embodiment evaluates the linear defect of the inspection panel 2 to be inspected. The inspection panel 2 of the present embodiment is a liquid crystal panel, but is a liquid crystal panel at the component stage before cutting and assembly.
Further, the line defect of the inspection panel 2 is a line-shaped defect portion having a transmittance different from that of surrounding pixels in the inspection panel 2. For example, in a display device such as a liquid crystal panel driven by a TFT circuit, this linear defect is caused in the X direction or Y direction (scanning line direction or data line direction) when a defect occurs in a part of the TFT circuit element. This occurs when a line-shaped display failure occurs. Therefore, when the entire screen is displayed in gray gradation, bright lines or dark lines appear compared to surrounding pixels, and these are called linear defects.
The linear defect detection apparatus 1 according to the present embodiment emphasizes and detects a linear defect portion in the inspection panel 2 if there is a linear defect portion.

  The linear defect detection apparatus 1 images an XY stage 3 movable in the XY direction, a backlight 4 placed on the XY stage 3, and an inspection panel 2 placed on the backlight 4. The apparatus includes a CCD camera 5, a driving device 6 that drives the inspection panel 2, a processing device 7 that is connected to the CCD camera 5 and the driving device 6, and a display device 8.

The XY stage 3 can move in the X direction shown in FIG. 1 and the Y direction orthogonal to the X direction in a plane along the upper surface of the XY stage 3.
The backlight 4 is used for displaying an inspection image on the inspection panel 2 which is a liquid crystal panel. That is, in the inspection panel 2, the transmittance of each pixel is controlled by the driving device 6, and thereby the amount of light transmitted from the backlight 4 is controlled, and an image displayed on the inspection panel 2 is displayed in various gradations. (Brightness) can be controlled.

The CCD camera 5 is supported by a Z stage (not shown) that can move in the Z direction orthogonal to the XY direction, and is configured to be able to adjust the focus of the CCD camera 5.
Accordingly, the positions of the inspection panel 2 and the CCD camera 5 can be relatively moved in the XYZ directions.
As this CCD camera 5, a color camera may be used. However, in this embodiment, since a change in brightness of the inspection panel 2 is only required to be photographed, a monochrome CCD camera is used.
The driving device 6 is controlled by the processing device 7 and performs a process of drawing a pattern for detecting a linear defect on the inspection panel 2.

The processing device 7 acquires and stores digital image data captured by the CCD camera 5 and performs a linear defect detection process described below.
Therefore, as shown in FIG. 2, the processing device 7 includes a captured image acquisition unit 71, an inspection region setting unit 72, a difference processing unit 73, a smoothing processing unit 74, a one-dimensional integration processing unit 75, and a moving average processing unit. 76, defect detection processing means 77, and defect determination means 78 are provided.

The picked-up image acquisition means 71 performs processing for acquiring a picked-up image by picking up an image of the inspection panel 2 made of a liquid crystal panel formed on a glass substrate with the CCD camera 5.
The inspection area setting means 72 performs processing for detecting and excluding the edge of the panel from the captured image, and setting the inspection area by excluding the non-inspection area.
The difference processing means 73 creates a difference image and performs difference processing between the captured image and the difference image.
The smoothing processing means 74 performs smoothing processing that smoothes the image after the difference processing and suppresses the influence of the black matrix.
The one-dimensional integration processing means 75 performs a one-dimensional integration process for integrating the pixel data in the vertical and horizontal directions and creating profile information.
The moving average processing means 76 performs moving average processing for emphasizing only defect components from the profile information.
The defect detection processing unit 77 performs a linear defect detection process for extracting a linear defect using a threshold value at a place where the data change between the pixels is large.
The defect determination means 78 determines whether or not the linear defect detection process has been performed on the entire surface of the glass substrate. If all the defect determination processes have been performed, a defect determination process for displaying the detection result on the display device 8 is performed.

[Linear defect detection method for inspection panel]
In the following, details of the processing in each means, that is, the “linear defect detection method of the inspection panel 2” will be described with reference to the flowchart of FIG.
First, as a preliminary preparation, since the backlight 4 is placed on the XY stage 3, a liquid crystal panel (inspection panel 2) before cutting and assembling to be inspected is set thereon. Above this, a CCD camera 5 for obtaining an image of the inspection panel 2 is installed.

And the outline of the linear defect detection method in this embodiment is as follows.
That is, after drawing a pattern for detecting a linear defect on the inspection panel 2 using the driving device 6, an image of the inspection panel 2 obtained by transmission illumination of the backlight 4 is passed through the lens of the CCD camera 5. The image is formed on the image sensor and stored in the processing device 7 as a digital image.
Then, the processing device 7 performs processing for detecting a linear defect on the stored image. Since the lens of the CCD camera 5 used at this time has a sufficient resolution for detecting a linear defect, an image obtained by the CCD camera 5 becomes a part of the inspection panel 2. In order to inspect the entire surface of the inspection panel 2, an image of the inspection panel 2 is acquired while moving the XY stage 3, and detection processing is performed. After the detection processing is completed, the processing result is output to the display device 8 attached to the processing device 7, and the detection result is confirmed.

The above linear defect detection method will be described in detail with reference to the flowchart shown in FIG.
[1. Captured image acquisition processing]
When the linear defect detection process is started, the captured image acquisition unit 71 of the processing device 7 moves the XY stage 3 to an appropriate position and controls the driving of the inspection panel 2 via the driving device 6. At the same time, a captured image acquisition process is performed in which the inspection panel 2 placed on the XY stage 3 is imaged using the CCD camera 5 and the captured image is stored in the memory of the processing device 7 (S1).
At this time, the brightness of the illumination and the set value of the CCD camera 5 are adjusted in advance. As a specific adjustment condition, the average luminance level in the image plane imaged by the CCD camera 5 is set to approximately the median value (for example, “2048” when the luminance value can be stored in 12 bits, that is, 4096 gradations). An example of the captured image 20 is shown in FIG. In the image 20 of FIG. 4, a linear defect 21 extending in the horizontal direction (X direction, scanning line direction) along the pixel and a linear defect 22 extending in the vertical direction (Y direction, data line direction) are displayed. ing.

[2. Inspection area setting process]
When the captured image acquisition processing step S1 is completed, the inspection region setting unit 72 of the processing device 7 determines whether or not the stored captured image 20 includes a region that is not the inspection target, and is outside the inspection target. If there is such a region, the non-target region is deleted from the image, and the inspection region setting process is performed to make the image only the inspection target region (S2).
Here, the non-target region is specifically a panel end (cut portion) or the like. Therefore, the inspection area setting means 72 scans one pixel in the vertical and horizontal directions at the end of the captured image 20 as shown by the arrows in FIG. As shown in FIG. 5, the display area of the inspection panel 2 is normally controlled so that the luminance value becomes approximately the median value, so that the display area 25 portion in the captured image 20 is displayed in gray. On the other hand, the end region of the inspection panel 2 is a black region 26 in the captured image 20, and the region where no pattern is formed (transparent glass region) is a white region 27 in the captured image 20. Therefore, the edge, that is, the boundary position between the display area 25 and the area 26 is searched from these luminance differences, and the presence or absence of the non-inspection area is confirmed.
If the inspection area setting unit 72 determines that there is a non-inspection area, the inspection area setting unit 72 performs processing to delete the non-target areas (areas 26 and 27 in FIG. 5) from the captured image. On the other hand, if the inspection area setting unit 72 determines that there is no non-inspection area, it does not perform the process of deleting the non-target area.

[3. Difference processing]
When the inspection area setting process step S2 ends, the difference processing means 73 executes a difference process for creating a difference image using the image from which the non-target area is deleted (S3).
That is, the difference processing means 73 duplicates the captured image stored in the memory and performs an averaging process on the captured image. Then, an image obtained by the averaging process is set as a difference image. This averaging process is a process for removing the influence of low-frequency components such as shading due to illumination on the image and peripheral dimming by the lens.

Hereinafter, this difference process will be described more specifically.
First, the difference processing means 73 creates a difference image using the image captured in the captured image acquisition processing step S1. That is, the difference processing means 73 performs an averaging process by applying a Gaussian filter having a larger size to the size of one pixel of the inspection panel 2 in the captured image. In the present embodiment, one pixel of the inspection panel 2 has a size of about 6 pixels in the captured image, whereas the averaging process is performed using a Gaussian filter having a size of 31 × 31 pixels. Thereby, a high frequency component and a linear defect component as represented by a black matrix are removed, and an image of only a low frequency component is created. Let this image be a difference image.

Next, the difference processing means 73 performs difference processing between the “captured image” captured in the captured image acquisition processing step S1 and the “difference image” created above. Assuming that the image thus obtained is a “difference processed image”, this difference processing can be expressed by the following Equation 1.
Difference processed image = captured image−difference image + intermediate value …… Equation 1
Here, the intermediate value is set to “2049” which is almost in the middle when the gradation of the captured image is 12 bits (4096 gradation).
The intermediate value is added in equation 1 because the intermediate value is added even if the luminance value of each pixel of the difference image is subtracted from the luminance value of each pixel of the captured image. This is because a positive value can be obtained. As a result, the luminance value of each pixel of the processed image becomes a value with reference to the intermediate gradation (gray), and it is easy to confirm the variation of each luminance.

[4. Smoothing process]
When the difference processing step S3 ends, the smoothing processing means 74 executes a smoothing process for removing high-frequency components on the difference processing image created in the difference processing step S3 (S4).
Specifically, the smoothing processing unit 74 applies a Gaussian filter to the difference processed image. This filter size is set to a size larger than the size of one pixel of the inspection panel 2 in the captured image, and is, for example, about twice as large as the pixel size. By applying this Gaussian filter, it is possible to suppress the influence of high-frequency components typified by a black matrix.
Further, the averaging process in the vertical direction (Y direction) is performed to further reduce the influence of high frequency components. By performing this averaging process, it is possible to suppress amplitude fluctuation during the next one-dimensional integration process. The size of the filter that performs the averaging process may be about half that of one pixel of the liquid crystal. For example, in the present embodiment, as described above, since one pixel of the inspection panel 2 is about 6 pixels on the captured image, the smoothing processing unit 74 applies a filter having a size of 3 × 1 pixel in the vertical direction. An averaging process is performed.
By such a smoothing process step S4, it is possible to remove only high-frequency components without erasing the linear defects, and to further increase the stability of the linear defect luminance in the integrating direction.

[5.1-dimensional integration processing]
When the smoothing processing step S4 is completed, the one-dimensional integration processing means 75 integrates the processed image 20A subjected to the smoothing processing pixel by pixel in the vertical direction (Y direction) as shown in FIG. Then, a one-dimensional integration process for storing the result in a one-dimensional array is executed (S5).
Although not shown, the one-dimensional integration processing means 75 also executes a one-dimensional integration process that integrates one pixel at a time in the horizontal direction (X direction) and stores the result in a one-dimensional array (S5).

[6. Moving average processing]
When the one-dimensional integration processing step S5 is completed, the moving average processing means 76 performs processing on each profile information (vertical profile information and horizontal profile information) as a result of the one-dimensional integration processing, and extracts defects. A moving average process is performed so that it is easy (S6).
Specifically, the moving average processing means 76 performs moving average to remove noise components. In the present embodiment, an average value of a total of five pixels, that is, a pixel of interest and two pixels before and after the pixel of interest, is obtained as the processing value of the pixel of interest, and the average value is obtained while moving the pixel of interest for each pixel. It is carried out. That is, first, an average of the data values of the five pixels at the pixel positions “1 to 5” is obtained to obtain a processing value for the pixel position “3” as the pixel of interest, and then the five pixels at the pixel positions “2 to 6”. Is the processing value of the pixel position “4” that is the pixel of interest. The moving average processing means 76 performs such moving average processing while sequentially moving the pixel of interest, and stores the result in the memory.

FIG. 7 shows profile information as a result of performing the moving average process, that is, the one-dimensional integration process step S5. Further, FIG. 8 shows profile information after the processing in which the moving average processing step S6 is performed. As shown in FIGS. 7 and 8, a fine noise component is removed by performing the moving average processing step S6.
The moving average processing means 76 performs this moving average processing step S6 on the profile information in the vertical direction and the profile information in the horizontal direction, respectively.

[7. Defect extraction process]
When the moving average processing step S6 ends, the defect detection processing unit 77 extracts a defect by examining the amount of change from the adjacent pixel position with respect to the profile information smoothed in the moving average processing step S6. Extraction processing is performed (S7).
The amount of change usually increases or decreases around zero. For this reason, when a change having a large change amount is extracted by comparing the change amount with a threshold value, it is usually necessary to provide a minus threshold value in addition to the plus threshold value.
On the other hand, in this embodiment, since defect determination is performed using a single threshold, the defect detection processing unit 77 obtains the absolute value of the amount of change. Next, the defect detection processing unit 77 obtains a moving average with respect to the change amount of the absolute value, and determines that the defect is not a defect if the moving average is equal to or greater than a predetermined threshold, and is not a defect if the moving average is less than the threshold.
FIG. 9 shows profile information and threshold values in the defect detection processing step S7. The threshold is set to about 0.3.
The defect detection processing means 77 also uses the vertical profile information obtained by performing the moving average processing on the vertical one-dimensional integration processing result and the moving average on the horizontal one-dimensional integration processing result in the defect detection processing step S7. The processing is executed for each of the processed profile information in the horizontal direction, and a linear defect along the vertical direction and a linear defect along the horizontal direction are detected.

[8. Full-screen end determination process]
When the defect detection processing step S7 is completed, the defect determination means 78 of the processing device 7 determines whether the entire screen of the inspection panel 2 has been processed (S8). That is, as described above, since the image obtained by the CCD camera 5 in the captured image acquisition processing step S1 is a part of the inspection panel 2, the entire surface of the inspection panel 2 is imaged and the processes of S1 to S7 are performed. Requires a plurality of processes. Therefore, if it is determined in S8 that the full screen processing is not completed, the processing device 7 returns to the captured image acquisition processing step S1 and continues the processing. In this case, the captured image acquisition means 71 moves the XY stage 3 and images a portion that has not yet been captured on the inspection panel 2 with the CCD camera 5 and continues the processing.

[9. Comprehensive judgment processing]
In the full screen end determination processing step S <b> 8, when it is determined that the processes of S <b> 1 to S <b> 7 have been completed for the entire screen of the inspection panel 2, the defect determination unit 78 captures each of the inspection panels 2. Based on the processing result of the defect detection processing step S7 for the image, a processing result image 30 as shown in FIG. 10 is displayed on the display device 8 to perform a comprehensive determination process (S9).
In the processing result image 30 of FIG. 10, the outermost periphery indicates the outer periphery of the glass substrate of the inspection panel 2. Each of the cells is one screen imaged by the CCD camera 5. A region 31 displayed in gray in the processing result image 30 corresponds to the liquid crystal surface on which the liquid crystal is provided. In this way, the liquid crystal surfaces are installed on a single glass substrate so that the plurality of liquid crystal surfaces are taken together, and the defect determination means 78 combines the results in order to determine whether or not there is a defect in each liquid crystal surface. Is displayed on the display device 8. Therefore, if the inspector confirms the processing result image 30 displayed on the display device 8, it can easily grasp which liquid crystal surface has a defect.
Note that FIG. 10 shows only vertical line defects. However, when a horizontal line defect is also detected, the horizontal line defect is also included in the region 31 of FIG. Show it.

According to the above-described embodiment, the following effects can be achieved.
(1) Since the difference processing unit 73 and the smoothing processing unit 74 perform the difference processing step S3 and the smoothing processing step S4, low-frequency components such as shading by illumination and peripheral light reduction by a lens, and a black matrix The influence of the high frequency component represented by can be suppressed. For this reason, it is possible to remove only the noise information without erasing the information of the actual linear defect portion, and it is possible to detect the linear defect with high accuracy.

(2) Since the one-dimensional integration processing step S5 is performed by the one-dimensional integration processing means 75, even if it is a light linear defect, that is, a linear defect with a small luminance difference with respect to the surroundings, the luminance value of the pixel in the defective portion Can be increased, and the difference from the integrated value of the normal portion can be increased to enhance the linear defect. For this reason, even a light linear defect can be detected accurately and automatically.

(3) Since the moving average processing step S6 by the moving average processing means 76 is performed, the noise component included in the integration result of the one-dimensional integration processing step S5 can be removed, and a linear defect is erroneously affected by the noise. Detection can be prevented and detection can be performed with high accuracy.

(4) In the defect detection processing step S7 by the defect detection processing means 77, since the amount of change between adjacent pixels is obtained, it is possible to emphasize a portion with a large change in luminance value, that is, a linear defect portion with high accuracy. It can be detected.
In addition, since the defect detection processing unit 77 compares the change amount with the threshold value after making it an absolute value, it is only necessary to set a single threshold value, and it is not necessary to set both positive and negative threshold values for comparison. The linear defect portion can be easily detected.

(5) As shown in FIG. 10, since the defect determination means 78 shows the linear defect portion in the planar state of the inspection panel 2, the inspector can easily grasp the occurrence state of the linear defect.

[Modification]
Although the best configuration for carrying out the present invention has been disclosed in the above description, the present invention is not limited to this. That is, since the embodiment does not limit the present invention, description of the names of members excluding some or all of the limitations on the shape, material, etc. is included in the present invention. is there.

For example, although the Gaussian filter is applied as the low-pass filter in the difference processing step S3 and the smoothing processing step S4, other low-pass filters may be used as long as the same function can be obtained.
Moreover, the linear defect detection apparatus 1 of this invention can be widely utilized not only for the display inspection of a liquid crystal panel but for the linear defect inspection of various display devices.
For example, in the embodiment, a transmissive liquid crystal panel is inspected, but a reflective panel such as LCOS (Liquid Crystal on Silicon) may be inspected.
Further, the present invention can be used for inspection of various display panels such as a plasma display, an organic EL display, and a DMD (direct mirror device) as an inspection object in the present invention. In particular, the present invention can be widely used when detecting a linear defect by removing the influence of a high-frequency component such as a black matrix in an inspection object in which an imaging pixel is included.

It is a figure which shows the structure of the linear defect detection apparatus which concerns on embodiment of this invention. It is a block diagram which shows the structure of the processing apparatus of the said embodiment. It is a flowchart which shows the linear defect detection method of the said embodiment. It is a figure which shows the captured image containing a linear defect. It is a figure which shows the captured image containing a panel edge part. It is explanatory drawing explaining a one-dimensional integration process. It is a graph which shows the profile information which shows a one-dimensional integration process result. It is a graph which shows the profile information which shows a moving average process result. It is a graph which shows the profile information which shows a defect detection process. It is a figure which shows the process result image of a comprehensive determination process.

Explanation of symbols

  DESCRIPTION OF SYMBOLS 1 ... Linear defect detection apparatus, 2 ... Inspection panel, 3 ... XY stage, 4 ... Back light, 5 ... CCD camera, 6 ... Drive apparatus, 7 ... Processing apparatus, 8 ... Display apparatus, 71 ... Captured image acquisition means 72 ... inspection area setting means, 73 ... difference processing means, 74 ... smoothing processing means, 75 ... one-dimensional integration processing means, 76 ... moving average processing means, 77 ... defect detection processing means, 78 ... defect determination means.

Claims (6)

A linear defect detection method for detecting a linear defect of the display device based on a captured image acquired by imaging a display device,
A captured image acquisition process for capturing the captured image by imaging the display device;
An inspection area setting process for setting an inspection object area in the captured image;
A difference processing step of creating a difference image from which the high-frequency component and the linear defect component have been removed from the captured image, and obtaining a difference processed image by subtracting the difference image from the captured image;
A smoothing process for removing high frequency components from the difference processed image;
A one-dimensional integration process step of integrating the luminance data of each pixel in the smoothed image in the vertical direction and the horizontal direction to create profile data in the vertical direction and the horizontal direction;
A moving average processing step of performing a moving average process on each profile data to remove a noise component and emphasize a linear defect component;
A linear defect is obtained by obtaining a difference value between adjacent pixel data from profile data that has been subjected to moving average processing, comparing this difference value with a preset threshold value, and extracting a portion where the change in profile data is large. Defect detection processing step to detect as,
A linear defect detection method characterized by comprising: The linear defect detection method according to claim 1,
The difference processing step creates a difference image by applying a Gaussian filter having a size larger than the size of a pixel of the display device in the captured image to an image obtained by duplicating the captured image, and generates the difference from the captured image. A linear defect detection method characterized in that a difference processed image is obtained by subtracting images and adding a preset setting value. In the linear defect detection method according to claim 1 or 2,
The smoothing processing step creates a smoothed image by applying a Gaussian filter having a size smaller than the pixel size of the display device in the captured image to the difference processed image. Detection method. In the linear defect detection method according to any one of claims 1 to 3,
In the defect detection processing step, a difference value between data of adjacent pixels is obtained, a moving average process is performed on the absolute value of the difference value, and the change in profile data is large by comparing the processing result with the threshold value. A linear defect detection method, characterized in that a portion is extracted and detected as a linear defect. In the linear defect detection method according to any one of claims 1 to 4,
The inspection area setting processing step detects a change in luminance at the outer periphery of the captured image, determines the position where the luminance has changed as a boundary between the inspection area and the inspection area, and sets the inspection area. Defect detection method. A linear defect detection apparatus for detecting a linear defect of the display device based on a captured image acquired by imaging a display device,
Captured image acquisition means for capturing the captured image by capturing the display device;
Inspection area setting means for setting an inspection object area in the captured image;
A difference processing unit that creates a difference image from which the high-frequency component and the linear defect component have been removed from the captured image, and calculates a difference processed image by subtracting the difference image from the captured image;
Smoothing processing means for removing high frequency components from the difference processed image;
One-dimensional integration processing means for integrating the luminance data of each pixel in the smoothed image in the vertical direction and the horizontal direction to create profile data in the vertical direction and the horizontal direction;
Moving average processing means for performing a moving average process on each profile data to remove a noise component and emphasize a linear defect component;
A linear defect is obtained by obtaining a difference value between adjacent pixel data from profile data that has been subjected to moving average processing, comparing this difference value with a preset threshold value, and extracting a portion where the change in profile data is large. Defect detection processing means for detecting as,
A linear defect detection apparatus comprising: