JP2006133196A - Detection method of pixel uneveness defection, detection device for pixel uneveness defection, detection program for pixel uneveness defection, and recording medium with the program stored - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a detection method of pixel uneveness defects, capable of accurately detecting the pixel uneveness defects existing in images. <P>SOLUTION: The detection technique of pixel uneveness defects detects pixel uneveness defects (a detection process ST300 for uneven pixels) from images acquired in an inspection image acquiring process ST100, and by a quantitative evaluation, based on the detected result in the detecting process of pixel uneveness, discrimination of pixel uneveness defects is conducted (a discrimination process ST400 of defects) in the images. The detection process ST300 of pixel uneveness is equipped with a domain-dividing process for dividing the image data to a plurality of regions of a predetermined area, a process (a detection process of extra pixels) for detecting extra pixels with brightness higher or lower than a predetermined threshold as compared with the surroundings in each region, a process (a counting process of extra pixels) for counting the number of detected extra pixels for each region, and a process (an extraction process of condidate pixel uneveness) for extracting the regions related to the pixel uneveness, based on the number of extra pixels counted by each region. <P>COPYRIGHT: (C)2006,JPO&NCIPI

Description

  The present invention relates to a pixel unevenness defect detection method, a pixel unevenness defect detection device, a pixel unevenness detection program, and a recording medium on which the program is recorded. For example, a display device such as a liquid crystal panel or a light valve of a projector that is an application product thereof. The present invention relates to a method for detecting a pixel non-uniformity defect that is a display defect in the inspection.

When manufacturing a display device such as a liquid crystal panel and a light valve of a projector which is an application product thereof, an inspection for display defects is performed. Display defects include point defects, line defects, surface defects, and the like, and a method for detecting these display defects has been proposed (Patent Document 1).
This Patent Document 1 discloses a method of binarizing an image using a threshold of three stages in order to separately extract point defects, line defects, and surface defects.

Japanese Patent Laid-Open No. 9-288037

In addition to point defects, line defects, and surface defects, pixel defects include pixel unevenness defects.
The pixel unevenness defect exists over a range where there is a luminance change in units of display elements due to defective polarization of the element.For example, when trying to display the entire screen with the same luminance, the luminance is too high compared to the surroundings. This is a phenomenon in which there are regions where pixels that are too low exist.
If there is such a pixel unevenness defect, there arises a problem that the image has a feeling of roughness or roughness and the image quality is lowered.
However, the pixel unevenness defect which is a defect of the display element unit is different from any of the point defect, the line defect, and the surface defect, and can be detected by the method of simply binarizing the image as disclosed in Patent Document 1. There wasn't. For this reason, the determination of the presence or absence of pixel unevenness defects depends on qualitative judgment by the human eye, and there are problems such as different judgments by people, and it is difficult to quickly process a large number of products. There was a problem.
Therefore, a method for detecting a pixel unevenness defect has been desired.

  An object of the present invention is to provide a pixel unevenness defect detection method, a pixel unevenness defect detection apparatus, a pixel unevenness defect detection program, and a recording medium on which the program is recorded, which accurately detect pixel unevenness defects present in a display image. .

  According to the pixel unevenness defect detection method of the present invention, an image displayed when an image displayed by a display device exists in a certain area in a certain area having pixels with higher or lower luminance than surrounding pixels. A pixel unevenness defect detection method for detecting a defective pixel unevenness defect, based on an inspection image acquisition step of acquiring an image from a display device to be inspected, and image data acquired in the inspection image acquisition step A pixel unevenness detecting step for detecting pixel unevenness in the image data; and a defect determining step for determining a defect of the display device based on a detection result of the pixel unevenness detecting step, wherein the pixel unevenness detecting step includes: A region dividing step for dividing the image data acquired in the inspection image acquiring step into a plurality of regions, and brightness higher than surrounding pixels for each region. Alternatively, an abnormal pixel detection step of detecting an abnormal pixel that is low and has a luminance difference with respect to a reference value set by surrounding pixels that is equal to or greater than a predetermined threshold, and the number of abnormal pixels detected in the abnormal pixel detection step for each region An abnormal pixel counting step of counting, and a pixel unevenness candidate extraction step of extracting candidate regions that become pixel unevenness based on the number of abnormal pixels for each of the regions counted in the abnormal pixel counting step. The step is characterized in that a defect of the display device is discriminated based on a quantitative evaluation of the region extracted in the pixel unevenness candidate extraction step.

In such a configuration, an image from a display device is acquired by taking an image (inspection image acquisition step), and the acquired image data is divided into a plurality of regions having a predetermined area (region division step).
In each region generated by the division, abnormal pixels whose luminance is higher or lower than a predetermined threshold compared to the surroundings are detected (abnormal pixel detection process), and the number of detected abnormal pixels is counted for each region. (Abnormal pixel counting process).
Then, a region connected to pixel unevenness is extracted based on the number of abnormal pixels counted for each region (pixel unevenness candidate extraction step).
By performing evaluation based on the quantitative value of the region extracted as a candidate that leads to pixel unevenness in this way, for example, the image by the display device has a pixel uneven defect and the display device is defective. Defect determination is performed (defect determination step).

The pixel unevenness defect is caused by a certain number or more of pixels having poor luminance display within a certain area, and cannot be detected only by processing such as simply binarizing the entire image as in the prior art. However, according to the present invention having the above-described configuration, pixels having a luminance value different from that of the periphery due to the luminance value being higher or lower than that of the surrounding pixels are detected. Pixels that lead to uneven defects can be picked up. Here, a slight difference in luminance compared to the surrounding area does not necessarily lead to pixel unevenness defects. However, a predetermined threshold is provided, and an abnormal pixel detected in the abnormal pixel detection step has a luminance difference of more than the predetermined threshold. Only certain abnormal pixels are detected. As a result, it is possible to detect an abnormal pixel having a luminance difference that is perceived as roughness of the image and causing a pixel unevenness defect to deteriorate the image.
Then, by detecting such abnormal pixels for each region and extracting a region in which the number of abnormal pixels is equal to or greater than a certain value, a region that leads to a pixel unevenness defect can be extracted.
In addition, since the regions that are pixel unevenness candidates can be extracted in this way, the extracted areas can be quantitatively analyzed such as the total area of the extracted regions, the total number of abnormal pixels in the extracted regions, and the average luminance value of the abnormal pixels. The pixel unevenness defect can be automatically (based on arithmetic processing) based on the correct value.
Then, the presence / absence of the pixel unevenness defect can be quickly and accurately determined, and the quality of the display device product can be determined.

Here, in the abnormal pixel detection step, in detecting a pixel having a higher or lower luminance than the surrounding pixels, the luminance is higher than any of these eight pixels compared to the eight pixels surrounding the target pixel. Alternatively, a low pixel may be detected, or a pixel whose luminance is higher or lower than any of these 24 pixels compared to the 24 pixels surrounding the pixel of interest may be detected. It is arbitrary how large the is taken.
In addition to detecting pixels with higher or lower luminance than any of the surrounding pixels, several high-luminance pixels (for example, five) are detected from the upper level in a certain range, and lower luminance pixels are lower-level. Some (for example, 5) may be detected.
Further, in the abnormal pixel detection process, when a pixel (abnormal pixel) whose luminance difference with respect to a reference value set by surrounding pixels is equal to or greater than a predetermined threshold is detected, one reference value is skipped from the pixel of interest. The average luminance of the 16 pixels surrounding the pixel of interest may be the average luminance, or the average luminance of 24 pixels surrounding the pixel of interest may be set, so that the surrounding pixels surrounding the pixel of interest are visually rough. An appropriate reference value may be set to detect that the luminance of the pixel of interest is separated. Also, the threshold value is not particularly limited.
As described above, in the abnormal pixel detection process, specific peripheral ranges, reference values, threshold values, and the like for detecting abnormal pixels can be set for various purposes such as the size, application, and quality standard of the display device to be inspected. It can be changed accordingly.

  In the present invention, the abnormal pixel detection step detects an extreme pixel that detects a maximum pixel that is a pixel having higher luminance than a surrounding region and a minimum pixel that is a pixel having low luminance for each region divided in the region division step. An extreme pixel detection process that performs detection filter processing, and a classification that classifies abnormal pixels that lead to pixel unevenness according to a predetermined threshold based on the luminance values of the maximum pixel and the minimum pixel detected in the extreme pixel detection process It is preferable to provide a process.

In this configuration, pixels having different luminance display from the surroundings are detected by extreme pixel detection filter processing (extreme pixel detection step).
Here, if the luminance is slightly different from that of the surrounding area, pixel unevenness is not caused. However, in the classification step, only pixels having a difference of a certain threshold or more with respect to the surrounding pixels are classified.
According to such a configuration, all extreme pixels (maximum pixels, minimum pixels) are first picked up by the extreme pixel detection filter process, and further, the aperture is narrowed down by the threshold value in the separation process. Therefore, an abnormal pixel that leads to pixel unevenness is not missed.
In addition, it is possible to appropriately determine the presence / absence of pixel unevenness by excluding extreme pixels that do not lead to pixel unevenness by classifying only abnormal pixels in the classification step, and excluding them from the object of determining the pixel unevenness defect.

  In the present invention, a voting space generating step for generating a voting space having a region corresponding to each region generated by dividing the image in the region dividing step, and abnormal pixels counted for each region in the abnormal pixel counting step. A voting step of voting a number to a corresponding region of the voting space, wherein the pixel unevenness candidate extracting step includes a region in which the number of abnormal pixels voted in each region of the voting space is equal to or greater than a predetermined threshold value. It is preferable to extract.

In such a configuration, in response to dividing the image to be inspected into a plurality of regions in the region dividing step, a voting space having a region corresponding to the region generated by image division is generated (voting space). Production process). The number of abnormal pixels counted in the abnormal pixel counting process is voted for each area of the voting space generated in this way (voting process). Then, based on the number of abnormal pixels voted for each area of the voting space in this way, an area that becomes a pixel unevenness candidate is extracted.
According to such a configuration, since data such as the number of abnormal pixels is voted in the voting space generated corresponding to the image to be inspected, the processing in this voting space directly corresponds to the image to be inspected. Thus, the coordinates of the region extracted as the pixel unevenness candidate in the voting space are directly the pixel unevenness candidate region in the inspection image. By performing the pixel unevenness detection process in the voting space in this manner, the pixel unevenness detection process can be easily associated with the inspection image.

  In the extreme pixel detection step, the maximum pixel and the minimum pixel are separately detected, so that the voting space generated in the voting space generation step is a high-luminance pixel voting space for voting the number of maximum pixels. Preferably, the low-luminance pixel voting space for voting the number of minimum pixels is separately generated.

  In the present invention, an abnormal pixel total count step for counting the total number of the abnormal pixels in all regions extracted in the pixel unevenness candidate extraction step, an average luminance calculation step for calculating an average luminance for the abnormal pixels, And the defect determination step compares the values calculated in the abnormal pixel total count step and the average luminance calculation step with respect to a threshold curve defined in advance in relation to the total number of abnormal pixels and the average luminance. It is preferable to determine the defect of the display device.

In such a configuration, the total number of abnormal pixels present in all the regions extracted as pixel unevenness candidates is counted (abnormal pixel total number counting step). Also, the average luminance of abnormal pixels is calculated (average luminance calculation step).
Then, in the defect determination step, it is determined whether there is a pixel unevenness defect in the image by the display device based on a threshold curve defined in advance by the relationship between the total number of abnormal pixels and the average luminance, and whether the display device is defective or non-defective. To do.
In this way, it is possible to determine whether the product is non-defective or defective by comparing the quantitative value of the number of abnormal pixels and the average luminance with the threshold curve. Can do. Then, it is possible to quickly and accurately determine the presence or absence of a pixel unevenness defect.

Here, in the defect determination step, the threshold curve defined in advance by the relationship between the total number of abnormal pixels and the average luminance is generally defined in an inversely proportional relationship, and when the number of abnormal pixels is small, the average luminance of the abnormal pixels is around. It is preferable to determine that the product is a non-defective product even if it is relatively deviated from the brightness of the product, and to determine that the product is a good product when the average brightness of the abnormal pixels is small from the surrounding brightness when the number of abnormal pixels is large. In other words, even if the number of abnormal pixels is small, if the average luminance of the abnormal pixels greatly deviates from the surroundings, a feeling of roughness occurs in the image. In such a case, it is determined as a defective product. Further, since the roughness is conspicuous if the number of abnormal pixels is large, it is determined that the product is defective even when the deviation from the average luminance around the abnormal pixels is small.
According to the threshold curve of such a relationship, a non-defective product / defective product can be determined on the basis of a determination criterion that approximates the roughness of an image that is felt by human eyes.

  In the present invention, a pixel unevenness area calculating step for calculating the total area of the regions extracted as pixel unevenness candidates based on the number of regions extracted in the pixel unevenness candidate extracting step, and calculating an average luminance for the abnormal pixels. An average luminance calculation step, wherein the defect determination step is calculated in the pixel unevenness area calculation step and the average luminance calculation step with respect to a threshold curve defined in advance in relation to the pixel unevenness area and the average luminance. It is preferable to determine the defect of the display device by comparing the values.

In such a configuration, the total area of the regions extracted as pixel unevenness candidates is calculated (pixel unevenness area calculating step), and the average luminance of abnormal pixels is calculated (average luminance calculating step).
Then, in the defect determination step, it is determined whether there is a pixel unevenness defect in the image by the display device or whether the display device is defective or non-defective based on a threshold curve defined in advance by the relationship between the pixel unevenness area and the average luminance. .
Since it is possible to determine whether the product is non-defective or defective by comparing the quantitative values of the pixel unevenness area and the average luminance with the threshold curve in this manner, it is possible to automatically determine pixel unevenness defects (in terms of arithmetic processing). it can. Then, it is possible to quickly and accurately determine the presence or absence of a pixel unevenness defect.

Here, in the defect determination step, the threshold curve defined in advance by the relationship between the pixel unevenness area and the average luminance is generally defined by an inversely proportional relationship, and when the pixel unevenness area is small, the average luminance of the abnormal pixel is around. It is preferable to determine that the product is non-defective even if it is relatively deviated from the brightness of the pixel, and to determine that the pixel is non-defective when the average luminance of the abnormal pixel is small from the surrounding luminance when the pixel unevenness area is large.
In other words, even if the pixel unevenness area is small, if the average luminance of the abnormal pixels is greatly deviated from the surroundings, the image has a rough feeling. In such a case, it is determined as a defective product. Further, since the roughness is conspicuous when the pixel unevenness area is large, it is determined that the defective product is defective even when the deviation from the periphery of the average luminance of the abnormal pixel is small.
According to the threshold curve of such a relationship, a non-defective product / defective product can be determined on the basis of a determination criterion that approximates the roughness of an image that is felt by human eyes.

  In the present invention, the inspection image acquisition step includes a background image acquisition step of acquiring a background image, an imaging step of capturing an image by the display device to be inspected, and the background from the image data captured in the imaging step. And a background image difference processing step of subtracting the image.

  When it is desired to detect pixel unevenness existing in an image by a display device to be inspected, for example, if a screen or lens on which an image is projected has dirt, the image to be inspected cannot be analyzed accurately. In this regard, according to the present invention, since a background image is captured in advance and dirt or the like present on the screen or lens is removed from the image by background image difference processing, only the image data to be inspected can be acquired. Pixel unevenness existing in the inspection image can be accurately detected.

  In addition, it is preferable to provide the display area extraction process which extracts a display area from the image data imaged at the said imaging process. For example, when the image is distorted into a barrel shape, a pincushion shape, or the like due to the influence of the lens, or when the image is inclined due to installation of an imaging device or the like, the display area is extracted by correcting the image, and this is used as the inspection target image.

  In the present invention, it is preferable that the image data processed in the background image difference processing step includes a pixel unevenness enhancement step of performing a filter process for removing a low frequency component and enhancing a high frequency component.

  An example of a filter that emphasizes such pixel unevenness is a secondary differential filter. For example, it is possible to obtain a luminance difference (luminance height) that the pixel of interest has with respect to surrounding pixels by a filtering process such as a secondary differential filter, and an abnormal pixel based on the magnitude of the luminance difference (luminance height). , And consequently pixel unevenness can be detected.

  According to the pixel unevenness defect detection apparatus of the present invention, an image displayed when a pixel having a higher luminance or a lower luminance than a surrounding pixel is present in a certain area within an image displayed by a display device is displayed. A pixel non-uniformity defect detection device that detects a defective pixel non-uniformity defect, based on inspection image acquisition means for acquiring an image from a display device to be inspected, and image data acquired by the inspection image acquisition means Pixel unevenness detection means for detecting pixel unevenness in the image data, and defect determination means for determining a defect of the display device based on a detection result by the pixel unevenness detection means, the pixel unevenness detection means, A region dividing unit that divides the image data acquired by the inspection image acquiring unit into a plurality of regions, and a luminance higher than the surrounding pixels for each region. Alternatively, an abnormal pixel detection unit that detects an abnormal pixel that is low and has a luminance difference with respect to a reference value set by surrounding pixels that is equal to or greater than a predetermined threshold, and the number of abnormal pixels detected by the abnormal pixel detection unit for each region. An abnormal pixel counting means for counting, and a pixel unevenness candidate extracting means for extracting a candidate area that becomes a pixel unevenness based on the number of abnormal pixels for each of the areas counted by the abnormal pixel counting means, and the defect determination The means is characterized by discriminating a defect of the display device based on a quantitative evaluation of the region extracted by the pixel unevenness candidate extraction means.

  According to the pixel unevenness defect detection program of the present invention, an image displayed when a pixel having a higher brightness or a lower brightness than a surrounding pixel is present within a certain area in an image displayed by a display device is displayed. An inspection image acquisition step of acquiring an image from a display device to be inspected in a computer incorporated in a pixel unevenness defect detection device that detects a defective pixel unevenness defect, and the image data acquired in the inspection image acquisition step A pixel non-uniformity detection step for detecting pixel non-uniformity in the image data, and a defect determination step for determining a defect of the display device based on a detection result of the pixel non-uniformity detection step. The pixel unevenness detecting step is a region division for dividing the image data acquired in the inspection image acquiring step into a plurality of regions. An abnormal pixel detection step for detecting an abnormal pixel whose luminance is higher or lower than the peripheral pixels for each region and whose luminance difference with respect to a reference value set by the peripheral pixels is equal to or greater than a predetermined threshold; and An abnormal pixel count step for counting the number of abnormal pixels detected in the pixel detection step for each region, and a candidate region that causes pixel unevenness based on the number of abnormal pixels for each region counted in the abnormal pixel count step A pixel unevenness candidate extracting step for extracting the display device, wherein the defect determining step determines a defect in the display device based on a quantitative evaluation of the region extracted in the pixel unevenness candidate extracting step. To do.

  The recording medium of the present invention is characterized in that the pixel unevenness defect detection program is recorded.

According to such a configuration, the same effects as those of the above-described invention can be achieved.
That is, by detecting abnormal pixels for each region and extracting a region where the number of abnormal pixels is greater than or equal to a certain value, a region that leads to a pixel unevenness defect can be extracted. Furthermore, since the pixel non-uniformity candidate regions can be extracted in this way, pixel non-uniformity defects can be determined automatically (in terms of arithmetic processing) based on quantitative values for these extracted regions. Then, the presence / absence of the pixel unevenness defect can be quickly and accurately determined, and the quality of the display device product can be determined.

Further, if a program is configured to cause a computer having a CPU (central processing unit) and a memory (storage device) to execute the above steps, parameters in each step can be easily changed. For example, various parameters such as the size of the area in the region dividing process, the filter and threshold value for detecting abnormal pixels, the threshold value in the pixel unevenness candidate extracting process, etc. can be used for the characteristics of the image to be inspected, the purpose of inspection, etc. Can easily be changed accordingly.
Then, the program may be recorded on a recording medium, and the recording medium may be directly inserted into the computer to install the program in the computer. A reading device that reads information on the recording medium is externally attached to the computer, and the reading device The program may be installed on the computer.
The program may be supplied and installed on a computer via a communication line such as the Internet, a LAN cable, a telephone line, or wirelessly.

DESCRIPTION OF EMBODIMENTS Hereinafter, embodiments of the present invention will be illustrated and described with reference to reference numerals attached to respective elements in the drawings.
(First embodiment)
A first embodiment according to a pixel unevenness defect detection method of the present invention will be described.
FIG. 1 is an overall configuration diagram of a pixel unevenness defect detection system 100 that realizes a pixel unevenness defect detection method.
This system 100 includes a liquid crystal light valve 112 to be inspected for whether there is a pixel unevenness defect, an image projection unit 110 that projects an image, a pattern generator 120 that drives the liquid crystal light valve 112, and an image projection unit. A projection unit 130 having a screen 134 for projecting an image projected from 110, a CCD camera (imaging means) 140 for capturing an image projected on the screen 134, and pixels based on image data captured by the CCD camera 140 And an arithmetic processing unit (pixel unevenness defect detection device) 150 that detects unevenness and determines pixel unevenness defects.

The image projection unit 110 is a light source 111, a liquid crystal light valve (display device) 112 that is driven by the pattern generator 120 and modulates the light from the light source 111, and a projection lens that collects and projects the light from the liquid crystal light valve 112. 114 and a mirror 113 that reflects the light from the projection lens 114 toward the screen 134.
The liquid crystal light valve 112 is a liquid crystal panel including a liquid crystal cell for each pixel. When each liquid crystal cell has a polarization defect or the like, a display defect such as a pixel unevenness defect occurs in a projected image.
The image projecting unit 110 imitates a projection type image display device such as a projector, for example.

The pattern generator 120 applies a drive signal to each switching element that drives each liquid crystal cell of the liquid crystal light valve 112 to drive the liquid crystal light valve 112 in a predetermined pattern.
Here, when performing pixel unevenness defect detection, it is inspected whether all display pixels exhibit the same luminance at the same input level, and the pattern generator 120 drives all liquid crystal cells at the same input level.

  The projection unit 130 includes a dark box 131 and a screen 134 provided in the dark box 131. The dark box 131 has an opening 132 into which light from the image projection unit 110 is incident, and an opening 133 for capturing an image on the screen 134 with the CCD camera 140.

The CCD camera 140 captures a projected image on the screen 134 and outputs the captured image data to the arithmetic processing unit 150.
Here, since the pixel unevenness defect is detected based on the image data captured by the CCD camera 140, the CCD camera 140 needs to have a function of capturing an image with the number of gradations exceeding the number of gradations of the liquid crystal light valve 112. There is.
For example, when 256 gradations are expressed by the liquid crystal light valve 112, the image data of the CCD camera 140 has the gradation number of 12-bit data in which “0” is black and “4095” is white.

The arithmetic processing unit 150 includes a CPU (central processing unit) and a predetermined memory (storage device), and by incorporating a pixel unevenness defect detection program, as shown in FIG. The functions of the pixel unevenness enhancement unit 220, the pixel unevenness detection unit 230, and the defect determination unit 240 are realized.
Here, the inspection image acquisition unit 210 includes a background image acquisition unit 211, an image data acquisition unit 212, a display area extraction unit 213, and a background difference processing unit 214. The pixel unevenness detection unit 230 includes an area division. Means 310, voting space generation means 320, voting means 330, abnormal pixel detection means 340, abnormal pixel count means 350, pixel unevenness candidate extraction means 360, abnormal pixel total count means 370, average luminance height A calculation unit (average luminance calculation unit) 380 and a pixel unevenness area calculation unit 390; and the abnormal pixel detection unit 340 further includes an extreme value pixel detection unit 341, an array table 342, and a classification unit 343. Prepare.
The operation of each means will be described with reference to the flowcharts of FIGS.

(Pixel unevenness detection method)
Next, the pixel unevenness defect detection method will be described with reference to the flowcharts of FIGS.
First, in ST100, an inspection image that is a target for detecting a pixel unevenness defect is acquired (inspection image acquisition step).
The inspection image acquisition process (operation of the inspection image acquisition means) will be described with reference to the flowchart of FIG.
In the inspection image acquisition step ST100, first, in ST110, the background image acquisition unit 211 performs a background image acquisition step of acquiring a background image serving as a base for eliminating the influence of dirt or the like on the screen 134.
This is because an image is projected on the screen 134 using the liquid crystal light valve 112 which is known to have almost no defects (in addition to pixel unevenness defects, point defects, line defects, and surface defects). This is done by taking an image. The captured image data is stored in the background image acquisition unit 211. At this time, about 20 images are picked up and the average of them is taken as the background image.

  Next, in ST120, the liquid crystal light valve 112 to be inspected is set on the image projecting means 110, and an image is projected toward the screen 134 (image projection process). In ST130, this image is captured by the CCD camera 140 (image). Data acquisition process). The captured image data is output to and stored in the image data acquisition unit 212.

In ST140, the background image difference processing means 213 performs a background image difference processing step of subtracting the background image from the image data to be inspected.
This is a step of generating a background difference image by subtracting the background image acquired in the background image acquisition step (ST110) from the image data acquired in the image data acquisition step (ST130). The luminance value is subtracted for each pixel to be processed.
The background difference image acquired in this way becomes an inspection image which is a detection target of the pixel unevenness defect.

In ST150, the display area extracting unit 214 performs a display area extracting step of extracting only the screen portion to be inspected from the image data.
The image projected on the screen 134 does not correspond to the entire surface of the screen 134. For example, in the display area extraction step (ST150), unnecessary portions are extracted from the image data when projected in the central area excluding the edge portion. Trimming to extract only the image in the display area. As a method for extracting the display area in this way, for example, an edge detection pattern for detecting four corners is prepared in advance, and the display area is specified by detecting the coordinates of the four corners of the display area by pattern matching. As an example.
Furthermore, depending on the inclination of the projection lens 114, the display area may be trapezoidal instead of rectangular, and depending on the inclination of the CCD camera 140, it may be captured as a trapezoid instead of a rectangle. Based on the coordinates of the four corners, it is converted into a rectangle.

After the inspection image is acquired in the inspection image acquisition step (ST100), the pixel unevenness enhancement unit 220 performs a pixel unevenness enhancement step of enhancing pixel unevenness existing in the inspection image in ST200.
This is, for example, a process of performing filter processing for removing low frequency components and enhancing high frequency components from the inspection image. An example of such a filter is the second-order differential filter shown in FIG. 7. When the second-order differential filter is applied to the 5 × 5 region in FIG. 8, the peripheral pixels of the center M pixel that is the target pixel are included. luminance level L _M for is calculated as follows.
Here, the luminance height L _M is expressed by the following equation, and the difference in luminance value (in the skipped one) of the luminance values of the center pixel to be focused on with respect to the average (reference value) of the luminance values of the peripheral pixels ( (Luminance height). Therefore, if the luminance value of the focused central pixel is equal to the average luminance value of the peripheral pixels, the luminance height is 0, and if the luminance value is higher or lower than the average luminance of the peripheral pixels, it becomes a positive or negative value.

L _M = M− (A + B + C + D + E + F + J + K + O + P + T + U + V + W + X + Y) / 16

Thus for all of the pixels by performing a filtering process for calculating the luminance level L _M of each pixel.
As an example, FIG. 9 shows the luminance value of each pixel before the filtering process, and FIG. 10 shows the luminance height of each pixel after the filtering process.

After the filtering process by the pixel unevenness enhancement step (ST200) is completed, a pixel unevenness detecting unit 230 performs a pixel unevenness detecting step for detecting a region where the pixel unevenness defect exists from the inspection image in ST300.
The pixel unevenness detection step (ST300) will be described with reference to the flowcharts of FIGS.
First, in ST301, an area size for dividing an inspection image into a plurality of areas is determined (division area size determining step).
The pixel unevenness defect is a state in which a certain number or more of pixels having different luminance compared to the surrounding area exist in a certain area (region). In other words, in order to detect such a state, the inspection image is divided into regions of a predetermined area, and the number of pixels having higher or lower luminance than the surroundings is detected and counted in this region. . Therefore, in the divided region size determination step (ST301), an appropriate region size is determined for detecting the pixel unevenness defect. For example, as shown in FIG. 11, one image is composed of a set of small pixels, and 100 × 100 pixels are used as one region.

Next, in ST302, the inspection image is divided by the region dividing means 310 according to the region size determined in the divided region size determining step (ST301) (region dividing step, see FIG. 11).
Further, two voting spaces having areas corresponding to the respective areas of the inspection image divided in the area dividing step (ST302) are generated by the voting space generating means 320 (see FIGS. 12 and 13).
The voting space is a plot table that plots the number of abnormal pixels detected for each region of the inspection image and the degree of abnormality (average luminance height of the abnormal pixels), and is generated as having the same size as the inspection image. The
And in each area | region of a test | inspection image, since a pixel with high brightness | luminance and a low pixel are counted separately, the high brightness | luminance pixel voting space for recording the number of pixels with high brightness | luminance as a voting space (refer FIG. 12) And a low-luminance pixel voting space (see FIG. 13) for recording the number of pixels with low luminance.

Next, in ST304, the extreme pixel detection unit 341 detects an extreme pixel in which the luminance is higher or lower than that of the surroundings for each region and the luminance difference with respect to the surrounding pixels is equal to or greater than a predetermined threshold (extreme pixel detection unit 341). Detection step).
This step is a step of performing abnormal pixel detection filter processing for detecting a maximal pixel that is a pixel having a higher luminance than the surroundings and a minimal pixel that is a pixel having a lower luminance.
This extreme pixel detection filter detects, for example, pixels that satisfy the following conditions as a maximum pixel and a minimum pixel with respect to the 3 × 3 pixels shown in FIG.
That is, in detecting the maximum pixel, the brightness height of the target pixel at the center with respect to the brightness heights (A, B, C, D, F, G, H, I) of the surrounding eight pixels. Detect that E is high. Further, in detecting the minimum pixel, the luminance height E of the central pixel of interest is lower than the luminance heights (A, B, C, D, F, G, H, I) of the surrounding eight pixels. Detect that.

When each pixel is processed by the abnormal pixel detection filter (abnormal pixel detection step ST304) and there is a maximum pixel or minimum pixel (ST305: YES), in ST306, the coordinate value (X , Y) and the luminance height of the pixel are arranged and stored in the arrangement table 342 (see FIGS. 15 and 16).
FIG. 15 shows the coordinate value of the maximal pixel and its luminance height, and FIG. 16 shows the coordinate value of the tiny pixel and its luminance height.

While the pixel of interest is sequentially moved (ST318) and filter detection is completed for all the pixels in the region (ST307: YES), in ST308, the data stored in the array table 342 (ST306) is set according to a predetermined threshold. Sorting is performed by the sorting unit 343 (sorting step).
When pixels having a high or low luminance level with respect to surrounding pixels are detected and stored in the array table 342 (FIGS. 15 and 16), even if the luminance level is higher or lower than the surrounding pixels, it is acceptable. If it is within the range, it will not lead to pixel unevenness defects.
Therefore, the luminance height is classified according to a predetermined threshold value, and only abnormal pixels that lead to pixel unevenness defects are extracted.
A maximum threshold value and a minimum threshold value are set for each of the maximum pixel and the minimum pixel, and for the maximum pixel, a pixel having a luminance height of 60 or more is extracted. For a minimum pixel, the luminance height is −60 or less. As an example, it is assumed that these pixels are extracted.

  Here, an abnormal pixel detection step for detecting abnormal pixels having higher or lower luminance than the surrounding pixels is configured for each region by the processing from the extreme pixel detection step of ST304 to the classification step of ST308. .

  In ST309, the number of extracted maximum abnormal pixels and the number of extracted minimum abnormal pixels are counted by the abnormal pixel counting means 350, respectively, and in ST310, the average of the extracted maximum abnormal pixels and the minimum abnormal pixels is averaged. Each brightness height is calculated.

In ST311, the counted maximum abnormal pixel number and the calculated average luminance height of the maximum abnormal pixel are plotted in the high luminance pixel voting space (see FIG. 12), and the counted minimum abnormal pixel number and the calculated The average luminance height of the minimal abnormal pixels is plotted in the low luminance pixel voting space (see FIG. 13). This step is executed by the voting means 330 (voting step).
12 and 13 show the results of voting the number of maximum abnormal pixels and minimum abnormal pixels.

A pixel unevenness candidate extraction step for extracting pixel unevenness candidates in ST313, after the investigation has been completed for all the areas in the inspection image (ST312: YES) while moving the areas in the inspection image (ST319). Is performed by the pixel unevenness candidate extraction means 360.
In each region of the high-luminance pixel voting space (FIG. 12) and the low-luminance pixel voting space (FIG. 13), the number of maximum abnormal pixels and minimum abnormal pixels is plotted. A region having a maximum abnormal pixel or a minimum abnormal pixel is extracted as a pixel unevenness defect candidate.
For example, extracting a region with a count number of 30 or more as a pixel unevenness defect candidate for the high-luminance pixel voting space (FIG. 12) and the low-luminance pixel voting space (FIG. 13) is an example.
This process is performed independently for the high-luminance pixel voting space (FIG. 12) and the low-luminance pixel voting space (FIG. 13).

Next, in ST314, a pixel unevenness area calculating step for calculating the area of the region extracted as the pixel unevenness candidate is performed by the pixel unevenness area calculating means 390. Further, in ST315, abnormal pixels (maximum abnormal pixels, minimal abnormal pixels) are calculated. The abnormal pixel total count process for counting the total number of pixels) is performed by the abnormal pixel total count means 370.
In the pixel unevenness area calculating step (ST314), the area _Smax of the region that is a pixel unevenness candidate in the high brightness pixel voting space (FIG. 12) is calculated, and the area that is a pixel unevenness candidate in the low brightness pixel voting space (FIG. 13). The area S _min is calculated.
In the abnormal pixel total count step (ST315), the total number w _Tmax of maximally abnormal pixels existing in the entire region of pixel unevenness candidates in the high luminance pixel voting space (FIG. 12) is counted, and the low luminance pixel voting space ( In FIG. 13), the total number w _Tmin of the minimally abnormal pixels existing in the entire region of the pixel unevenness candidate is counted.

Next, in ST316, an all-maximum average luminance height calculation step (average luminance calculation step) is performed, and in ST317, an all-minimum average luminance height calculation step (average luminance calculation step) is performed. This step is executed by the average luminance height calculation means 380.
In the total maximum average brightness height calculation step (ST316), the average brightness height for all maximum abnormal pixels existing in all areas that are high brightness pixel unevenness candidates in the high brightness pixel voting space (FIG. 12) is calculated. calculate.
Similarly, in the all-minimum average luminance height calculation step (ST317), the average luminance for all the minimal abnormal pixels existing in all the regions that are candidates for low-luminance pixel unevenness in the low-luminance pixel voting space (FIG. 13). Calculate the height.
Here, for example, in the high luminance pixel voting space (FIG. 12), 1, 2 extracted region as a high luminance pixel unevenness candidate, denoted ..., k, the number of up to ... N _max, the k-th region Where w _{max, k} is the number of maximally abnormal pixels existing in, and L _{max, k} is the maximal average luminance height of this k-th region, the total maximal average luminance height is expressed by the following equation.

Thus, the area S _max of the region which is a pixel unevenness candidate obtained in the high luminance pixel voting space (FIG. 12), the total number of maximum abnormal pixels w _Tmax , the total maximum average luminance height L _Tmax , and the low luminance pixels The area S _min of the region which is a pixel unevenness candidate obtained in the voting space (FIG. 13), the total number of minimum abnormal pixels w _Tmin , and the total minimum average luminance height L _Tmin are respectively stored in a predetermined table (see FIG. 17). The

Then, based on the result in the pixel unevenness defect detection step (ST300), in ST400, the defect determination unit 240 performs a defect determination step for determining whether the liquid crystal light valve 112 to be inspected is a good product or a defective product.
In the defect determination step (ST400), the pixel unevenness area _Smax of the maximum abnormal pixel obtained in the pixel unevenness candidate detection step (ST300), the maximum maximum pixel number _wTmax , the total maximum average luminance height _LTmax , and the minimum abnormal pixel When the pixel unevenness area S _min , the total number of minimum pixels w _Tmin , and the total minimum average luminance height L _Tmin are plotted on a graph, and all the points are in the non-defective range, the liquid crystal light valve 112 is non-defective. Judge.
There are two graphs, the first graph and the second graph, for plotting the above values (S _max , w _Tmax , L _Tmax , S _min , w _Tmin , L _Tmin ).
As shown in FIG. 17, the first graph is a graph (w) showing the relationship between the average luminance heights L _Tmax and L _Tmin (total maximum or minimum) relative to the total number of abnormal pixels w _Tmax and w _Tmin (maximum or minimum). _Tmax− L _Tmax , w _Tmin −L _Tmin ).
As shown in FIG. 18, the second graph is a graph (S _max −L _Tmax , showing the relationship between the average luminance heights L _Tmax and L _Tmin (total maximum or total minimum) with respect to the pixel unevenness areas S _max and S _min . S _min −L _Tmin ).

In the first graph (FIG. 17) and the second graph (FIG. 18), threshold curves for defining the non-defective range are set.
All threshold curves are generally in an inversely proportional relationship. For example, the threshold curve R of w _Tmax −L _Tmax (the total maximum average luminance height L _{Tmax with} respect to the maximum abnormal pixel total number w _Tmax ) in the first graph (FIG. 18). ₁ , if the maximum abnormal pixel total number w _Tmax is small, it is determined that the average luminance height L _Tmax is a good range even if the average luminance height _LTmax is (somewhat) large, and if the average luminance height _LTmax is small, the maximum abnormal pixel total number w _Tmax is large. However, the relationship to be judged as the non-defective range is specified.
Further, for example, the threshold curve _{R 2} of the second graph S _max -L _Tmax (total maximum average brightness level for the pixel unevenness area S _max L _Tmax) in (19), the average luminance smaller pixel unevenness area S _max Even if the height L _Tmax is large, it is determined as a non-defective range, and when the average luminance height L _Tmax is small, a relationship is determined to be determined as a non-defective range even if the pixel unevenness area S _max is large.

FIGS. 18 and 19 show how all the relationships are plotted in the first graph and the second graph. In the first graph of FIG. 18, the point of w _Tmax −L _Tmax is “◯”, the point of w _Tmin −L _Tmin is “×”, and in the second graph of FIG. 19, the point of S _max −L _Tmax is The points “Δ” and S _min −L _Tmin are represented by “□”.

  In ST500, it is determined whether all the points are in the non-defective range. If all the points are in the non-defective range (ST500: YES), it is determined that the liquid crystal light valve 112 to be inspected is a non-defective product. If it is determined that even one point is out of the non-defective range (ST500: NO), it is determined that the liquid crystal light valve 112 to be inspected is defective.

According to such a pixel unevenness defect detection method, the following effects can be obtained.
(1) Since a certain number of pixels with poor display of luminance exist within a certain area (region), a pixel unevenness defect occurs. Thus, the entire image is simply binarized as in the prior art. Although it could not be detected by processing alone, according to the above configuration, abnormal pixels are detected for each region (abnormal pixel detection steps ST304 to ST308), and regions where the number of abnormal pixels is equal to or greater than a certain value are extracted. Thus, it is possible to extract a region that leads to a pixel unevenness defect (pixel unevenness candidate extraction step ST313).

(2) it is possible to extract a region to be a pixel unevenness candidate (pixel unevenness candidate extraction step ST 313), for these extracted area, the total area S _max of the extracted area, S _min, the total number of abnormal pixels of the extracted region A pixel unevenness defect can be automatically (calculated) determined based on quantitative values such as w _Tmax and w _Tmin and average luminance heights L _Tmax and L _{Tmin of} abnormal pixels (defect determination step ST400). ). As a result, it is possible to quickly and accurately determine whether there is a pixel unevenness defect, and to determine whether the product of the liquid crystal light valve 112 is good or bad.

(3) All extreme pixels (maximum pixels, minimum pixels) are picked up in the extreme pixel detection step (ST304), and further, the aperture is narrowed down with the threshold value in the classification step (ST308). Abnormal pixels that lead to pixel unevenness are not missed. In addition, it is possible to appropriately determine the presence or absence of pixel unevenness by excluding extremal pixels that do not lead to pixel unevenness by separating only abnormal pixels in the classification step (ST308), and excluding them from the object of determining pixel unevenness defects. Can do.

(4) A voting space (high luminance pixel voting space, low luminance pixel voting space) having an area corresponding to the area generated by dividing the image data in the area dividing process (ST302) is generated (voting space generating process) ST303) Since data such as the number of abnormal pixels is voted in the voting space (voting step ST311), the processing in this voting space corresponds to the image to be inspected as it is, and is extracted as a pixel unevenness candidate in the voting space. The coordinates of the area to be used are the pixel unevenness candidate areas in the inspection image as they are.
By performing the pixel unevenness detection process in the voting space in this manner, the pixel unevenness detection process can be easily associated with the inspection image.

(5) In the defect determination step (ST400), the abnormal pixel total number w _Tmax , w _Tmin , average luminance height L _Tmax , L _Tmin , pixel unevenness area S _max , S _min are shown in the first graph (FIG. 18) and the second graph. By plotting in FIG. 19 and comparing the plotted points with threshold curves (R ₁ , R _2, etc.), it is possible to determine whether the product is a non-defective product or a defective product. Defects can be identified. As a result, it is possible to quickly and accurately determine the presence or absence of pixel unevenness defects.

(6) In the defect determination step (ST400), for example, a threshold curve R _1, it is determined that the average luminance level L _Tmax is (slightly) greater in the non-defective range The less the maximum error pixel total w _Tmax, The average If the luminance height L _Tmax is small, the relationship is determined to be a non-defective range even if the maximum abnormal pixel total number w _Tmax is large. According to such a threshold curve, it approximates the roughness of the image felt by human eyes. It is possible to determine whether the product is non-defective or defective based on the determined criteria.

(7) Since the background image captured in advance is subtracted from the image data to be inspected in the background image difference processing step (ST140), dirt and the like existing on the screen 134 are removed. Therefore, only image data to be inspected can be acquired, and pixel unevenness existing in the inspection image can be accurately detected.

(8) In the pixel unevenness emphasizing step (ST200), a luminance difference (luminance height) that the pixel of interest has with respect to surrounding pixels can be obtained by the filtering process of the secondary differential filter, and each pixel is a surrounding pixel. Can be obtained, and abnormal pixels can be detected and consequently pixel unevenness can be detected based on the magnitude of the luminance difference (luminance height).

It should be noted that the present invention is not limited to the above-described embodiment, and modifications, improvements, and the like within the scope that can achieve the object of the present invention are included in the present invention.
For example, in the inspection image acquisition step ST100, the display area extraction step ST150 and the background image difference processing step ST140 may be omitted depending on circumstances. For example, if an image is projected on the entire screen 134 in the image projection step ST120 or only a part necessary for capturing an image with the CCD camera 140 is captured, the display area extraction step (ST150) is performed. There is no need.
Further, in ST100, the case where an image is captured by the liquid crystal light valve 112, which is known to have no defect when acquiring a background image, has been described as an example. However, the light is applied to the screen 134 without the liquid crystal light valve 112 being provided. , And an image of light projected on the screen 134 may be taken as a background image. As a result, a background image including dirt, defects, etc. existing in the optical path (mirror 113, projection lens 114, screen 134) excluding the liquid crystal light valve 112 can be obtained.

  In the pixel unevenness enhancement step ST200, the luminance difference that each pixel has with respect to surrounding pixels is calculated by the secondary differential filter processing, and the pixel unevenness detection step ST300 is performed based on this luminance difference. Of course, such filter processing is not necessarily required, and the pixel unevenness detection step ST300 may be performed based on the luminance value of the pixel.

In the defect determination step, when the maximum abnormal pixel total number w _Tmax , the average luminance height L _Tmax , the pixel unevenness area S _max, etc. are plotted on a graph and defect determination is performed by comparison with threshold curves (for example, R ₁ , R ₂ ). However, the specific means of defect determination can be variously changed, and the defect determination may be performed based on various data of the pixel unevenness candidate area detected in the pixel unevenness defect detection step ST300.

  The present invention can be used, for example, to determine whether a non-defective product or defective product is inspected by checking whether there is a pixel unevenness defect in a display device such as a liquid crystal panel or a light valve of a projector that is an application product thereof.

1 is an overall configuration diagram of a pixel unevenness defect detection system that realizes a pixel unevenness defect detection method in a first embodiment of the present invention. FIG. The figure which shows the structure of the arithmetic processing part in 1st Embodiment. 5 is a flowchart illustrating a processing procedure of a pixel unevenness defect detection method in the first embodiment. 5 is a flowchart illustrating a processing procedure of an inspection image acquisition process in the first embodiment. 5 is a flowchart illustrating a processing procedure of a pixel unevenness detection step in the first embodiment. 5 is a flowchart illustrating a processing procedure of a pixel unevenness detection step in the first embodiment. The figure which shows a secondary differential filter in 1st Embodiment. The figure which shows the pixel which applies a secondary differential filter in 1st Embodiment. The figure which shows an example of the luminance value of each pixel. The figure which shows an example of the brightness | luminance height of each pixel calculated by applying a secondary differential filter. The figure which shows the example which divided the area | region of image data in 1st Embodiment. The figure which shows high-intensity pixel voting space in 1st Embodiment. The figure which shows the low-intensity pixel voting space in 1st Embodiment. The figure which shows an example of the pixel to which an extreme value pixel detection filter is applied. The figure which shows an example of the arrangement | sequence table which arranged the maximum pixel in 1st Embodiment. The figure which shows an example of the arrangement | sequence table which arranged the minimum pixel in 1st Embodiment. The table which shows the various data in the area | region of the pixel nonuniformity candidate detected at the pixel nonuniformity detection process in 1st Embodiment. The figure which shows the 1st graph of a defect discrimination | determination process in 1st Embodiment. The figure which shows the 2nd graph of a defect discrimination | determination process in 1st Embodiment.

Explanation of symbols

DESCRIPTION OF SYMBOLS 100 ... Pixel unevenness defect detection system, 110 ... Image projection means, 111 ... Light source, 112 ... Liquid crystal light valve, 113 ... Mirror, 114 ... Projection lens, 120 ... Pattern generator, 130 ... Projection part, 131 ... Dark box, 132 ... Opening, 133 ... opening, 134 ... screen, 140 ... CCD camera, 150 ... arithmetic processing unit, 210 ... inspection image acquisition means, 211 ... background image acquisition means, 212 ... image data acquisition means, 213 ... background difference processing means , 214 ... Display area extracting means, 220 ... Pixel unevenness enhancing means, 230 ... Pixel unevenness detecting means, 240 ... Defect determining means, 310 ... Area dividing means, 320 ... Voting space generating means, 330 ... Voting means, 340 ... Abnormal pixels Detection means, 341... Extreme pixel detection means, 342... Array table, 343. Normal pixel counting means, 360 ... pixel irregularity candidate extraction unit, 370 ... error pixel total count means, 380 ... average brightness level calculating means, 390 ... pixel uneven area calculating means, R 1 _... threshold curve, R 2 _... threshold curve, S _max ... pixel uneven area, S _min ... pixel uneven area.

Claims (10)

A pixel that detects a pixel unevenness defect that causes a displayed image to be defective when a pixel having a luminance higher than that of surrounding pixels or a pixel having a lower luminance within a certain area in an image displayed by a display device exists within a certain area. An uneven defect detection method,
An inspection image acquisition step of acquiring an image from a display device to be inspected;
A pixel unevenness detecting step of detecting pixel unevenness in the image data based on the image data acquired in the inspection image acquiring step;
A defect determination step of determining a defect of the display device based on a detection result by the pixel unevenness detection step,
The pixel unevenness detection step includes
A region dividing step of dividing the image data acquired in the inspection image acquiring step into a plurality of regions;
An abnormal pixel detection step for detecting an abnormal pixel whose luminance is higher or lower than the surrounding pixels for each region and whose luminance difference with respect to a reference value set by the surrounding pixels is equal to or greater than a predetermined threshold;
An abnormal pixel counting step of counting the number of abnormal pixels detected in the abnormal pixel detection step for each region;
A pixel unevenness candidate extraction step of extracting candidate regions that become pixel unevenness based on the number of abnormal pixels for each of the regions counted in the abnormal pixel counting step,
The defect determination step includes
A pixel unevenness defect detection method, comprising: determining a defect of the display device based on a quantitative evaluation of the region extracted in the pixel unevenness candidate extraction step. The pixel unevenness defect detection method according to claim 1,
The abnormal pixel detection step includes
An extreme pixel detection step for performing an extreme pixel detection filter process for detecting a maximum pixel that is a pixel having a higher luminance than a surrounding pixel and a minimum pixel that is a pixel having a lower luminance for each region divided in the region division step; ,
A non-uniform pixel separation step of classifying abnormal pixels that are connected to a non-uniform pixel defect by a predetermined threshold based on the luminance values of the maximum pixel and the minimum pixel detected in the extreme pixel detection step. Defect detection method. In the pixel nonuniformity defect detection method of Claim 1 or Claim 2,
A voting space generating step for generating a voting space having a region corresponding to each region generated by dividing the image in the region dividing step;
A voting step for voting the number of abnormal pixels counted for each region in the abnormal pixel counting step to a corresponding region of the voting space, and
The pixel unevenness candidate extracting step includes extracting a region where the number of abnormal pixels voted for each region of the voting space is equal to or greater than a predetermined threshold. In the pixel nonuniformity defect detection method in any one of Claims 1-3,
An abnormal pixel total count step for counting the total number of abnormal pixels in all the regions extracted in the pixel unevenness candidate extraction step;
An average luminance calculation step for calculating an average luminance for the abnormal pixel, and
The defect determination step compares the values calculated in the abnormal pixel total count step and the average luminance calculation step with respect to a threshold curve defined in advance in relation to the total number of abnormal pixels and the average luminance. A pixel unevenness defect detection method characterized by performing defect discrimination. In the pixel nonuniformity defect detection method in any one of Claims 1-4,
A pixel unevenness area calculating step of calculating the total area of the regions extracted as pixel unevenness candidates based on the number of regions extracted in the pixel unevenness candidate extracting step;
An average luminance calculation step for calculating an average luminance for the abnormal pixel, and
The defect determining step compares the values calculated in the pixel unevenness area calculating step and the average luminance calculating step with respect to a threshold curve defined in advance by the relationship between the pixel unevenness area and the average luminance. A pixel unevenness defect detection method characterized by performing defect discrimination. In the pixel nonuniformity defect detection method in any one of Claims 1-5,
The inspection image acquisition step includes
A background image acquisition step of acquiring a background image;
An imaging step of capturing an image by the display device to be inspected;
And a background image difference processing step of subtracting the background image from the image data picked up in the image pickup step. The pixel unevenness defect detection method according to claim 6,
A pixel unevenness defect detecting method comprising: a pixel unevenness enhancing step of performing a filter process for removing a low frequency component and enhancing a high frequency component from the image data processed in the background image difference processing step. A pixel that detects a pixel unevenness defect that causes a displayed image to be defective when a pixel having a luminance higher than that of surrounding pixels or a pixel having a lower luminance within a certain area in an image displayed by a display device exists within a certain area. An uneven defect detection device,
Inspection image acquisition means for acquiring an image from a display device to be inspected;
Pixel unevenness detection means for detecting pixel unevenness in the image data based on the image data acquired by the inspection image acquisition means;
Defect determining means for determining a defect of the display device based on a detection result by the pixel unevenness detecting means,
The pixel unevenness detecting means includes
Area dividing means for dividing the image data acquired by the inspection image acquiring means into a plurality of areas;
Abnormal pixel detection means for detecting an abnormal pixel whose luminance is higher or lower than the surrounding pixels for each region and whose luminance difference with respect to a reference value set by the surrounding pixels is a predetermined threshold value or more;
Abnormal pixel counting means for counting the number of abnormal pixels detected by the abnormal pixel detecting means for each region;
Pixel unevenness candidate extraction means for extracting candidate areas that become pixel unevenness based on the number of abnormal pixels for each of the areas counted in the abnormal pixel counting means,
The defect determination means includes
A pixel unevenness defect detection apparatus, wherein a defect of the display device is determined based on a quantitative evaluation of a region extracted by the pixel unevenness candidate extraction unit. A pixel that detects a pixel unevenness defect that causes a displayed image to be defective when a pixel having a luminance higher than that of surrounding pixels or a pixel having a lower luminance within a certain area in an image displayed by a display device exists within a certain area. In the computer built in the mura defect detection device,
An inspection image acquisition step of acquiring an image from a display device to be inspected;
A pixel unevenness detecting step of detecting pixel unevenness in the image data based on the image data acquired in the inspection image acquiring step;
A defect determination step of determining a defect of the display device based on a detection result by the pixel unevenness detection step,
The pixel unevenness detection step includes
A region dividing step of dividing the image data acquired in the inspection image acquiring step into a plurality of regions;
An abnormal pixel detection step for detecting an abnormal pixel whose luminance is higher or lower than the surrounding pixels for each region and whose luminance difference with respect to a reference value set by the surrounding pixels is equal to or greater than a predetermined threshold;
An abnormal pixel counting step of counting the number of abnormal pixels detected in the abnormal pixel detection step for each region;
A pixel unevenness candidate extraction step of extracting candidate regions that become pixel unevenness based on the number of abnormal pixels for each of the regions counted in the abnormal pixel counting step,
The defect determination step includes
A computer-readable pixel unevenness defect detection program characterized in that a defect of the display device is determined based on a quantitative evaluation of a region extracted in the pixel unevenness candidate extraction step.   The recording medium which recorded the pixel nonuniformity defect detection program of Claim 9.

Images