JP6041858B2 - Method and apparatus for detecting display unevenness of display device - Google Patents

Description

  The present invention relates to a method and apparatus for detecting display unevenness of a display device.

  Taking an image displayed by the display device and detecting the display unevenness of the display device by comparing the captured image data with the image data for display means that the image data for display is obtained with the correction content determined based on the detected display data. It is useful to eliminate display unevenness by correcting the correction. Related techniques are disclosed in Japanese Patent Office Published Patent Publication Nos. 2010-57149 and 2005-150349.

JP 2010-57149 A JP 2005-150349 A

  In detecting the display unevenness of the display device, the present inventors examined a specific analysis method of image data obtained by photographing an image displayed by the display device in order to further improve the detection accuracy. In the process, it was found that the specific analysis method of the image data affects the detection accuracy of display unevenness. The present invention has been made in view of such problems, and an object of the present invention is to provide a method and apparatus for detecting display unevenness of a display device with higher accuracy.

Aspects of the present invention include
Based on output image data obtained by photographing an image displayed by the display device, a pixel value acquisition step of acquiring a pixel value of each pixel of the display device;
A differentiation step of obtaining a differential pixel value of each pixel of the display device;
An uneven area detecting step for detecting an area where the display unevenness occurs in the display device based on a distribution of pixels in which the differential pixel value of the display device exceeds a predetermined unevenness determination threshold;
An intensity value acquisition step of acquiring a display unevenness intensity value of the generation area based on the pixel value or the differential pixel value of each pixel belonging to the generation area;
Display unevenness detecting step for detecting the occurrence area where the intensity value exceeds a predetermined unevenness intensity threshold as the display unevenness; and
The present invention relates to a method for detecting display unevenness of a display device including:

Another aspect of the present invention is:
A pixel value obtaining unit that assigns each pixel value of output image data obtained by photographing an image displayed by the display device to each pixel of the display device, and obtains a pixel value of each pixel of the display device;
Differentiating means for obtaining a differential pixel value of each pixel of the display device;
Pixel value comparison means for comparing the differential pixel value of each pixel of the display device with a predetermined unevenness determination threshold;
An uneven area detecting unit that detects an occurrence area of the display unevenness in the display device based on a distribution of pixels in which the differential pixel value exceeds the unevenness determination threshold;
Intensity value acquisition means for acquiring a display unevenness intensity value of the generation area based on the pixel value or the differential pixel value of each pixel belonging to the generation area;
Intensity value comparing means for comparing the intensity value with a predetermined uneven intensity threshold;
Display unevenness detecting means for detecting the occurrence area where the intensity value exceeds a predetermined unevenness intensity threshold as the display unevenness;
The display unevenness detecting device of the display device comprising:

  According to the aspect of the present invention, the unevenness generation area of the display device is first detected from the distribution of pixels in which the differential pixel value exceeds the unevenness determination threshold. Then, the occurrence area in which the display unevenness intensity value of each occurrence area acquired from the pixel value or the differential pixel value of each occurrence area exceeds the unevenness intensity threshold is finally detected as display unevenness. For this reason, first, an area that is a candidate for display unevenness is primarily detected as an area where display unevenness occurs. Then, by narrowing down the detected occurrence area by the intensity value of the display unevenness, display unevenness that is recognized as visual unevenness is finally detected. Therefore, the display unevenness of the display device can be detected with higher accuracy than simply detecting the display unevenness from the distribution of the differential pixel values.

It is explanatory drawing which shows the state which detects the display nonuniformity of a display device using the display nonuniformity detection apparatus which concerns on the 1st Embodiment of this invention. It is a flowchart which shows the display nonuniformity detection procedure which the display nonuniformity detection apparatus of FIG. 1 performs. It is a flowchart which shows the specific procedure of the differentiation process of FIG. It is explanatory drawing which shows the acquisition principle of the integration pixel value in the integration process of FIG. 3 about the pixel of the left side vicinity area of a liquid crystal panel display. It is explanatory drawing which shows the acquisition principle of the integration pixel value in the integration process of FIG. 3 about the pixel of the upper side vicinity area of a liquid crystal panel display. It is explanatory drawing which shows the acquisition principle of the integration pixel value in the integration process of FIG. 3 about the pixel of the area near the upper left corner of a liquid crystal panel display. It is a flowchart which shows the specific procedure of the differentiation process of FIG. It is explanatory drawing which shows the principle of the emphasis process of FIG. (A), (b) is explanatory drawing which shows partially the image and each pixel value of the output image data before the differentiation process of the liquid crystal panel display of FIG. (A), (b) is explanatory drawing which shows partially the image and each pixel value of the output image data after the differentiation process of the liquid crystal panel display of FIG. It is explanatory drawing which shows the state which provided the label value to the pixel in which the differential pixel value shown in FIG.10 (b) exceeded the nonuniformity determination threshold value. It is explanatory drawing which shows the positional relationship of a foreground (FG) and background (BG) in the generation | occurrence | production area | region of display unevenness required in order to obtain | require a SEMU value. It is a flowchart which shows the test | inspection procedure of the liquid crystal panel display using the detection result of the display nonuniformity by the display nonuniformity detection apparatus of FIG.

  Hereinafter, embodiments of a display unevenness detection apparatus to which the display unevenness detection method of the present invention is applied will be described. The display unevenness detection apparatus of the present invention may be configured in an in-line format incorporated in an inspection line (not shown) or the like in the manufacturing process of the display device, or configured in a stand-alone format separated from the inspection line. Also good.

  Examples of display devices that can detect display unevenness with the display unevenness detection apparatus of the present invention include a liquid crystal panel display, a plasma panel display, and an organic EL display. In the following embodiments, a case where the display device is a liquid crystal panel display will be described as an example.

  As shown in FIG. 1, the display unevenness detection apparatus 1 of the present embodiment is configured in a stand-alone format, and an image such as a test pattern displayed on a liquid crystal panel display 3 (corresponding to a display device) is displayed by a CCD camera 5. The display unevenness of the liquid crystal panel display 3 is detected from the output image data obtained by photographing. The display unevenness detection device 1 can be configured by, for example, a personal computer or the like if there is no problem in processing capability.

  The display unevenness detection device 1 includes a CPU (Central Processing Unit), a RAM (Random Access Memory), a ROM (Read Only Memory), a hard disk, and the like. The CPU executes a program stored in the ROM or the hard disk, thereby executing display unevenness detection processing on the liquid crystal panel display 3.

  As shown in FIG. 2, the display unevenness detection process performed by the display unevenness detection device 1 includes an output image data acquisition process (step S1), an addressing and moire removal process (step S3), and a differential process (step S5), differential threshold determination (primary threshold determination, binarization) processing (step S7), uneven intensity calculation processing (step S9), uneven intensity threshold determination (secondary threshold determination) processing (step S11), and result output Processing (step S13) is included.

  In the output image data acquisition process in step S1, for example, a CCD camera that displays an image such as a test pattern on the liquid crystal panel display 3 based on the input image data supplied from the display unevenness detection device 1 to the liquid crystal panel display 3 and shoots the display screen. The display unevenness detection apparatus 1 acquires the video signal from 5 as output image data of the liquid crystal panel display 3.

  Here, the display unevenness generated in the liquid crystal panel display 3 includes the brightness unevenness and the color unevenness, and the display unevenness detecting device 1 of the present embodiment can detect both the brightness unevenness and the color unevenness. For this purpose, the liquid crystal panel display 3 displays an image suitable for detecting luminance unevenness and an image suitable for detecting color unevenness by appropriately changing the RGB value pattern. Then, the display unevenness detection apparatus 1 performs the display unevenness detection operation according to the following procedure for each image.

  In the addressing process of step S3, each pixel of the CCD sensor of the CCD camera 5 is assigned to each pixel of the liquid crystal panel display 3, and each pixel value of the liquid crystal panel display 3 is determined from the pixel value of each pixel of the CCD sensor constituting the output image data. Determine the pixel value of a pixel.

  Even when the display device is an organic EL panel display or a plasma panel display, an addressing process is performed using one light emitting element as one pixel.

  By the way, the liquid crystal panel display 3 and the CCD sensor each have a lattice pattern in which pixels are arranged in a matrix. Since the CCD camera 5 has more pixels than the liquid crystal panel display 3, image light from one pixel of the liquid crystal panel display 3 is received by a plurality of pixels of the CCD camera 5. Therefore, the pixel value of each pixel of the liquid crystal panel display 3 is determined based on the pixel having the highest pixel value among the plurality of pixels of the CCD camera 5 corresponding to the pixel, for example.

  At this time, if the CCD camera 5 has pixels that are integral multiples of the liquid crystal panel display 3, no phase difference occurs between the pixel period of the CCD camera 5 and the pixel period of the liquid crystal panel display 3. Therefore, when the pixels of the liquid crystal panel display 3 emit light with the same pixel value, the pixels with the highest pixel value among the pixels of the CCD camera 5 corresponding to each pixel have the same pixel value. Therefore, moire fringes do not occur in the captured image of the CCD camera 5 that captured the display image of the liquid crystal panel display 3.

  However, if the CCD camera 5 has a number of pixels that is not an integral multiple of the liquid crystal panel display 3, a phase difference occurs between the pixel period of the CCD camera 5 and the pixel period of the liquid crystal panel display 3. Therefore, even when each pixel of the liquid crystal panel display 3 emits light with the same pixel value, the pixel with the highest pixel value among the pixels of the CCD camera 5 corresponding to each pixel does not have the same pixel value. This causes moiré fringes in the captured image of the CCD camera 5 that captured the display image on the liquid crystal panel display 3.

  If the output image data from the CCD camera 5 is used for detecting the display unevenness of the liquid crystal panel display 3 while including the moire fringes, there is a possibility that the display unevenness is erroneously detected.

  In step S3, therefore, the moire removal process is performed together with the addressing process. In the moiré removal processing, for example, output using the technique of adding or averaging the pixel values of each pixel of the CCD sensor and the surrounding pixels proposed by the applicant of the application according to Japanese Patent Application Laid-Open No. 2004-317329. Remove moire components in image data.

  Even when the display device is an organic EL panel display or a plasma panel display, it is effective to perform the same moire removal process together with the addressing process because the light emitting elements are arranged in a matrix lattice pattern. However, the moire removal process is not essential, and the moire removal process may be omitted when the generation of the moire fringes is not hindered when the display unevenness is detected.

  In the differentiation process of step S5, the pixel value of each pixel of the liquid crystal panel display 3 after the addressing process and the moire removal process performed in step S3 is differentiated to obtain a differential pixel value. The differential pixel value may be obtained by a general differentiation process for obtaining a difference between pixel values of the target pixel and its surrounding pixels.

  However, in the present embodiment, in the differentiation process in step S5, as shown in FIG. 3, the pixel value enhancement process (step S51), the integration process (step S51), and the liquid crystal panel display of each pixel of the liquid crystal panel display 3 are performed. Difference processing between the pixel value of each pixel 3 and the integral pixel value (step S53) is performed.

  In the integration process in step S51, the pixel value of each pixel of the liquid crystal panel display 3 after the addressing process and the moire removal process performed in step S3 of FIG. 2 is averaged and integrated with the pixel values of the surrounding pixels using a spatial filter. Then, the integrated pixel value is acquired.

  The spatial filter used here needs to have a matrix shape that covers display unevenness that may occur in the liquid crystal panel display 3. Therefore, the spatial filter has a kernel size corresponding to display unevenness that may occur in the liquid crystal panel display 3. For example, if there is a possibility that the display unevenness has a size of 100 × 100 pixels of the liquid crystal panel display 3 at the maximum, the spatial filter used for integration is also of a kernel size of 100 × 100. Incidentally, the value of each kernel is “1”, and the coefficient is the reciprocal of the number of kernels (= 1 / (100 × 100)).

  When the integration process of step S51 is performed using the spatial filter, if the target pixel to be integrated (the acquisition target pixel of the integrated pixel value) approaches any of the upper, lower, left, and right outer peripheries of the liquid crystal panel display 3, the spatial filter Some kernel rows protrude outside the liquid crystal panel display 3.

  4 and 5, the pixel value of each pixel of the liquid crystal panel display 3 is integrated with the integration spatial filter 40 (kernel size spatial filter corresponding to the shape and size of the display unevenness of the display device). A case where integration is performed using will be described as an example. In the example described here, the kernel size of the spatial filter 40 is schematically set to 7 × 7 as in the samples shown on the rightmost side of FIG. 4 and the bottom side of FIG. In this spatial filter 40, each kernel value is “1”, and the coefficient of each kernel is the reciprocal of the total number of kernels (= 1 / (7 × 7)).

  First, FIG. 4 shows the relationship between the positional relationship of the spatial filter 40 with respect to the left side 31 of the liquid crystal panel display 3 and the effective kernel sequence. In this example, when the pixels from the left side 31 to the third pixel are integrated by the spatial filter 40 (see the third example from the top in FIG. 4), the leftmost kernel column (1 to 3) Column) protrudes beyond the left side 31 of the liquid crystal panel display 3.

  Since there is no corresponding pixel column in the kernel column that protrudes outside the liquid crystal panel display 3, it is necessary to invalidate this kernel column in the integration process. Therefore, the kernel is invalidated (kernel value = “0”) for the kernel string of the spatial filter 40 that protrudes outside the left side 31.

  Further, when the pixels after the fourth pixel from the left side 31 are integrated by the spatial filter 40 (see the fourth and subsequent examples from the top in FIG. 4), the entire spatial filter 40 fits inside the liquid crystal panel display 3. . In this case, since corresponding pixels exist for all kernels, in principle, no invalid kernel (column) is necessary.

  It should be noted that the integration target pixel existing in the area near the right side of the liquid crystal panel display 3 may be integrated using the spatial filter 40 whose contents are reversed in the left-right direction in FIG.

  As shown in FIG. 5, the same can be done when the spatial filter 40 is close to the upper side 35 of the liquid crystal panel display 3. That is, when the integration target pixel is from the upper side 35 to the third pixel (see the third example from the right in FIG. 5), the upper kernel column (1st to 3rd columns) of the spatial filter 40 is the liquid crystal panel. It protrudes beyond the upper side 35 of the display 3. Therefore, the kernel is invalidated (kernel value = “0”) for the kernel string of the spatial filter 40 that protrudes outside the upper side 35.

  In addition, when the integration target pixel is the fourth and subsequent pixels from the upper side 35 (see the fourth to the leftmost example from the right in FIG. 5), the entire spatial filter 40 fits inside the liquid crystal panel display 3. In principle, there is no need to set an invalid kernel (column) in the spatial filter 40.

  It should be noted that the integration target pixel existing in the area near the lower side of the liquid crystal panel display 3 may be integrated using the spatial filter 40 whose contents are upside down in FIG.

  By the way, in the liquid crystal panel display 3 using the backlight, particularly when the light source is arranged at the outer peripheral edge of the screen and guided to the center of the screen using the light guide plate, the outer periphery of the screen is caused by the attenuation of light in the light guide plate. Shading is likely to occur in which the brightness at the center of the screen is relatively lower than the brightness in the vicinity. This shading may also occur in a display device that does not use a backlight, such as a plasma panel display or an organic EL panel display.

  Therefore, in the present embodiment, when the integration target pixel is near the outer periphery of the liquid crystal panel display 3, the spatial filter 40 has the directionality in the same direction as the extension direction of the nearby side, and the extension of the side is performed. The sensitivity in the direction orthogonal to the current direction is reduced, and shading correction is performed on the integrated pixel value.

  For example, in the vicinity of the left side 31 of the liquid crystal panel display 3 shown in FIG. 4, it is assumed that shading is likely to occur in the vicinity area 33 of the left side 31 that spans seven pixels from the left side 31 to the sensitivity correction line 32. In that case, when the target pixel to be integrated by the spatial filter 40 exists in the neighborhood area 33, the spatial filter 40 is given directionality in the extending direction of the left side 31. In principle, the effective kernel columns in the direction orthogonal to the left side 31 (horizontal direction) are three columns, and the kernel size is vertical × horizontal = 7 × 3.

  However, in the uppermost example of FIG. 4, an effective kernel sequence that overlaps the pixel adjacent to the left of the central integration target pixel of the spatial filter 40 protrudes from the left side 31 of the spatial filter 40 to the outside. The effective kernel size is vertical × horizontal = 7 × 2.

  Similarly, in the vicinity of the upper side 35 of the liquid crystal panel display 3 shown in FIG. 5, it is assumed that shading is likely to occur in the vicinity area 37 of the upper side 35 over the 7-pixel width from the upper side 35 to the sensitivity correction line 36. In that case, when the target pixel to be integrated by the spatial filter 40 exists in the neighborhood area 37, the spatial filter 40 is given directionality in the extending direction of the upper side 35. Then, the effective kernel rows in the direction orthogonal to the upper side 35 (vertical direction) are basically three rows, and the kernel size is vertical × horizontal = 3 × 7.

  However, in the rightmost example of FIG. 5, the effective kernel size that overlaps with the upper adjacent pixel of the integration target pixel protrudes outward from the upper side 35 of the spatial filter 40. Is vertical × horizontal = 2 × 7.

  If shading is likely to occur in the integration target pixel existing in the area near the right side over the 7-pixel width from the right side of the liquid crystal panel display 3 to the sensitivity correction line (not shown), the image shown in FIG. Integration processing may be performed using the spatial filter 40 having the above contents. Similarly, when shading is likely to occur in the integration target pixel existing in the area near the lower side over the width of 7 pixels from the lower side of the liquid crystal panel display 3 to the sensitivity correction line (not shown), Integration processing may be performed using the spatial filter 40 having the inverted contents.

  As described above, for the integration target pixels in the vicinity of the outer periphery of the liquid crystal panel display 3, the spatial filter 40 to be used has a direction in the extending direction of the nearby side, and the effective kernel size is set to the vertical × horizontal direction. = 7 × 2 or 2 × 7, or 7 × 3 or 3 × 7, the spatial filter 40 can be used to simultaneously perform shading correction when integrating the pixel values of the integration target pixels.

  In addition, when the integration target pixel exists inside the liquid crystal panel display 3 with respect to the sensitivity correction lines 32 and 36, the effective kernel size of the spatial filter 40 used for integration of the pixel is basically 7 × 7. Can do. However, the effective kernel size of the spatial filter 40 is vertical × horizontal = 7 × as soon as the integration target pixel moves from the vicinity areas 33 and 37 to the inside of the liquid crystal panel display 3 beyond the sensitivity correction lines 32 and 36. Changing from 3 or 3 × 7 to 7 × 7 is not preferable because the integral characteristic changes abruptly.

  Therefore, as long as the integration target pixel located inside the neighboring areas 33 and 37 is near the sensitivity correction lines 32 and 36, the effective kernel size of the spatial filter 40 is set to the vertical x as the distance from the neighboring areas 33 and 37 increases. The width may be gradually changed to 7 × 5, 5 × 7, or 7 × 7.

  By the way, the above-described kernel size of vertical × horizontal = 7 × 7 is merely an example for explanation, and the kernel size of the spatial filter is arbitrary as long as it is a size corresponding to display unevenness that may occur in the liquid crystal panel display 3. . And about the spatial filter used for integration of the integration object pixel near the outer periphery of the liquid crystal panel display 3, like the spatial filter 40 shown in FIG.4 and FIG.5, the direction orthogonal to the extension direction of a near edge | side is used. The number of effective kernel columns is variable, and the sensitivity has directionality.

  For example, when the spatial filter has a kernel size of 15 × 15, the effective kernel size of the spatial filter is set to vertical × horizontal = 15 × 3 or 3 as the integration target pixel moves away from the neighboring areas 33 and 37. From x15 to 15x5 or 5x15, 15x7 or 7x15, 15x9 or 9x15, 15x11 or 11x15, 15x13 or 13x15, 15x15 Can be changed through many stages.

  Even when it is not necessary to consider the above-described shading correction, like the spatial filter 40 shown in FIGS. 4 and 5, it is the same as or close to the kernel row on the outer peripheral side of the liquid crystal panel display 3 as an invalid kernel. The kernel of the number of columns may be invalidated also in the kernel column on the center side of the liquid crystal panel display 3. By doing so, the directionality (sensitivity) of the spatial filter 40 in the direction orthogonal to the left side 31 and the upper side 35 (or the right side and the lower side) of the liquid crystal panel display 3 can be made equal to the integration target pixel. it can. In other words, when it is not necessary to consider shading correction, whether or not to set an invalid kernel row on the center side of the liquid crystal panel display 3 is arbitrary.

  Further, at the four corners of the liquid crystal panel display 3, for example, as shown in FIG. 6, two neighboring areas 33 and 37 of the left side 31 and the upper side 35 overlap. Therefore, when there are integration target pixels in the region 39 where the neighboring areas 33 and 37 overlap, the kernel columns invalidated by the spatial filter 40 shown in FIGS. What is necessary is just to invalidate the kernels in the upper and lower rows. Also in this case, when it is not necessary to consider the shading correction, whether or not to set invalid kernel rows in the center side of the liquid crystal panel display 3 in the vertical direction and the horizontal direction is arbitrary.

  By the way, the display unevenness of the liquid crystal panel display 3 includes a display having a certain size in both the vertical and horizontal directions, and a linear unevenness having a small size in the vertical or horizontal direction. Compared with display unevenness that has a certain size in both the vertical and horizontal directions, the linear unevenness has a smaller range (area) of the unevenness. Therefore, when integration processing is performed, the integrated pixel value is lowered by being pulled by the pixel values of surrounding pixels. It tends to be difficult to detect as unevenness.

  Therefore, in the differentiation process in step S5 of FIG. 2 performed when detecting linear unevenness having a small size in the vertical direction or the horizontal direction, as shown in FIG. 7, the integration process in step S51 in FIG. Before executing the difference process in the same manner, the enhancement process (step S50) may be performed as a preprocess.

  In the enhancement processing in step S50, the pixel values of the vertical or horizontal linear unevenness are averaged in the extending direction of the linear unevenness to reduce the noise component. FIG. 8 shows a case where enhancement processing of linear unevenness extending in the vertical direction is performed. In this case, the spatial filter 50 (emphasis spatial filter) for emphasis processing that has directionality in the vertical direction (in which effective kernels are arranged) is used as in the case of linear unevenness. This spatial filter 50 has an n × n kernel size and only one vertical column at the center in the horizontal direction is a valid kernel (kernel value = “1”), and the others are invalid kernels (kernel value = “0”). The coefficient of the effective kernel is an inverse number (= 1 / n) of the effective kernel number n. FIG. 8 shows a case where n = 9.

  In the linear unevenness enhancement process in step S50 of FIG. 7 using the spatial filter 50, the pixel values of the linear unevenness portion are averaged with the pixel values of peripheral pixels having the same effective kernel number n in the vertical direction. As a result, the vertical boundary of the linear unevenness is clarified, and the display unevenness is easily detected.

  It should be noted that a spatial filter (not shown) for enhancement processing that has directionality in the horizontal direction may be used for the enhancement processing of the linear unevenness extending in the horizontal direction. Further, with respect to dot-like defects that are densely arranged in a dot shape, the pixel value is lowered in accordance with the pixel values of the peripheral pixels by performing this emphasis process, so that it is difficult to be erroneously detected as display unevenness.

  When performing the enhancement processing in step S50 described above, the integration processing in step S51 in FIG. 7 is performed using the pixel values of each pixel of the liquid crystal panel display 3 after the enhancement processing, and the integration pixel value is obtained. Become. During the integration process, as described with reference to FIGS. 4 and 5, a part of the kernel string of the spatial filter 50 is changed according to the positional relationship between the integration target pixel and the outer periphery of the liquid crystal panel display 3. You may make it invalidate.

  Next, in the difference process in step S53 in FIG. 3 or FIG. 7, the difference between the pixel value of each pixel of the liquid crystal panel display 3 before the integration process in step S51 and the integration pixel value after the integration process in step S51 is calculated. This is obtained and obtained as a differential pixel value of each pixel of the liquid crystal panel display 3. This completes the differentiation process in step S5 of FIG.

  Note that the two graphs shown beside step S51 in FIG. 3 show pixel value distributions in one horizontal line of the liquid crystal panel display 3 before and after the integration process in step S51. As can be seen by comparing these two graphs, the low frequency component of the change in the pixel value of the liquid crystal panel display 3 is extracted when the integration process in step S51 of FIG. 3 or FIG. 7 is performed. When an offset of the pixel value occurs over the entire pixel of the liquid crystal panel display 3, this offset is included in the extracted low frequency component.

  Also, the graph shown beside step S53 in FIG. 3 shows the pixel value distribution in one horizontal line with the liquid crystal panel display 3 after the difference processing in step S53. As can be seen from this graph, when the difference processing in step S53 of FIG. 3 and FIG. 7 described above is performed, only the high frequency component is extracted by removing the low frequency component from the pixel value change of the liquid crystal panel display 3. Even when an offset of the pixel value occurs over the entire pixel of the liquid crystal panel display 3, the offset is excluded as a low frequency component.

  Therefore, in the differentiation process in step S5 in FIG. 2, the difference between the pixel values of the target pixel and its surrounding pixels is obtained by performing the integration process and the difference process in steps S51 and S53 in FIG. 3 and FIG. Compared with performing general differentiation processing, it is possible to detect the pixel area of the liquid crystal panel display 3 having a gap in pixel value between peripheral pixels due to display unevenness with high accuracy.

  Incidentally, in the integration process in step S51 of FIG. 3, the pixel values of the respective pixels of the liquid crystal panel display 3 after the addressing process and the moire removal process performed in step S3 of FIG. 2 are integrated. On the other hand, in the integration process in step S51 of FIG. 7, the pixel values of the pixels of the liquid crystal panel display 3 after the enhancement process in step S50 are integrated. That is, even in the same integration process, the pixel value of each pixel of the liquid crystal panel display 3 used for the integration process is different between the integration process in step S51 in FIG. 3 and the integration process in step S51 in FIG.

  Therefore, when detecting unevenness having a certain size in both the vertical and horizontal directions and the linear unevenness in the vertical or horizontal direction as display unevenness, the differential processing according to the procedure of FIG. 3 and the differential processing according to the procedure of FIG. It is necessary to do each. In that case, the differentiation process according to the procedure of FIG. 3 and the differentiation process according to the procedure of FIG. 7 may be performed serially or in parallel.

  Here, the image and pixel value of the output image data of the liquid crystal panel display 3 before and after the differentiation process in step S5 of FIG. 2 will be described with reference to FIGS.

  First, it is assumed that the image display unevenness shown in FIG. 9A exists in the output image data of the liquid crystal panel display 3 before the differential processing in step S5 of FIG. The pixel value of the corresponding pixel of the liquid crystal panel display 3 at this time is a value as shown in FIG. In order to facilitate the description, in FIG. 9B, the pixel value of each pixel is represented by a value (average value = 100) obtained by normalizing the luminance indicating the shading of the screen instead of each RGB value.

  Therefore, when the differentiation process in step S5 in FIG. 2 is performed on the pixel value shown in FIG. 9B, the pixel value becomes 1000 only for pixels having a pixel value higher than the average, as shown in FIG. 10B. The pixel values of other pixels are 0. When this is represented by an image, as shown in FIG. 10A, the contrast difference between the display unevenness and the periphery thereof becomes larger than the contrast difference before the differentiation processing shown in FIG. It is clear.

  Next, in the differential threshold value determination (primary threshold value determination) process in step S7 of FIG. 2, the pixel value shown in FIG. 10B, that is, the differential pixel value of each pixel of the liquid crystal panel display 3 is compared with the unevenness determination threshold value. And binarize. The unevenness determination threshold value is a threshold value for determining whether or not a pixel in an area where display unevenness of the liquid crystal panel display 3 may occur (display unevenness generation area) is based on the differential pixel value.

  Then, as illustrated in FIG. 11, a label value is assigned to a pixel whose differential pixel value exceeds the unevenness determination threshold value, and “0” is assigned to a pixel whose differential pixel value is equal to or less than the unevenness determination threshold value. The label value is a value uniquely given to each display unevenness generation area, with an aggregate of adjacent pixels exceeding the unevenness determination threshold as one display unevenness generation area. Therefore, the same label value is assigned to the pixels in the same display unevenness occurrence area. An integer greater than or equal to “1” is used for the label value.

  Subsequently, in the unevenness intensity calculation process of step S9 in FIG. 2, the intensity of display unevenness is calculated for each display unevenness occurrence area. For the strength of display unevenness, for example, a SEMU (SEMI MURA) value standardized by Semiconductor Equipment and Materials International (SEMI, registered trademark) can be used. Here, a method for calculating the SEMU value will be described.

The calculation of the SEMU value requires the average contrast Cx of the display unevenness generation area, the area Sx of the display unevenness generation area, and the density Cjnd of the display unevenness of the human detection limit. The average contrast Cx is the luminance of the display unevenness occurrence area (the average luminance value of the pixels in the area) expressed as a percentage when the luminance of the peripheral pixels of the display unevenness generation area is 100%. The area Sx is expressed in mm ² . The density Cjnd of the display unevenness of the detection limit is expressed by a function F (Sx) of the area Sx of the display unevenness occurrence area.

  In order to obtain the above-described average contrast Cx for each occurrence area of display unevenness, it is necessary to set a foreground (Foreground: FG) and a background (Back Ground: BG) for each occurrence area. For example, in the case of the display unevenness generation area having the shape shown in FIGS. 9A and 10A, the display unevenness generation area is FG as shown in FIG. 12, and the periphery is 2 pixels away from FG. A circular area having a width of 2 pixels is BG. Therefore, an average luminance value is obtained for each pixel belonging to FG and each pixel belonging to BG, and set as an FG value and a BG value.

Next, the following formula (1)
Cx = (FG value−BG value) / BG value (1)
The average contrast Cx is obtained from the FG value and the BG value.

Moreover, using the following formula (2),
Cjnd = F (Sx) = 1.97 × (1 / Sx ^0.33 ) +0.72 (2)
The display unevenness density Cjnd of the detection limit is obtained.

And using the following formula (3),
SEMU value = | Cx | / Cjnd (3)
Obtain the SEMU value.

  As described above, since the average contrast Cx and the area Sx are used to calculate the SEMU value, it is necessary to know the exact shape of the display unevenness generation area. From this point of view, the linear unevenness that specifies the area where the display unevenness is generated by performing the emphasis process in step S50 of FIG. 7 may be excluded from the target for calculating the unevenness intensity based on the SEMU value. The reason is that, in the case of linear unevenness, the shape recognized as the display unevenness generation area may be slightly changed from the original linear unevenness shape by the preceding emphasis processing.

  Further, in the unevenness intensity threshold value determination (secondary threshold value determination) process in step S11 of FIG. 2, the unevenness intensity value (SEMU value) of the display unevenness occurrence area calculated in step S9 is compared with the intensity threshold value. The intensity threshold value is a threshold value for determining a display unevenness occurrence area that is finally detected as display unevenness based on the uneven intensity value. This intensity threshold value is set to the lowest unevenness intensity value of the display unevenness occurrence area detected as display unevenness.

  A display unevenness occurrence area where the unevenness intensity value exceeds the intensity threshold is detected as display unevenness. On the other hand, a display unevenness occurrence area where the unevenness intensity value does not exceed the intensity threshold is not detected as display unevenness. Finally, the detected display unevenness is output to the outside of the display unevenness detection device 1 as display unevenness detection result information by associating the pixel position and the unevenness intensity value in the liquid crystal panel display 3 in the result output process of step S13. To do.

  The above is the whole of the display unevenness detection processing of the liquid crystal panel display 3 performed by the display unevenness detection device 1. In the present embodiment, step S3 in the flowchart of FIG. 2 is processing corresponding to the pixel value acquisition means (pixel value acquisition step) in the claims. Further, in the present embodiment, step S5 in FIG. 2 is processing corresponding to differentiation means (differentiation step) in the claims, and step S7 in FIG. 2 is pixel value comparison means in the claims. The processing corresponds to the uneven area detecting means (area detecting step).

  Furthermore, in this embodiment, step S9 in FIG. 2 is processing corresponding to the intensity value acquisition means (intensity value acquisition step), and step S11 in FIG. 2 is intensity value comparison means in the claims. In addition, the processing corresponds to display unevenness detecting means (display unevenness detecting step).

  Further, in the present embodiment, step S50 in the flowchart of FIG. 7 is processing corresponding to emphasis means (emphasis step) in the claims. Furthermore, in this embodiment, step S51 in the flowcharts of FIGS. 3 and 7 is processing corresponding to the integration means (integration step) in the claims, and step S53 in FIGS. 3 and 7 is charged. The processing corresponds to the difference means (difference step) in the paragraph.

  The display unevenness detection result information output from the display unevenness detection device 1 is, for example, for input image data for eliminating display unevenness that is stored and held in accordance with the presence or absence of individual display unevenness and the content thereof. It can be used to generate correction data. In particular, if the display unevenness detection device 1 is provided in-line on a shipping inspection line or the like of the liquid crystal panel display 3, the display unevenness detection process can be linked with the preceding and subsequent processes.

  In that case, a controller (not shown) that manages the shipping inspection line as a whole and a unit controller (not shown) that individually manages each process on the line, the display unevenness detection device 1 provides the display unevenness detection process. However, the following procedure is executed.

  That is, as shown in FIG. 13, in step S101, the display unevenness detection apparatus 1 described with reference to the flowchart of FIG. 2 performs the display unevenness detection process, and subsequently the display unevenness detection apparatus 1 outputs the display. The presence or absence of display unevenness is detected from the unevenness detection result information (step S103). If there is no display unevenness (NO in step S103), it is determined as a non-defective product and the inspection process for the liquid crystal panel display 3 to be inspected is terminated.

  On the other hand, if there is display unevenness (YES in step S103), the number of times that the display unevenness detection device 1 outputs the inspection result information indicating that display unevenness has been detected for the liquid crystal panel display 3 is compared with the set number of times. (Step S105). If the number of outputs exceeds the set number (YES in step S105), it is determined as a defective product, and the inspection process for the liquid crystal panel display 3 to be inspected is terminated.

  On the other hand, when the output number of detection result information indicating that display unevenness has been detected does not exceed the set number of times (NO in step S105), input image data for eliminating display unevenness detected by the display unevenness detection device 1 is output. Correction data generation processing is performed (step S107).

  The correction data generation processing is executed by a unit controller included in a correction data generation unit (not shown) of the shipping inspection line. The generated correction data is newly written or overwritten and updated by a unit controller in a flash memory (not shown) of a driver circuit built in the liquid crystal panel display 3. If the correction data is appropriate, when the input image data is input to the driver circuit, the input image data is corrected by the correction data read from the flash memory so as to cancel the display unevenness. The display unevenness from the display screen 3 is eliminated.

  Then, after the correction data generation process of step S107, the process returns to step S101 again, and the display unevenness detection process by the display unevenness detection apparatus 1 described with reference to the flowchart of FIG. 2 is performed. Therefore, if display unevenness is detected by the display unevenness detection device 1 even after the display unevenness detection process and the correction data update of the liquid crystal panel display 3 are repeated a set number of times, the liquid crystal panel display 3 is determined to be defective. Will be.

  As described above, according to the display unevenness detection apparatus 1 of the present embodiment, the pixel value of each pixel of the liquid crystal panel display 3 is acquired from the output image data of the display image of the liquid crystal panel display 3 acquired from the CCD camera 5. In addition, the differential pixel value is acquired and compared with the unevenness determination threshold value. Then, an adjacent pixel group whose differential pixel value has exceeded the unevenness determination threshold is first detected as a display unevenness occurrence area.

  Subsequently, the intensity of display unevenness is calculated for each display unevenness occurrence area, the value is compared with the unevenness intensity threshold value, and when the unevenness intensity threshold is exceeded, the occurrence area is finally detected as display unevenness. I did it.

  For this reason, first, an area that is a candidate for display unevenness is primarily detected as an area where display unevenness occurs. Then, by narrowing down the detected generation area by the height of the intensity value of display unevenness, clear display unevenness that can be recognized visually is finally detected. Therefore, the display unevenness of the liquid crystal panel display 3 can be detected with higher accuracy than simply detecting the display unevenness from the distribution of the differential pixel values.

  Note that the configuration in which the differentiation process of the flowchart of FIG. 7 including the enhancement process is performed together with the differentiation process of the flowchart of FIG. 3 to detect linear unevenness may be omitted. Further, during the integration process in step S51 of FIGS. 3 and 7, as described with reference to FIGS. 4 and 5, according to the positional relationship between the integration target pixel and the outer periphery of the liquid crystal panel display 3, A configuration for invalidating a part of the kernel string of the spatial filter 50 may be omitted. Further, the intensity of display unevenness may be evaluated by a value other than the SEMU value.

  As described at the beginning, the display unevenness detection method of the present invention and the display unevenness detection apparatus to which this method is applied include a plasma panel display and an organic EL display in addition to the liquid crystal panel display 3 described in the above embodiment. It is also possible to detect display unevenness in a display device such as the above.

  The present invention can be widely applied when detecting display unevenness of a display device by image processing.

DESCRIPTION OF SYMBOLS 1 Display unevenness detection apparatus 3 Liquid crystal panel display 5 CCD camera 31 Left side 32,36 Sensitivity correction line 33,37 Neighborhood area 35 Upper side 39 Area 40,50 Spatial filter

Claims (2)

Based on output image data obtained by photographing an image displayed by the display device, a pixel value acquisition step of acquiring a pixel value of each pixel of the display device;
Each pixel value of the display device is integrated by averaging each pixel value of the display device with a pixel value of a surrounding pixel using a kernel size spatial filter corresponding to the shape and size of the display unevenness of the display device, and each pixel of the display device An integration step for obtaining an integration pixel value of
A difference step of obtaining a differential pixel value of each pixel of the display device by a difference between the pixel value and the integral pixel value in each pixel of the display device;
An uneven area detecting step for detecting an area where the display unevenness occurs in the display device based on a distribution of pixels in which the differential pixel value of the display device exceeds a predetermined unevenness determination threshold;
Calculate the average contrast of the generation area based on the pixel value or the differential pixel value of each pixel belonging to the generation area, and calculate the density of the display unevenness of the detection limit based on the area of the generation area, An intensity value acquisition step of acquiring a SEMU value indicating the display unevenness intensity of the occurrence area based on the calculated average contrast and the display unevenness density of the detection limit;
A display unevenness detecting step of detecting the occurrence area in which the SEMU value exceeds a predetermined unevenness intensity threshold as the display unevenness ,
When the acquisition target pixel of the integration pixel value belongs to any one of the vicinity areas from the outer periphery of the display device to the sensitivity correction line where shading is likely to occur, in the integration step, the vicinity area to which the acquisition target pixel belongs The pixel value of the acquisition target pixel is integrated using the spatial filter in which the sensitivity of light in the direction orthogonal to the extending direction is lowered so as to have directionality in the extending direction of the side corresponding to The display unevenness detection method of the display device. A pixel value obtaining unit that assigns each pixel value of output image data obtained by photographing an image displayed by the display device to each pixel of the display device, and obtains a pixel value of each pixel of the display device;
Each pixel value of the display device is integrated by averaging each pixel value of the display device with a pixel value of a surrounding pixel using a kernel size spatial filter corresponding to the shape and size of the display unevenness of the display device, and each pixel of the display device An integration means for obtaining an integral pixel value of
Difference means for obtaining a differential pixel value of each pixel of the display device by a difference between the pixel value and the integrated pixel value in each pixel of the display device;
Pixel value comparison means for comparing the differential pixel value of each pixel of the display device with a predetermined unevenness determination threshold;
An uneven area detecting unit that detects an occurrence area of the display unevenness in the display device based on a distribution of pixels in which the differential pixel value exceeds the unevenness determination threshold;
Calculate the average contrast of the generation area based on the pixel value or the differential pixel value of each pixel belonging to the generation area, and calculate the density of the display unevenness of the detection limit based on the area of the generation area, Intensity value acquisition means for acquiring a SEMU value indicating the display unevenness intensity of the occurrence area based on the calculated average contrast and the display unevenness density of the detection limit;
Intensity value comparing means for comparing the SEMU value with a predetermined uneven intensity threshold;
Display unevenness detecting means for detecting the occurrence area where the SEMU value exceeds a predetermined unevenness intensity threshold as the display unevenness ,
The integration means, when the acquisition target pixel of the integration pixel value belongs to any one of the vicinity areas from each outer periphery of the display device to a sensitivity correction line where shading is likely to occur, the neighboring area to which the acquisition target pixel belongs The pixel value of the acquisition target pixel is integrated using the spatial filter in which the sensitivity of the light in the direction orthogonal to the extending direction is lowered so as to have directionality in the extending direction of the side corresponding to
Display unevenness detection device for display device.