JP2861944B2 - Display panel defect inspection apparatus and method - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an apparatus and a method for inspecting a display panel of a display panel such as a liquid crystal display, and more particularly to a defect inspection apparatus for a display panel which automatically inspects a display uneven defect having a small difference in density and brightness. And methods.

[0002]

2. Description of the Related Art FIG. 7 is a flowchart showing a conventional display panel defect inspection method. The method shown in FIG. 7 includes a step S1 for moving and positioning the camera, a step S2 for displaying a black image on the display panel, a step S3 for capturing a black image, and a step S3 for displaying a white image on the display panel.
4, step S5 for capturing a white image, step S6 for obtaining a change in the gray value of the image and dividing the area, and when the test object is a color liquid crystal, the binarization processing is performed after the cancellation processing for crosstalk is performed. Steps S7 to S9 for performing the inspection for the number of deducted defective pixels, Steps S10 to S13 for performing the black image display and performing the bright point defect inspection, and Step S14 for displaying the gray image by the same processing and inspecting the defective pixels. To S17 and step S18 for inspecting line defects (for example, see Japanese Patent Laid-Open No. 5-219).
535).

Although this display panel defect inspection method has a linear defect detection function, it is sometimes impossible to detect an irregular defect because the shape of the defect is unspecified. In addition, half-bright point defect detection is performed by binarized image processing of a gray image, but a non-uniform defect, which is a collection of half-bright point defects, is not an absolute value of display luminance but a luminance difference between a background and a defective portion. In order for a person to recognize, a technique for changing a threshold value according to the inspection area is required (for example, see Japanese Patent Application Laid-Open No. 4-208834).

[0004] For the above reasons, the inspection for uneven defects has been performed visually.

[0005]

In the conventional defect inspection method for a display panel, it is difficult to set a visual field required for the inspection because the size and shape of the irregular defect are unspecified. The difference in luminance between the defective portion and the background, which is a normal portion, was so small that it was difficult to determine a threshold value.

[0006]

A display panel defect inspection apparatus according to the present invention comprises a photoelectric conversion input unit for converting an image of a display panel, which is an object to be inspected, into original image data; A hierarchical processing unit for creating a plurality of image data forming a hierarchy having the original image data at the lowest level as inspection area data so that each area divided in a shape becomes a pixel of the image data in the upper layer, A two-dimensional statistical data processing unit that performs texture analysis processing on the area data to create two-dimensional statistical data, and a non-uniformity detecting unit that detects a non-uniform defect in the image of the display panel that is the inspection object from the two-dimensional statistical data. Preferably, the hierarchization processing unit inspects each of the areas obtained by dividing the image data of the name hierarchy into a grid pattern as pixels of the image data of the upper hierarchy, except for the top image data. Created as area data, the two-dimensional statistical data is a set of luminance combinations of adjacent pixels for all pixels on the inspection area data, and the adjacent pixels of all the pixels on the inspection area data without uneven defects are determined in advance. A set of combinations of luminances is obtained as a normal luminance range, and the unevenness detection unit detects unevenness if the combination of luminances of adjacent pixels on the inspection area data obtained from the display panel as the inspection target is outside the normal luminance range. It is possible to determine that the defect is a defect, and to further include an inspection result output unit that displays the position of the non-uniform defect detected by the non-uniformity detection unit on the original image data.

In the defect inspection method for a display panel according to the present invention, each area obtained by converting a liquid crystal panel to be inspected into original image data using a photoelectric conversion input unit and dividing the image data of one layer into a lattice. A plurality of image data forming a hierarchy having the original image data at the lowest level is created as inspection area data so that the pixels become pixels of the image data at a level higher by one hierarchy, and a texture solution process is performed on the inspection area data. An uneven defect in an image on a display panel as the inspection object is detected based on the obtained two-dimensional statistical data.

[0008]

Next, embodiments of the present invention will be described in detail with reference to the drawings.

FIG. 1 is a block diagram of a display panel defect inspection apparatus according to an embodiment of the present invention.

The defect inspection apparatus for a display panel shown in FIG.
As means for inputting an image of a target object, a photoelectric conversion input unit 1 for converting and inputting an image into original image data 2 and an image memory 3 for storing the original image data 2 are provided.

FIG. 2 shows an example of data 2 including an uneven defect. An image of a liquid crystal display having 640 × 400 pixels, which is an object, is input using a 4096-pixel one-dimensional CCD camera as a photoelectric conversion input unit 1. As the unevenness defect 11 on the display screen, there are a small-size linear unevenness defect and a large planar unevenness defect. As for the luminance of the portion of the uneven defect 11, when the black-and-white difference is 256 gradations, the luminance difference from the background is about 5 gradations.

As means for creating inspection area data 5 by adjusting the inspection field for extracting uneven defects 11 of an unspecified size, a hierarchical processing unit 4 is provided. The hierarchization processing unit 4 performs region division using the original image data 2 as the lowest micro hierarchy, and integrates the regions hierarchically to create macro regions and hierarchies. The number of hierarchies is created in accordance with the number of display surface pixels to be inspected, and the number n of hierarchies increases as the number of pixels of the original image data 2 increases.

FIG. 3 shows an example of a process in which the hierarchization processing section 4 hierarchizes into two hierarchies. In FIG. 3, the micro hierarchy 12 is divided into P × Q areas in a grid pattern. Each divided area is composed of M × N pixels arranged in a grid. Each of the divided areas is regarded as one pixel and P
A macro hierarchy 13 consisting of × Q pixels is created. The luminance of the pixel of the macro hierarchy 13 is an average of the luminance of M × N pixels constituting the divided region of the micro hierarchy 12 corresponding to the pixel.

Except for the lowest macro layer, the processing performed by the two-dimensional statistical data processing unit 6 and thereafter described below is performed for each of the divided areas except for the lowest macro layer for each layer created by the layering processing unit 4. In the micro hierarchy shown in FIG. 3, the processing of the two-dimensional statistical data processing unit 16 and below is performed for each of the P × Q divided areas.
The hierarchical processing unit 4 outputs the image data of the hierarchically divided area as inspection area data 5.

FIG. 4 is a view for explaining the inspection area data 5.
The horizontal axis represents the X direction, and the vertical axis represents the Y direction.
Y). As means for creating two-dimensional statistical data 7 which is a type of texture analysis, a two-dimensional statistical data processing unit 6 is provided. Creation of a combination of luminance values of two pixels having a fixed inspection area distance is processed for all pixels in the inspection area data 5.

With respect to image data having uniform luminance of about 200 gradations without unevenness, adjacent pixels, ie, adjacent in the X direction as (Xa, Ya) and (Xb, Yb) shown in FIG. FIG. 5 shows an example of two-dimensional statistical data in the case where the luminance values of pixels and pixels adjacent in the Y direction are used, assuming that one of the adjacent pixels has luminance a and the other has luminance b. Shown in However, calculation processing is performed as normalization so that the average value of the luminance distribution in the inspection area data becomes 128, and the existing combination of luminance is indicated by a symbol *. In the entire inspection area data, there is a luminance distribution of five or more gradations that cannot be recognized by the operator. However, since the luminance difference between adjacent pixels is within five gradations, luminance a = luminance b, and a straight line in the vicinity. There are brightness combinations.

FIG. 6 shows an example of two-dimensional statistical data of image data containing an uneven defect. In this example, since the combination of the luminance of the periphery of the linear uneven defect or the elliptical defect and the background is a combination of the low luminance and the average luminance, it is recorded in the ◎ portion. Is recorded in the ● part because the luminance combination is lower than the average.

As means for extracting and judging the combination of luminance of the uneven defect portion by processing the two-dimensional statistical data,
An unevenness detection unit 8 is provided. A case where a set of luminance combinations of two-dimensional statistical data created from inspection area data without uneven defects is defined as a normal luminance range, and a luminance combination not within the normal luminance range exists in the two-dimensional statistical data of the inspection area data 5 Is extracted as a non-uniform defect, and the combination of the luminance is set as the inspection result 9. Note that the inspection area data having no unevenness defect is created using a non-defective liquid crystal panel and stored in the unevenness detection unit 8 in advance.

As an example of unevenness detection, when the data shown in FIG. 6 is calculated using the two-dimensional statistical data without unevenness defects shown in FIG. 5 as a normal luminance range, the results are as follows. The combination of the luminances of ● inside the uneven defect is detected as an uneven defect. An inspection result output unit 10 is provided as a means for displaying and recording the inspection result of the uneven defect.
The position on the original image data 2 where the combination of the brightness of the uneven defect exists in the inspection result 9 detected by the unevenness detecting unit 8 is obtained,
The position is displayed on the display.

In the example shown in FIG. 2, the linear unevenness defect where the micro hierarchy is detected as the inspection area data 5 is a collection of points,
The planar unevenness defect in which the macro hierarchy is detected as the inspection area data 5 is indicated by a square.

In the original image data 2, the difference in luminance between adjacent pixels in all pixels is a small difference of 5 gradations or less, but the greater the distance between the pixels, the greater the difference in luminance. Such unevenness in which the boundary is not clear is not detected when the micro hierarchy created by the hierarchical processing unit 4 is used as the inspection area data 5, but is detected because the difference in luminance between adjacent pixels increases when the macro hierarchy is used. Become like

[0022]

As described above, the defect inspection apparatus and method for a display panel according to the present invention creates two-dimensional statistical data by texture analysis processing on image data on which hierarchical processing has been performed, thereby forming a lower hierarchical layer. Image data of various sizes and shapes including small linear unevenness defects can be detected, and unevenness defects whose boundaries are not clear can be detected by image data of a higher hierarchy.

[Brief description of the drawings]

FIG. 1 is a block diagram of a display panel defect inspection apparatus according to an embodiment of the present invention.

FIG. 2 is a diagram showing original image data 2 in FIG.

FIG. 3 is a diagram illustrating a hierarchical processing performed by a hierarchical processing unit 4 in FIG. 1;

FIG. 4 is a diagram showing inspection area data 5 in FIG.

FIG. 5 is a diagram showing two-dimensional statistical data 7 in FIG. 1 for image data without uneven defects;

FIG. 6 is a diagram illustrating two-dimensional statistical data 7 in FIG. 1 corresponding to image data including an uneven defect;

FIG. 7 is a flowchart showing a conventional display panel defect inspection method.

[Description of Signs] 1 photoelectric conversion input unit 3 image memory 4 hierarchical processing unit 6 two-dimensional data processing unit 8 unevenness detection unit 10 inspection result output unit 11 unevenness defect

Continuation of the front page (58) Field surveyed (Int.Cl. ⁶ , DB name) G01M 11/00 G01N 21/88 G06T 7/00

Claims (6)

(57) [Claims] 1. A photoelectric conversion input unit for converting an image on a display panel, which is an inspection object, into original image data, and each area obtained by dividing one layer of image data into a grid pattern has pixels of one layer higher image data A hierarchical processing unit that creates a plurality of image data forming a hierarchy with the original image data as the lowest order as inspection area data, and performs a texture analysis process on the inspection area data to generate two-dimensional statistical data. A defect inspection apparatus for a display panel, comprising: a two-dimensional statistical data processing unit to be created; and a non-uniformity detection unit that detects a non-uniform defect in an image of a display panel as the inspection object from the two-dimensional statistical data. 2. The hierarchization processing unit creates, as inspection area data, areas obtained by dividing the image data of each hierarchy into pixels in the form of pixels of the image data of the upper hierarchy, excluding the top image data. 2. The defect inspection apparatus for a display panel according to claim 1. 3. The defect inspection apparatus for a display panel according to claim 1, wherein the two-dimensional statistical data comprises a set of combinations of luminances of adjacent pixels for all pixels on the inspection area data. 4. A set of combinations of luminances of adjacent pixels for all pixels on the inspection area data having no irregularity defect is determined in advance as a normal luminance range. 4. The display panel defect inspection apparatus according to claim 3, wherein if the combination of luminances of adjacent pixels on the obtained inspection area data is outside the normal luminance range, the defect is determined to be an uneven defect. 5. The defect detection device for a display panel according to claim 1, further comprising an inspection result output unit that indicates a position on the original image data of the uneven defect detected by the uneven detection unit. 6. A liquid crystal panel to be inspected is converted into original image data by using a photoelectric conversion input section, and each area obtained by dividing one layer of image data into a lattice is a pixel of one layer higher image data. A plurality of image data forming a hierarchy having the original image data at the lowest level as inspection area data, and performing a texture analysis process on the inspection area data. A defect inspection method for a display panel, comprising detecting an uneven defect in an image of a display panel as an object.