Display defect detection system and display defect detection method

Abstract

A display defect detection system and method are disclosed. The display defect detection system includes: a preprocessing circuit that receives a captured image of a test pattern displayed on a display panel as a panel image including diagonal unevenness, and preprocesses the panel image to output a preprocessed image; and an unevenness detection circuit that calculates a final feature value of the diagonal unevenness based on the edge mapping of the preprocessed image by multiplying a first feature value reflecting the brightness difference of the diagonal unevenness with a second feature value reflecting the shape ratio of the diagonal unevenness, and detects the display position of the diagonal unevenness based on the final feature value of the diagonal unevenness.

Description

[0001] Cross-reference to related applications

[0002] This application claims the benefit of Korean Patent Application No. 10-2021-0178480, filed on December 14, 2021, which is incorporated herein by reference as fully set forth herein. Technical Field

[0003] This disclosure relates to display defect detection systems and methods. Background Technology

[0004] Display defects are defects that occur during the manufacturing process of display panels. Their causes include malfunctions in processing facilities, worker errors, minor imperfections caused by dust or particles, and defects in the films used. When display defects occur, the screen characteristics of the display panel are uneven and exhibit a mura state. Various types, sizes, and levels of mura exist, and in particular, mura has various types, such as circular, linear, informal, and repetitive coating types.

[0005] There are methods where workers directly inspect display panels for unevenness using their eyes. However, in such methods, workers' eyes may become fatigued, the detection rate of minor defects may decrease, and the inspection results may change due to workers' subjective decisions, making it difficult to expect consistently high levels of inspection results.

[0006] In non-uniformity detection methods, the difficulty level is high because the boundaries, brightness variations, and contrast of the captured image are not clearly shown. Conventional algorithms are based on contrast ratio, standard deviation (STD), and amplitude, but they struggle to improve the detection accuracy of diagonal non-uniformity. Diagonal non-uniformity exhibits various line directions and lengths, thus limiting the accuracy of related detection techniques. Summary of the Invention

[0007] To overcome the aforementioned problems in related technologies, this disclosure provides a display defect detection system and a display defect detection method that can improve the detection accuracy of diagonal unevenness.

[0008] To achieve these objectives and other advantages and in accordance with the purposes of this disclosure, as embodied and broadly described herein, a display defect detection system includes: a preprocessing circuit that receives a captured image of a test pattern displayed by a display panel as a panel image including diagonal unevenness, and preprocesses the panel image to output a preprocessed image; and an unevenness detection circuit that, based on the edge mapping of the preprocessed image, calculates a final characteristic value of the diagonal unevenness by multiplying a first characteristic value reflecting the brightness difference of the diagonal unevenness by a second characteristic value reflecting the shape ratio of the diagonal unevenness, and detects the display position of the diagonal unevenness based on the final characteristic value of the diagonal unevenness.

[0009] In another aspect of this disclosure, a display defect detection method includes: receiving a captured image of a test pattern displayed by a display panel as a panel image including diagonal unevenness, and preprocessing the panel image to output a preprocessed image; calculating a final feature value of diagonal unevenness by multiplying a first feature value reflecting the brightness difference of the diagonal unevenness with a second feature value reflecting the shape ratio of the diagonal unevenness based on the edge mapping of the preprocessed image; and detecting the display position of the diagonal unevenness based on the final feature value of the diagonal unevenness. Attached Figure Description

[0010] The accompanying drawings are included to provide a further understanding of this disclosure and are incorporated in and constitute a part of this application. The drawings illustrate embodiments of the disclosure and, together with the specification, serve to explain the principles of the disclosure. In the drawings:

[0011] Figure 1 This is a diagram illustrating a display defect detection system according to an embodiment of the present disclosure;

[0012] Figure 2 It is shown Figure 1 A diagram of the defect detection circuit.

[0013] Figure 3 It is shown by Figure 2 A diagram showing the preprocessing process performed by the preprocessing circuit;

[0014] Figure 4 and Figure 5 It is shown by Figure 1 A diagram showing the uneven detection process performed by the display defect detection circuit;

[0015] Figure 6 This is a diagram illustrating an example of rotating the edge-mapped image of each of the diagonal unevenness candidates n times based on the number of angles "n"; and

[0016] Figure 7This is a diagram illustrating an example of calculating the maximum ratio value among the projection ratio values ​​between the x-axis and y-axis directions for n angles as the shape ratio feature value of the corresponding diagonal unevenness candidate. Detailed Implementation

[0017] In the following description, the present disclosure will be given more fully with reference to the accompanying drawings, in which exemplary embodiments of the disclosure are illustrated. However, the present disclosure may be implemented in many different forms and should not be construed as limited to the embodiments set forth herein; rather, these embodiments are provided so that the disclosure will be comprehensive and complete and will fully convey the concept of the disclosure to those skilled in the art.

[0018] The advantages and features of this disclosure, and its implementation methods, will be illustrated by the following description of embodiments with reference to the accompanying drawings. However, this disclosure may be implemented in various forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of this disclosure to those skilled in the art. Furthermore, this disclosure is defined only by the scope of the claims.

[0019] The shapes, dimensions, ratios, angles, numbers, etc., disclosed in the accompanying drawings used to describe various embodiments of this disclosure are merely exemplary and are not limited thereto. Throughout the document, the same reference numerals refer to the same elements. As used herein, unless the term "only" is used, the terms "comprising," "having," "including," etc., indicate that additional parts may be added. As used herein, unless the context clearly indicates otherwise, the singular forms "a," "an," and "the" are intended to include the plural forms as well.

[0020] Even without explicit statement, elements in various embodiments of this disclosure will be construed as including tolerances.

[0021] When describing positional relationships, for example, when the positional relationship between two parts is described as "on top of", "above", "below", and "near", one or more other parts may be placed between the two parts, unless "exactly" or "directly" is used.

[0022] It should be understood that although the terms "first," "second," etc., may be used herein to describe various elements, these elements should not be limited by these terms. These terms are used only to distinguish one element from another. For example, without departing from the scope of this disclosure, a first element may be referred to as a second element, and similarly, a second element may be referred to as a first element.

[0023] In the following description, detailed descriptions of known functions or configurations will be omitted where it is determined that such detailed descriptions would unnecessarily obscure the essential points of this disclosure. Hereinafter, embodiments of this disclosure will be described in detail with reference to the accompanying drawings.

[0024] Figure 1 This is a diagram illustrating a display defect detection system 100 according to an embodiment of the present disclosure.

[0025] Reference Figure 1 The display defect detection system 100 may include a display device, a brightness measuring instrument (CMR), and a display defect detection circuit 40.

[0026] The display device can be an electroluminescent display device, but is not limited to this, and can be applied to various types of display devices. For example, the display device can be implemented in various types, such as liquid crystal display devices, electrophoretic display devices, electrowetting display devices, and quantum dot display devices. In this embodiment, the electroluminescent display device will be mainly described below.

[0027] The display device may include a controller 10, a panel driver 20, and a display panel 30.

[0028] A screen comprising multiple pixel lines can be disposed in the display panel 30, and multiple pixels P can be included in each of the multiple pixel lines. Here, "pixel line" can refer to a group of signal lines and pixels P that are adjacent to each other in one direction. The signal lines can include multiple data lines DL for supplying a data voltage Vdata to the pixel P, multiple reference voltage lines RL for supplying a reference voltage Vref to the pixel P, multiple gate lines GL for supplying a scan signal to the pixel P, and multiple high-level power lines for supplying a high-level pixel voltage EVDD to the pixel P.

[0029] The pixels P of the display panel 30 can be arranged in a matrix to configure a pixel array and provide a screen for displaying images. Each pixel P can be connected to one of the data lines DL, one of the reference voltage lines RL, one of the high-level power lines, and one of the gate lines GL. Each pixel P can also be supplied with a low-level pixel voltage EVSS from the panel driver 20.

[0030] Each pixel P may include, but is not limited to, a light-emitting device EL, a driving thin-film transistor (TFT) DT, multiple switching TFTs ST1 and ST2, and a storage capacitor Cst. The driving TFT DT and the switching TFTs ST1 and ST2 may each be implemented using NMOS transistors, but are not limited to.

[0031] The light-emitting device (EL) can be a light-emitting device that emits light with an intensity corresponding to the pixel current applied from the driving TFT DT. The EL can be implemented using an organic light-emitting diode (OLED) including an organic light-emitting layer, or using an inorganic light-emitting diode (LED) including an inorganic light-emitting layer. The anode of the EL can be connected to the second node N2, and its cathode can be connected to the input terminal of the low-level pixel voltage EVSS.

[0032] The driving TFT DT can be a driving element that generates pixel current based on its gate-source voltage. The first electrode of the driving TFT DT can be connected to a first node N1, and the first electrode of the driving TFT DT can be connected to the input terminal of the high-level pixel voltage EVDD via a high-level power line, and the second electrode of the driving TFT DT can be connected to a second node N2.

[0033] Switching TFTs (e.g., a first switching TFT and a second switching TFT) ST1 and ST2 can be switching elements that set the gate-source voltage of the driving TFT DT and connect the second electrode of the driving TFT DT to the reference voltage line RL.

[0034] A first switch TFT ST1 can be connected between the data line DL and the first node N1, and can be turned on based on the scan signal SCAN from the gate line GL. The first switch TFT ST1 can be turned on during image driving programming. When the first switch TFT ST1 is turned on, a data voltage Vdata can be applied to the first node N1. The second electrode of the first switch TFT ST1 can be connected to the gate line GL, the first electrode of the first switch TFT ST1 can be connected to the data line DL, and the second electrode of the first switch TFT ST1 can be connected to the first node N1.

[0035] The second switch TFT ST2 can be connected between the reference voltage line RL and the second node N2, and can be turned on based on the scan signal SCAN from the gate line GL. The second switch TFT ST2 can be turned on during image driving programming, and a reference voltage Vref can be applied to the second node N2. The second electrode of the second switch TFT ST2 can be connected to the gate line GL, the first electrode of the second switch TFT ST2 can be connected to the reference voltage line RL, and the second electrode of the second switch TFT ST2 can be connected to the second node N2.

[0036] The storage capacitor Cst can be connected between the first node N1 and the second node N2, and can maintain the gate-source voltage of the driving TFT DT during a specific time period.

[0037] The controller 10 may include a timing controller and a data modulator.

[0038] The timing controller can refer to timing signals input from the host system (e.g., vertical synchronization signal Vsync, horizontal synchronization signal Hsync, dot clock signal DCLK, and data enable signal DE) to control the operating timing of the control driver 20.

[0039] The data modulator can modulate the video data DATA input from the host system with reference to a predetermined non-uniformity compensation table. Compensation values ​​corresponding to the location information of target non-uniformities detected by the display defect detection circuit 40 can be stored in the non-uniformity compensation table in grayscale. The data modulator can transmit the modulated image data MDATA to the panel driver 20 via an internal interface.

[0040] The panel driver 20 can drive the pixels P contained in the screen of the display panel 30. The panel driver 20 may include a data driver that drives the data line DL connected to the pixel P, a gate driver that drives the gate line GL connected to the pixel P, and a power driver that drives the reference voltage line RL connected to the pixel P and the high-level power line connected to the pixel P.

[0041] The data driver can convert image data MDATA obtained through modulation based on controller 10 into a data voltage Vdata, and can supply this data voltage Vdata to the screen. The data voltage Vdata can be supplied to the screen via data line DL.

[0042] The gate driver can generate a scan signal SCAN based on the control of the controller 10, and can provide the scan signal SCAN to the screen based on the application timing of the data voltage Vdata. The scan signal SCAN can be supplied to the screen through the gate line GL, and therefore, the pixel line to which the data voltage Vdata will be applied can be selected. The gate driver can be formed directly in the non-display area outside the screen of the display panel 30.

[0043] The luminance measuring instrument (CMR) can capture a test pattern displayed on the display panel 30 to generate a panel image including diagonal unevenness. The luminance measuring instrument (CMR) can be a camera-based CMR, but is not limited to this.

[0044] The display defect detection circuit 40 can preprocess and edge-process the panel image input from the luminance meter (CMR), and can calculate the final characteristic value of the diagonal unevenness by multiplying a first characteristic value reflecting the luminance difference of the diagonal unevenness with a second characteristic value reflecting the shape ratio of the diagonal unevenness. The display position of the diagonal unevenness can be detected based on this final characteristic value. The display position of the diagonal unevenness detected by the display defect detection circuit 40 can be based on a constructed unevenness compensation table. In the unevenness compensation table, the position information of the diagonal unevenness can be read address information used to read the compensation value.

[0045] Figure 2 It is shown Figure 1 The diagram shows the defect detection circuit. Figure 3 It is shown by Figure 2 The diagram shows the preprocessing process performed by the preprocessing circuit. Figure 4 and Figure 5 It is shown by Figure 1 The diagram shows the uneven detection process performed by the display defect detection circuit. Figure 6 This is a diagram illustrating an example of rotating the edge-mapped image of each of the diagonal unevenness candidates n times based on the number of angles "n". Figure 7 This is a diagram illustrating an example of calculating the maximum ratio value among the projection ratio values ​​between the x-axis and y-axis directions for n angles as the shape ratio feature value of the corresponding diagonal unevenness candidate.

[0046] Reference Figure 2 The defect detection circuit 40 may include a preprocessing circuit 50 and an unevenness detection circuit 60.

[0047] The preprocessing circuit 50 can receive a captured image of the test pattern displayed on the display panel as a panel image including diagonal unevenness, and can preprocess the panel image to output a preprocessed image.

[0048] The preprocessing process performed by the preprocessing circuit 50 may include: such as Figure 3 The preprocessing circuit 50 includes the following steps: receiving a panel image (S501); cropping the panel image to extract a region of interest (ROI) including diagonal unevenness and changing the size of the ROI to a predetermined size (S502); normalizing the color or brightness of the ROI (S503); and removing noise components from the ROI that hinder the detection of diagonal unevenness (S504). To remove noise components from the ROI, the preprocessing circuit 50 may include a filtering-based algorithm (Gaussian filtering, bilateral filtering, etc.).

[0049] Performance biases and noise components in panel images, based on the shooting environment or panel characteristics, may reduce the accuracy of unevenness detection, and such side effects can be reduced through preprocessing.

[0050] The unevenness detection circuit 60 may include an edge detector 60A, a block calculator 60B, an unevenness candidate selector 60C, a shape ratio calculator 60D, and a final feature value calculator 60E. (See below for reference.) Figures 4 to 7 Detailed description of the unevenness detection circuit 60.

[0051] The edge detector 60A can generate an edge-mapped image that includes edge information about the preprocessed image (see...). Figure 4 (The edge detection image). For this purpose, the edge detector 60A may include a Sobel filter. The edge detector 60A can detect edge features in the x-axis and y-axis directions by using the Sobel filter, and can perform a weighted average of the edge features to generate an edge mapping image. The edge mapping image generated by the Sobel filter can be a binary image.

[0052] The Block Calculator 60B can divide an edge-mapped image into blocks and calculate the block edge feature values ​​corresponding to each edge-mapped block image (see [link]). Figure 4 (Block-generated image). When performing arithmetic operations on the edge-mapped image on a block-by-block basis, the computation speed can be very fast, and block edge feature values ​​with enhanced local properties can be obtained. In order not to miss the unevenness located in the boundary regions between blocks, the block calculator 60B can perform arithmetic operations on partially overlapping adjacent blocks.

[0053] Uneven candidate selector 60C can select edge-mapped block images whose block edge feature values ​​are greater than or equal to a reference value from edge-mapped block images as uneven candidate images (see [link]). Figure 5 The uneven candidate selector 60C can select edge-mapped block images with block edge feature values ​​that are among the top 10% from the edge-mapped block images as uneven candidate images, and can filter and remove edge-mapped block images with block edge feature values ​​lower than the top 10%, thereby improving the accuracy of unevenness detection. (See reference...) Figure 5 The uneven candidate selection image can detect and filter out edge-mapped blocks whose edge feature values ​​are below the top 10%. The uneven candidate image can be a binary image.

[0054] Shape Ratio Calculator 60D can, for example Figure 6 and Figure 7As in the example, while rotating the candidate image of unevenness based on the number of angles "n", the diagonal unevenness included in the candidate image is projected in the x-axis and y-axis directions at each angle, and the projection ratio value corresponding to the angle with the largest projection ratio value between the x-axis and y-axis directions can be calculated as the shape ratio feature value of the diagonal unevenness (see [reference]). Figure 5 (Image generated by the ratio of uneven shapes).

[0055] like Figure 7 As shown, the projection ratio value between the x-axis and y-axis directions can include: a first ratio value obtained by dividing the x-axis projection value (proj_x) by the y-axis projection value (proj_y); and a second ratio value obtained by dividing the y-axis projection value (proj_y) by the x-axis projection value (proj_x). The shape ratio calculator 60D can calculate n first ratio values ​​and n second ratio values ​​for all angles 1 to n, and can calculate the largest ratio value among the n first ratio values ​​and n second ratio values ​​as the shape ratio feature value for diagonal unevenness.

[0056] The final feature value calculator 60E can multiply a first feature value, which is a block edge feature value and a second feature value, which is a shape ratio feature value and a diagonal unevenness feature value, to calculate a luminance-shape ratio feature value that indicates the final feature value of the diagonal unevenness (see [reference]). Figure 5 Edge-ratio feature extraction image).

[0057] In addition, Figure 5 The image shows that the uneven shape ratio generated image is similar to the edge-ratio feature extraction image, but the shape ratio feature values ​​of the diagonal unevenness included in the uneven shape ratio generated image may be different from the brightness-shape ratio feature values ​​included in the edge-ratio feature extraction image.

[0058] As described above, the unevenness detection circuit 60 according to this embodiment can also consider the tilt of the diagonal unevenness and the horizontal-vertical ratio of the diagonal unevenness during the unevenness detection process, and therefore, it can more accurately detect diagonal unevenness that cannot be detected by conventional unevenness detection algorithms.

[0059] The unevenness detection circuit 60 according to this embodiment can accurately detect diagonal unevenness, which is different from point unevenness, band unevenness, informal unevenness and general linear unevenness.

[0060] The display defect detection method according to the embodiments of this disclosure can be based on the above-described display defect detection system.

[0061] A display defect detection method according to embodiments of the present disclosure may include the following steps: receiving a captured image of a test pattern displayed on a display panel as a panel image including diagonal unevenness, and preprocessing the panel image to output a preprocessed image; calculating a final feature value of diagonal unevenness by multiplying a first feature value reflecting the brightness difference of the diagonal unevenness with a second feature value reflecting the shape ratio of the diagonal unevenness based on the edge mapping of the preprocessed image; and detecting the display position of the diagonal unevenness based on the final feature value of the diagonal unevenness.

[0062] In a display defect detection method, the process of calculating the final feature value of diagonal unevenness may include the following steps: generating an edge-mapped image that includes edge information about a preprocessed image; dividing the edge-mapped image into blocks and calculating a block edge feature value corresponding to each edge-mapped block image; selecting edge-mapped block images from the edge-mapped block images whose block edge feature values ​​are greater than or equal to a reference value as unevenness candidate images; while rotating the unevenness candidate images at multiple angles, projecting the diagonal unevenness included in the unevenness candidate images in the x-axis and y-axis directions for each of the multiple angles, and calculating the projection ratio value corresponding to the angle with the largest projection ratio value between the x-axis and y-axis directions as the shape ratio feature value of the diagonal unevenness; and multiplying the block edge feature value of the unevenness candidate image indicating a first feature value with the shape ratio feature value of the diagonal unevenness indicating a second feature value to calculate the brightness-shape ratio feature value indicating the final feature value of the diagonal unevenness.

[0063] In embodiments of this disclosure, the slope and lateral-to-vertical ratio of diagonal unevenness can also be considered during the unevenness detection process, thus enabling more accurate detection of diagonal unevenness that cannot be detected by conventional unevenness detection algorithms. Furthermore, embodiments of this disclosure can accurately detect diagonal unevenness that differs from point-like, band-like, informal, and general linear unevenness.

[0064] The effects of this disclosure are not limited to the examples above, and various other effects may be included in the specification.

[0065] Although the present disclosure has been specifically shown and described with reference to exemplary embodiments thereof, those skilled in the art will understand that various changes in form and detail may be made therein without departing from the spirit and scope of the disclosure as defined by the appended claims.

Claims

1. A display defect detection system, comprising: a pre-processing circuit that receives a captured image of a test pattern displayed by a display panel as a panel image including a diagonal unevenness, and pre-processes the panel image to output a pre-processed image; and an unevenness detection circuit that calculates a final feature value of the diagonal unevenness by multiplying a first feature value reflecting a luminance difference of the diagonal unevenness with a second feature value reflecting a shape ratio of the diagonal unevenness based on an edge map of the pre-processed image, and detects a display position of the diagonal unevenness based on the final feature value of the diagonal unevenness.

2. The display defect detection system of claim 1, wherein, The pre-processing procedure performed by the pre-processing circuit includes: a procedure of cropping the panel image to extract a region of interest (ROI) including the diagonal unevenness and changing a size of the ROI to a predetermined size; a procedure of normalizing a color sense or a luminance of the ROI; and a procedure of removing a noise component that hinders detection of the diagonal unevenness from the normalized ROI.

3. The display defect detection system of claim 1, wherein, The unevenness detection circuit includes: an edge detector that generates an edge map image including edge information about the pre-processed image; a block calculator that divides the edge map image in units of blocks, and calculates a block edge feature value corresponding to each of the edge map block images; an unevenness candidate selector that selects, from the edge map block images, an edge map block image in which the block edge feature value is greater than or equal to a reference value as an unevenness candidate image; a shape ratio calculator that, while rotating the unevenness candidate image at a plurality of angles, projects a diagonal unevenness included in the unevenness candidate image in an x-axis direction and a y-axis direction for each of the plurality of angles, and calculates a projection ratio value corresponding to an angle in which a projection ratio value between the x-axis direction and the y-axis direction is greatest as a shape ratio feature value of the diagonal unevenness; and a final feature value calculator that multiplies the block edge feature value of the unevenness candidate image indicating the first feature value with the shape ratio feature value of the diagonal unevenness indicating the second feature value to calculate a luminance-shape ratio feature value indicating a final feature value of the diagonal unevenness.

4. The display defect detection system of claim 3, wherein, Each of the edge map image and the unevenness candidate image is a binary image.

5. The display defect detection system of claim 3, wherein, The unevenness candidate selector selects, from the edge map block images, an edge map block image in which the block edge feature value is a top 10% as the unevenness candidate image, and filters and removes an edge map block image in which the block edge feature value is lower than the top 10%.

6. The display defect detection system of claim 3, wherein, The projection ratio value between the x-axis direction and the y-axis direction includes: a first ratio value obtained by dividing an x-axis direction projection value by a y-axis direction projection value; and a second ratio value obtained by dividing the y-axis direction projection value by the x-axis direction projection value. 2.A display defect detection method, comprising: receiving a captured image of a test pattern displayed by a display panel as a panel image including a diagonal unevenness; pre-processing the panel image to output a pre-processed image; calculating a final feature value of the diagonal unevenness by multiplying a first feature value reflecting a luminance difference of the diagonal unevenness with a second feature value reflecting a shape ratio of the diagonal unevenness based on an edge map of the pre-processed image; and detecting a display position of the diagonal unevenness based on the final feature value of the diagonal unevenness. The pre-processing procedure includes: cropping the panel image to extract a region of interest (ROI) including the diagonal unevenness and changing a size of the ROI to a predetermined size; normalizing a color sense or a luminance of the ROI; and removing a noise component that hinders detection of the diagonal unevenness from the normalized ROI. The calculating of the final feature value includes: generating an edge map image including edge information about the pre-processed image; dividing the edge map image in units of blocks, and calculating a block edge feature value corresponding to each of the edge map block images; selecting, from the edge map block images, an edge map block image in which the block edge feature value is greater than or equal to a reference value as an unevenness candidate image; while rotating the unevenness candidate image at a plurality of angles, projecting a diagonal unevenness included in the unevenness candidate image in an x-axis direction and a y-axis direction for each of the plurality of angles, and calculating a projection ratio value corresponding to an angle in which a projection ratio value between the x-axis direction and the y-axis direction is greatest as a shape ratio feature value of the diagonal unevenness; and multiplying the block edge feature value of the unevenness candidate image indicating the first feature value with the shape ratio feature value of the diagonal unevenness indicating the second feature value to calculate a luminance-shape ratio feature value indicating a final feature value of the diagonal unevenness. Each of the edge map image and the unevenness candidate image is a binary image. The selecting of the unevenness candidate image includes: selecting, from the edge map block images, an edge map block image in which the block edge feature value is a top 10% as the unevenness candidate image, and filtering and removing an edge map block image in which the block edge feature value is lower than the top 10%. The projection ratio value between the x-axis direction and the y-axis direction includes: a first ratio value obtained by dividing an x-axis direction projection value by a y-axis direction projection value; and a second ratio value obtained by dividing the y-axis direction projection value by the x-axis direction projection value.

7. The display defect detection system of claim 6, wherein, The shape ratio calculator calculates a plurality of first ratio values and a plurality of second ratio values for the plurality of angles, and calculates one maximum ratio value among the plurality of first ratio values and the plurality of second ratio values as a shape ratio feature value of the diagonal unevenness.

8. The display defect detection system of claim 3, wherein, The edge detector detects edge features in x-axis and y-axis directions by using a Sobel filter, and weighted averages the edge features to generate the edge map image. 9.A display defect detection method, comprising: receiving a captured image of a test pattern displayed by a display panel as a panel image including diagonal unevenness, and pre-processing the panel image to output a pre-processed image; calculating a final feature value of the diagonal unevenness by multiplying a first feature value reflecting a luminance difference of the diagonal unevenness with a second feature value reflecting a shape ratio of the diagonal unevenness based on an edge map of the pre-processed image; and detecting a display position of the diagonal unevenness based on the final feature value of the diagonal unevenness.

10. The display defect detection method of claim 9, wherein, The pre-processing of the panel image includes: cropping the panel image to extract a region of interest (ROI) including the diagonal unevenness, and changing a size of the ROI to a predetermined size; normalizing a color sense or luminance of the ROI; and removing noise components impeding detection of the diagonal unevenness from the normalized ROI.

11. The display defect detection method of claim 9, wherein, The calculating of the final feature value of the diagonal unevenness includes: generating an edge map image including edge information about the pre-processed image; dividing the edge map image in units of blocks, and calculating a block edge feature value corresponding to each of the edge map block images; selecting, from the edge map block images, an edge map block image in which the block edge feature value is greater than or equal to a reference value as an unevenness candidate image; projecting the diagonal unevenness included in the unevenness candidate image in x-axis and y-axis directions while rotating the unevenness candidate image at a plurality of angles, and calculating a projection ratio value corresponding to an angle in which a projection ratio value between the x-axis direction and the y-axis direction is greatest as a shape ratio feature value of the diagonal unevenness; and multiplying the block edge feature value of the unevenness candidate image indicating the first feature value with the shape ratio feature value of the diagonal unevenness indicating the second feature value to calculate a luminance-shape ratio feature value indicating the final feature value of the diagonal unevenness.

12. The display defect detection method of claim 11, wherein, Each of the edge map image and the unevenness candidate image is a binary image.

13. The display defect detection method of claim 11, wherein, The selecting of the unevenness candidate image includes: selecting, from the edge map block images, an edge map block image in which the block edge feature value is in a top 10% as the unevenness candidate image; and filtering and removing an edge map block image in which the block edge feature value is lower than the top 10%.

14. The display defect detection method of claim 11, wherein, The projection ratio value between the x-axis direction and the y-axis direction includes: a first ratio value obtained by dividing an x-axis direction projection value by a y-axis direction projection value; and a second ratio value obtained by dividing the y-axis direction projection value by the x-axis direction projection value.

15. The display defect detection method of claim 14, wherein, calculating the projection ratio value as the shape ratio feature value includes: calculating a plurality of first ratio values and a plurality of second ratio values for the plurality of angles; and calculating a maximum ratio value among the plurality of first ratio values and the plurality of second ratio values as the diagonal unevenness shape ratio feature value.

16. The display defect detection method of claim 11, wherein, generating the edge map image includes detecting edge features in x-axis and y-axis directions by using a Sobel filter, and generating the edge map image by weighted averaging the edge features.

Images