Inspection method, inspection device, and program

Abstract

A method for inspecting a display panel by a computer comprises: an acquisition step (S11) for acquiring an abnormal part image which is an image including an abnormal part of a pixel region of a display panel obtained by performing image processing using a background subtraction method on an inspection image of the pixel region; a generation step (S12) for generating, from the abnormal part image, a label image obtained by converting a region showing the abnormal part to a color corresponding to the defect pattern of the abnormal part using the learned generation model; and a determination step (S13) for determining, on the basis of the color of the region in the label image, the possibility of whether or not a defect pattern of an abnormal portion is a penetration defect that is a defect that does not emit light due to deterioration of the functional layer in the pixel region, the defect pattern including a black spot defect and a penetration defect.

Description

Technical Field

[0001] The present disclosure relates to an inspection method, an inspection device, and a program. Background Art

[0002] In the production process of organic EL (Electro Luminescence) display panels, various inspections are performed to ensure product quality.

[0003] Among various inspection steps, a DS (Dark Spot) inspection is performed in an inspection step for inspecting display defects in a pixel region. The DS inspection is performed to inspect whether there is a black spot-like display defect (hereinafter referred to as a penetration defect) caused by a defect in a moisture barrier layer.

[0004] In the DS inspection, an operator checks an enlarged image of a pixel area, and if a region where moisture has penetrated (hereinafter referred to as a penetration region) is reflected in the light-emitting layer of the pixel area, it is determined whether it is a penetration defect based on the size of the penetration region.

[0005] However, in addition to penetration defects, poor display in the pixel area also includes poor display caused by black spots, etc., so the operator needs to judge the bad mode (hereinafter referred to as the defect mode). The judgment of this defect mode and the measurement of the size of the penetration area depend on the judgment of the operator. For this reason, there will be differences in the judgment criteria due to different operators, and even for the same operator, the judgment criteria will fluctuate over time. This will cause the problem of over-killing of good products as defective products, or under-killing of defective products as good products.

[0006] In this regard, for example, Patent Document 1 proposes an image classification method that automatically classifies images in which defects are detected during appearance inspection.

[0007] (Prior art literature)

[0008] (Patent Document)

[0009] Patent Document 1 Japanese Patent Application Publication No. 2017-054239 Summary of the invention

[0010] Problems to be solved by the invention

[0011] However, the penetration defect in the pixel area of ​​the organic EL display panel is not considered as a defect type in Patent Document 1. That is, even if the image classification method of Patent Document 1 is used, the defect type can only be classified as one of foreign matter, defect, and bubble, but the penetration defect in the pixel area of ​​the organic EL display panel cannot be classified.

[0012] In view of the above situation, the present disclosure aims to provide an inspection method and the like that can automatically determine a defect mode in a pixel region of a display panel.

[0013] Means for solving problems

[0014] In order to achieve the above-mentioned purpose, an inspection method involved in one embodiment of the present invention is an inspection method for a display panel performed by a computer, the inspection method comprising the following steps: an acquisition step, acquiring an abnormal portion image, the abnormal portion image is an image of an abnormal portion of the pixel area obtained by performing image processing using a background difference method on an inspection image of the pixel area of ​​the display panel; a generation step, using a learned generation model, generating a label image from the abnormal portion image, the label image is an image after transforming the area showing the abnormal portion into a color corresponding to a defect mode of the abnormal portion; and a judgment step, judging, based on the color of the area in the label image, the possibility of whether the defect mode of the abnormal portion is a penetration defect, the penetration defect being a defect caused by the functional layer of the pixel area deteriorating and not emitting light, the defect mode including a penetration defect and a non-luminous black spot defect caused by a short circuit or an open circuit in the pixel area.

[0015] Based on this, the defect mode in the pixel area of ​​the display panel can be automatically determined.

[0016] It may also be that the inspection method further includes a CNN judgment step. Before the judgment step, a classification result showing a defect pattern of the abnormal part is obtained from the abnormal part image using a learned CNN model. The CNN refers to a Convolutional Neural Network. In the judgment step, the possibility of whether the defect pattern of the abnormal part is a penetration defect is judged based on the classification result obtained in the CNN judgment step and the color of the area in the label image.

[0017] It may also be that, for example, in the judgment step, when the defect pattern shown by the classification result obtained by the CNN judgment step is consistent with the defect pattern shown by the color of the area in the label image, the computer judges whether there is a possibility that the defect pattern of the abnormal part is a penetration defect, and in the case of inconsistency, the inconsistency is notified to the operator so that the operator can judge whether there is a possibility that the defect pattern of the abnormal part is a penetration defect. The inspection method further includes the following steps: a measurement step, when it is judged in the judgment step that there is a possibility that the defect pattern of the abnormal part is a penetration defect, measuring whether the size of the area in the label image is above a specified value; and a penetration defect judgment step, when the size of the area is above the specified value in the measurement step, judging that the abnormal part is a penetration defect.

[0018] And it may also be that, for example, the inspection method includes the following steps: a measuring step, when it is judged in the judging step that there is a possibility that the defect mode of the abnormal part is a penetration defect, measuring whether the size of the area in the label image is above a specified value; and a penetration defect judging step, when the size of the area is above the specified value in the measuring step, judging that the abnormal part is a penetration defect.

[0019] It may also be, for example, that the learned generative model is learned using a learning abnormal part image and a learning label image prepared as training data, wherein the learning abnormal part image is an image obtained by performing image processing using a background difference method on an inspection image of a pixel area of ​​the display panel, and the learning label image is an image in which an area showing the abnormal part projected in the abnormal part image is transformed into a color corresponding to a defect mode of the abnormal part, the defect mode of the abnormal part showing a black spot defect, a penetration defect, or normal.

[0020] Here, for example, the learned generative model may be a neural network model based on GAN (Generative Adversarial Networks: Generative Adversarial Networks). Also, for example, the learned generative model may be a neural network model of Pix2Pix.

[0021] Furthermore, for example, the abnormal portion image for learning may be subjected to histogram adjustment so that a background region other than a region showing the abnormal portion becomes uniformly white.

[0022] In addition, these general or specific methods can be implemented by devices, methods, integrated circuits, computer programs, or computer-readable recording media such as CD-ROMs, or by any combination of systems, methods, integrated circuits, computer programs, and recording media.

[0023] Effects of the Invention

[0024] The present disclosure can provide an inspection method and the like that can automatically determine a defect mode in a pixel region of a display panel. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Figure 1 A schematic configuration of an inspection system including an inspection device according to an embodiment is shown.

[0026] Figure 2A An example of an enlarged image of an inspection image used in DS inspection according to the embodiment is shown.

[0027] Figure 2B This is an example of an enlarged image of an inspection image used in the DS inspection according to the embodiment.

[0028] Figure 3 This is a schematic diagram for explaining the mechanism of occurrence of penetration defects.

[0029] Figure 4A This is an example of the size of the permeation area shown in the enlarged image of the inspection image used in the DS inspection according to the embodiment.

[0030] Figure 4B This is an example of the size of the permeation area shown in the enlarged image of the inspection image used in the DS inspection according to the embodiment.

[0031] Figure 5 This is an example of the hardware configuration of a computer that realizes the functions of the inspection device according to the embodiment by software.

[0032] Figure 6 This is a block diagram showing an example of the functional configuration of the inspection device according to the embodiment.

[0033] Figure 7 Another example of an enlarged image of the inspection image of the organic EL display panel used in the DS inspection according to the embodiment is shown.

[0034] Figure 8 Shows from Figure 7 An example of an abnormal portion image obtained from an enlarged image of the inspection image shown.

[0035] Fig. 9An example of a label image according to the embodiment is shown.

[0036] Fig.10 An example of an image pair prepared as training data according to the embodiment is shown.

[0037] Fig.11A Other examples of image pairs prepared as training data according to the embodiment are shown.

[0038] Fig. 11B Another example of an image pair prepared as training data according to the embodiment is shown.

[0039] Fig. 11C Another example of an image pair prepared as training data according to the embodiment is shown.

[0040] Fig.11D Another example of an image pair prepared as training data according to the embodiment is shown.

[0041] Fig.12 It is conceptually shown that Figures 10 to 11D A diagram illustrating a method in which image pairs are used as training data to enable a generative model to learn.

[0042] Fig.13 This is a diagram for conceptually explaining the size measurement of the color label portion of the label image according to the embodiment.

[0043] Fig.14 This is a flowchart showing the operation of the inspection device according to the embodiment.

[0044] Fig.15 This is a block diagram showing an example of the functional configuration of an inspection device according to a modification of the embodiment.

[0045] Fig.16 This is a flowchart showing a part of the operation of the inspection device according to the modification of the embodiment. DETAILED DESCRIPTION

[0046] The following detailed description is made of the embodiments of the present disclosure using the accompanying drawings. In addition, the embodiments to be described below are all specific examples of the present disclosure. The numerical values, shapes, materials, specifications, constituent elements, configuration positions of constituent elements, connection methods, steps, order of steps, etc. shown in the following embodiments are examples, and their main purpose is not to limit the present disclosure. Moreover, the constituent elements of the following embodiments that are not recorded in the independent technical solutions of the present disclosure will be described as arbitrary constituent elements. Moreover, each figure is not necessarily a rigorous illustration. In each figure, the same number is given to substantially the same structure, and there are cases where repeated descriptions are omitted or simplified.

[0047] (Implementation Method)

[0048] Hereinafter, the inspection device and the like according to the present embodiment will be described.

[0049] [1. Check the system]

[0050] Hereinafter, the inspection device 10 according to the present embodiment will be described with reference to the drawings.

[0051] Figure 1 The schematic structure of the inspection system including the inspection device 10 involved in this embodiment is shown. In this embodiment, the inspection object of the inspection device 10 is an organic EL display panel 30 as an example for description. The inspection object of the inspection device 10 can be a display panel using a quantum dot light emitting element (QLED: Quantum dot Light Emitting Diode).

[0052] Figure 1 The inspection system shown includes an inspection device 10 , an imaging device 20 , a mounting table 21 , and a mounting table driving unit 22 .

[0053] The inspection device 10 is a device that automatically performs DS inspection, and the DS inspection checks whether there is a penetration defect in the pixel area of ​​the organic EL display panel 30. The penetration defect is a black spot-like display defect caused by a defect in the moisture barrier layer as described above. More specifically, the penetration defect involved in the present embodiment is a display defect caused by non-luminescence due to degradation of the functional layer including the light-emitting layer in the pixel area, which is one of the defect modes. Typically, it is the degradation of the functional layer including the light-emitting layer caused by moisture. Moreover, the penetration defect is mostly manifested as a display defect in which moisture penetrates into the light-emitting layer in the pixel area. In other words, the defect manifested by the penetration defect is that in the enlarged image of the pixel area, an area (penetration area) where moisture has penetrated is reflected in the light-emitting layer of the organic EL display panel 30, and the size of the penetration area is above a specified value. The generation mechanism of the penetration defect will be described later.

[0054] The camera device 20 captures the inspection target area on the organic EL display panel 30, and is composed of, for example, a CCD (Charge Coupled Device) or a CMOS (Complementary Metal-Oxide Semiconductor). More specifically, the camera device 20 captures the inspection target area, i.e., the pixel area, on the organic EL display panel 30 to obtain an inspection image of the organic EL display panel 30. In addition, although the camera device 20 is controlled by the inspection device 10, it may also be controlled by another computer.

[0055] The mounting table 21 holds the organic EL display panel 30 .

[0056] The stage drive unit 22 is composed of a ball screw, a guide rail, and a motor, and moves the stage 21 relative to the imaging device 20. The stage drive unit 22 is controlled by the inspection device 10, but may be controlled by another computer.

[0057] Figure 2A as well as Figure 2B This is an example of an enlarged image of an inspection image used in the DS inspection according to the present embodiment. Figure 2A An example is shown in which a penetration region that becomes a penetration defect is reflected in an enlarged image 91 of a pixel region of the organic EL display panel 30. Figure 2B An example is shown in which a black dot (black dot region) that is a black dot defect is reflected in an enlarged image 92 of a pixel region of an organic EL display panel 30. In addition, a black dot is a pixel that does not emit light (lights up) due to a short circuit or an open circuit in the pixel region, but may also include a pixel with low light brightness.

[0058] In addition, when the operator performs the DS inspection in person, the outline shape is used to identify whether the penetration area or black spot is reflected in the enlarged image of the pixel area. Since the penetration area is the area where water has penetrated in the light-emitting layer of the pixel area, and the area where water has penetrated will expand due to the influence of time and become an enlarged area, it has the characteristic of smooth outline, so it can be distinguished from the black spot area with uneven outline. However, the outline is difficult to identify, and the operator may mistakenly judge the penetration area as a black spot area.

[0059] Figure 3 This is a schematic diagram for explaining the mechanism of occurrence of penetration defects. Figure 3 Schematically shows an example of a cross-sectional view of a pixel region of an organic EL display panel 30. The organic EL display panel 30 is Figure 3As shown, the organic EL display panel 30 includes: a glass substrate 311, a thin film transistor layer 312 formed on the glass substrate 311, a light emitting layer 313 formed on the thin film transistor layer 312, and a protective film 314 formed on the light emitting layer 313. The protective film 314 functions as a moisture barrier layer to block water. In addition, in the organic EL display panel 30, an upper substrate is formed on the protective film 314 by a filling material 315 such as an adhesive. Figure 3 In the example shown, the upper substrate is formed of a color filter layer 316 and a glass substrate 317. The color filter layer 316 includes a pixel region and a black matrix (BM) for partitioning the pixel region. Alternatively, the upper substrate may be a substrate formed of a flexible polarizing plate or the like.

[0060] like Figure 3 As shown in FIG. 1 , when the protective film 314 of the organic EL display panel 30 is formed, foreign matter 320 is included, thereby forming a gap in the protective film 314. That is, it can be regarded as a defect of the moisture blocking layer occurring in the organic EL display panel 30. Figure 3 In the figure, moisture represented by H2O descends from the color filter layer 316 and the like to the protective film 314. In addition to moisture contained in the color filter layer 316, moisture is also contained in the filling material 315. Then, the moisture that descends to the protective film 314 invades the gap of the protective film 314 and adheres to the light-emitting layer 313. In this way, the moisture that descends to the protective film 314 invades the protective film 314 through the invasion path 321 and adheres to the light-emitting layer 313. Then, the moisture adhesion area extends along the direction of, for example, arrow 322, that is, along the light-emitting layer 313. In this way, the area where moisture penetrates expands in the light-emitting layer of the pixel area. In addition, as a factor that promotes the extension, it is dominated by heat, so the penetration area will expand due to the influence of temperature and time. For this reason, the size of the penetration area that becomes a defective product during DS inspection is determined according to the size of the penetration area that is considered a good product at the end of the product life of the organic EL display panel. Here, the size of the penetration area that becomes a defective product during DS inspection is, for example, on the order of tens of microns.

[0061] Figure 4A as well as Figure 4B This is an example of the size of the permeation area reflected in the enlarged image of the inspection image used in the DS inspection according to the present embodiment. Figure 4A as well as Figure 4B The size of the see-through area reflected in the enlarged image of the pixel area of ​​the organic EL display panel 30 is shown.

[0062] In addition, when the operator performs the DS inspection himself, the size of the penetration area reflected in the enlarged image of the pixel area is measured. Figure 4A as well as Figure 4BAs shown in FIG. 1 , since the permeation area is partially blocked by the pixel division, ie, BM, it is difficult for the operator to recognize the end of the permeation area. Thus, there may be a difference in the size measurement of the permeation area depending on the operator.

[0063] [1-1. Hardware Configuration of Inspection Device 10]

[0064] Before describing the functional configuration of the inspection device 10 according to the present embodiment, Figure 5 An example of the hardware configuration of the inspection device 10 according to the present embodiment will be described.

[0065] Figure 5 An example of the hardware configuration of the computer 1000 that realizes the functions of the inspection device 10 according to the present embodiment by software is shown.

[0066] Computer 1000 Figure 5 As shown, a computer includes an input device 1001, an output device 1002, a CPU 1003, an internal storage 1004, a RAM 1005, a GPU 1006, a reading device 1007, a transceiver 1008, and a bus 1009. The input device 1001, the output device 1002, the CPU 1003, the internal storage 1004, the RAM 1005, the GPU 1006, the reading device 1007, and the transceiver 1008 are connected via the bus 1009.

[0067] The input device 1001 is a device that becomes a user interface such as input keys, a touch pad, a touch screen display, etc., and accepts operations from the user. In addition, in addition to accepting contact operations from the user, the input device 1001 can also be configured to accept operations performed by voice or remote operations such as a remote controller.

[0068] The output device 1002 is used together with the input device 1001 and is composed of a touch panel or a touch screen display, etc., and notifies the user of information that the user should know.

[0069] The internal storage 1004 is a flash memory or the like. In addition, at least one of a program for realizing the functions of the inspection device 10 and an application utilizing the functions of the inspection device 10 may be stored in advance in the internal storage 1004. In addition, the internal storage 1004 may store a model of a neural network (a generation model, etc.), acquired learning data, parameters of an intermediate layer of the model, the order of performing image processing such as a background difference method, the order of performing judgments such as a non-black spot judgment and a DS judgment described later, and the like.

[0070] RAM 1005 is a random access memory (Random Access Memory) and is used to store data and the like when executing a program or an application.

[0071] The GPU 1006 is a graphics processing unit that copies programs, applications, and data stored in the internal storage 1004 to a dedicated RAM built into the GPU and performs graphics processing according to instructions in the programs or applications.

[0072] The reading device 1007 reads information from a recording medium such as a USB (Universal Serial Bus) memory. The reading device 1007 reads the program or / and application from the recording medium recording the program or application, and stores the program or application in the internal storage 1004.

[0073] The transceiver 1008 is a communication line for wireless or wired communication. The transceiver 1008 can communicate with a server device connected via a network, download the above-mentioned program and / or application from the server device, and store it in the internal storage 1004.

[0074] CPU 1003 is a central processing unit (Central Processing Unit), which copies the program and application stored in the internal storage 1004 to RAM 1005, and sequentially reads and executes the instructions in the program and / or application from RAM 1005.

[0075] [1-2. Functional Configuration of Inspection Device 10]

[0076] Next, use Figure 6 Each functional configuration of the inspection device 10 according to the present embodiment will be described.

[0077] Figure 6 This is a block diagram showing an example of the functional configuration of the inspection device 10 according to the present embodiment.

[0078] Inspection device 10 Figure 6 As shown, the inspection device 10 includes an image acquisition unit 101, a label image generation unit 102, a non-black spot determination unit 103, a size measurement unit 104, and a DS determination unit 105. In addition, in the inspection device 10, the size measurement unit 104 and the DS determination unit 105 are not essential and may be placed outside.

[0079] [1-2-1. Image acquisition unit 101]

[0080] The image acquisition unit 101 acquires an abnormal portion image including an abnormal portion of a pixel region obtained by performing image processing using a background difference method on an inspection image of the pixel region of the organic EL display panel 30 .

[0081] In this embodiment, the image acquisition unit 101 acquires an inspection image of a pixel region of the organic EL display panel 30 used in DS inspection from the camera device 20. Furthermore, the image acquisition unit 101 performs image processing using a background difference method on the acquired inspection image to generate a background difference image, i.e., an abnormal portion image, including an abnormal portion of the pixel region of the organic EL display panel 30. In addition, the image acquisition unit 101 can realize various functions such as an inspection image acquisition function, an image processing function, and an abnormal portion image generation function by executing a control program stored in a memory by a processor in a computer that realizes the functions of the inspection device 10.

[0082] Here, the background difference method refers to image processing that extracts objects that do not exist in the background image but exist in the observed image by comparing the observed image with the background image. In the present embodiment, the background difference method is image processing that extracts only the abnormal portion by taking the difference between the abnormal pixel image and the normal pixel image to remove the normal brightness distribution information from the abnormal pixel image. The abnormal pixel image is an image that includes abnormal pixels containing penetration areas, black spots, etc., and is an enlarged image of the inspection image of the pixel area of ​​the organic EL display panel 30. The normal pixel image is an image that includes normal pixels that do not contain penetration areas, black spots, etc., and is an enlarged image of the inspection image of the pixel area of ​​the organic EL display panel 30.

[0083] Furthermore, the normal pixel image is located at a position different from the abnormal portion in the inspection image of the pixel region of the organic EL display panel 30, and is an image that does not include the abnormal portion. In other words, the normal pixel image is an image of the same scale as the image of the region including the abnormal portion in the inspection image. In addition, the normal pixel image may be an image obtained from the inspection image, or may be an image of the same scale as the inspection image and prepared in advance.

[0084] The following use Figure 7 as well as Figure 8 An example of generating an abnormal portion image by the background subtraction method will be described. Figure 7 This is another example of an enlarged image of the inspection image of the organic EL display panel 30 used in the DS inspection according to the present embodiment. Figure 7 (a) shows an abnormal pixel image 41 including a permeation region, that is, an abnormal portion 31a, Figure 7 (b) shows a normal pixel image 42 . Figure 8 Shows from Figure 7 The figure shows an example of an abnormality image 43 obtained by enlarging the inspection image.

[0085] That is, the image acquisition unit 101 performs image processing using a background difference method. Figure 7 The abnormal pixel image 41 shown in (a) is Figure 7 The image acquisition unit 101 can generate a pixel image including the difference between the normal pixel image 42 shown in (b) to remove the normal brightness distribution information from the abnormal pixel image, thereby extracting only the abnormal part. Figure 8 An abnormal portion image 43 of the abnormal portion 31b is shown. Figure 8 The abnormal part image 43 shown is an image in which only Figure 7 An image of the abnormal portion 31a shown in (a).

[0086] In this way, the image acquisition unit 101 can acquire the abnormal part image.

[0087] [1-2-2. Label Image Generating Unit 102]

[0088] The label image generation unit 102 generates a label image from the abnormal part image acquired by the image acquisition unit 101 using the learned generation model, and the label image is an image in which the area showing the abnormal part is transformed into a color corresponding to the defect mode of the abnormal part. In addition, the label image generation unit 102 can realize the generation function of the label image using the learned generation model by executing the control program stored in the memory by the processor of the computer realizing the function of the inspection device 10. The learned generation model is generated by learning the learning abnormal part image and the learning label image prepared as training data. The learning abnormal part image is an image obtained by performing image processing using the background difference method on the inspection image of the pixel area of ​​the organic EL display panel 30. The learning label image is an image in which the area showing the abnormal part reflected in the abnormal part image is transformed into a color corresponding to the defect mode of the abnormal part. The defect mode of the abnormal part shows a black spot defect, a penetration defect, or normal.

[0089] Fig. 9 An example of the label image 44 according to the present embodiment is shown. Fig. 9 The label image 44 shown is an image including a color label portion 31c. The color label portion 31c is Figure 8 The region of the abnormal portion 31 b of the abnormal portion image 43 shown is converted into a color region (color label) corresponding to the defect mode of the abnormal portion 31 b.

[0090] In this embodiment, the label image generating unit 102 generates, for example, Figure 8 The abnormal part image 43 shown in FIG. 4 generates a color label portion 31c. Fig. 9In the label image 44 shown, the color label portion 31c is a portion in which the region of the abnormal portion 31b in the abnormal portion image 43 is converted into a color label corresponding to the defect mode of the abnormal portion 31b. In addition, the defect mode of the abnormal portion 31b is, for example, the above-mentioned penetration defect, black spot defect, normal, etc. In the color label portion 31c, the defect in the region of the abnormal portion 31b that is divided between pixels, that is, the BM occlusion is compensated by the color corresponding to the defect mode.

[0091] In the present embodiment, a generative model is used to generate the label image 44. The generative model is a neural network model that compensates for one of the paired images for learning. Here, the generative model involved in the present embodiment is, for example, a neural network model of Pix2Pix. Pix2Pix is ​​a generative model that uses a neural network to automatically extract the relationship hidden between the paired images for learning, and compensates for the other based on one of the paired images using the extracted relationship. In addition, the generative model involved in the present embodiment can be of any configuration as long as it is composed of a neural network model such as GAN (Generative Adversarial Networks) that performs adversarial generative learning of image pairs. In other words, the generative model involved in the present embodiment can be of any configuration as long as it is a neural network model based on GAN.

[0092] Next, an example of a method for learning a generative model for generating a label image from an abnormal portion image will be described.

[0093] Fig.10 An example of an image pair prepared as training data according to the present embodiment is shown. Fig.10 The abnormal part image 61 and its label image 62 shown are examples of an image pair prepared as training data. Fig.10 (a) shows an abnormal portion image 61 including an abnormal portion 61a as a permeation area. Fig.10 (b) shows a label image 62 including a color label portion 62b, which is a portion in which the region of the abnormal portion 61a is transformed (replaced) into a color (shaded in the figure) corresponding to the case where the defect mode of the abnormal portion 61a of the abnormal portion image 61 is a penetration defect. In addition, in the color label portion 62b, the defect portion of the abnormal portion 61a that is divided between pixels, that is, BM occlusion, is transformed into a color corresponding to the case where the defect mode of the abnormal portion 61a is a penetration defect.

[0094] FIG. 11A to FIG. 11D Another example of the image pair prepared as training data according to the present embodiment is shown.

[0095] Fig.11A (a) shows an example of a normal image, that is, an abnormal portion image without an abnormal portion. Fig.11A (b) shows the label image when the defect pattern of the abnormal part image is normal. Fig.11A There is no abnormal part in the abnormal part image shown in (a), so Fig.11A The label image of (b) becomes an image showing a normal white color label portion.

[0096] Fig. 11B (a) shows an example of an abnormal portion image including an abnormal portion as a permeation area. Fig. 11B (b) shows a label image including a color label portion, which is a portion corresponding to the situation drawn with hatching when the defect mode of the abnormal portion image is a penetration defect.

[0097] Fig. 11C (a) shows an example of an abnormal portion image including an abnormal portion as a black dot. Fig. 11C (b) shows a label image including a color label portion, which is a portion corresponding to the case where the defect mode of the abnormal portion image is a black dot defect and is drawn with hatching.

[0098] Fig.11D (a) shows an example of an abnormal portion image including an abnormal portion in which a permeated area and black dots are mixed. Fig.11D (b) shows a label image including a color label portion, which is a portion drawn with a shade corresponding to the defect mode of each abnormal portion of the abnormal portion image.

[0099] Fig.12 Conceptually, the Figures 10 to 11D The image pairs shown are used as training data to enable the generative model to learn. Fig.12 In the example of the abnormal part image inputted, in order to conceptually represent the abnormal part image, the pixels are represented by white frames, and the abnormal part is shaded except for the part blocked by BM.

[0100] When learning a generative model for generating a label image from an abnormal part image, first prepare a plurality of Figures 10 to 11D The abnormal part image and the label image shown in FIG. Fig.12 As shown, the neural network based on the GAN constituting the generative model is supervised learning in a manner that the input is an abnormal part image and the output is a corresponding label image. In this way, a generative model that generates a label image when an abnormal part image is input can be obtained.

[0101] In addition, Fig.12 Although the neural network constituting the generative model involved in this embodiment is conceptually shown in the figure, it can be constituted by a neural network such as a GAN that performs adversarial generative learning of image pairs such as pix2pix as described above. Furthermore, the neural network constituting the generative model involved in this embodiment can be constituted by a neural network that can obtain a generative model that generates a label image when an abnormal part image is input, and can be any configuration.

[0102] In this way, the label image generating unit 102 can generate a label image including a color label portion that is a portion depicted and distinguished in a color corresponding to the defect mode of each abnormal portion of the abnormal portion image, from the abnormal portion image acquired by the image acquiring unit 101 .

[0103] In addition, the abnormal part image for learning in the image pair prepared when the generative model is learned can also be histogram-adjusted in a manner that the background area of ​​the area showing the abnormal part is uniformly whitened, that is, in a manner that it is converted into a uniform white color represented by a grayscale value of 255. According to this, since only the information required by the generative model when generating the label image can be extracted in advance, it is possible to form a generative model that generates the label image with higher accuracy and learn such a generative model. In addition, when the abnormal part image for learning is used as an image for which the histogram is adjusted, the inspection image for inspection can also use an image for which the histogram is adjusted.

[0104] [1-2-3. Non-black spot determination unit 103]

[0105] The non-black spot determination unit 103 determines the possibility that the defect mode of the abnormal part is a penetration defect of the light-emitting layer in which water penetrates into the pixel area based on the color of the area in the label image. In addition, the non-black spot determination unit 103 can realize the determination function by, for example, executing a control program stored in a memory by a processor in a computer that realizes the function of the inspection device 10.

[0106] In this embodiment, the non-black spot determination unit 103 obtains the label image generated by the label image generation unit 102, and determines whether the defect mode of the abnormal part corresponding to the color label part is not a black spot defect according to the color of the color label part included in the obtained label image. Fig. 9 In this case, the non-black dot determination unit 103 determines that the defect mode of the abnormal portion 31b corresponding to the color label portion 31c is not a black dot defect based on the color of the color label portion 31c of the acquired label image 44.

[0107] In this way, the non-black dot determination unit 103 can determine that there is a possibility that the defect mode of the abnormal portion 31 b corresponding to the color label portion 31 c of the acquired label image 44 is a penetration defect.

[0108] When the defect mode of the abnormal portion 31 b corresponding to the color label portion 31 c is normal, the non-black dot determination unit 103 ends the processing on the abnormal portion image because the inspection result in the DS inspection is OK (good product).

[0109] [1-2-4. Size measuring unit 104]

[0110] The size measuring unit 104 measures whether the size of the area of ​​the color label portion in the label image is greater than a predetermined value when the non-black spot judging unit 103 judges that the defect mode with the abnormal portion is likely to be a penetration defect. In addition, the size measuring unit 104 can realize the measurement function through image processing by executing a control program stored in a memory, for example, by a processor in a computer realizing the function of the inspection device 10.

[0111] Fig.13 This is a diagram conceptually illustrating the size measurement of the color label portion 31 c of the label image 44 according to the present embodiment.

[0112] In this embodiment, the size measuring unit 104 obtains the defect mode determined by the non-black dot determination unit 103 as not being a black dot defect, for example Fig. 9 The label image 44 shown is processed to measure, for example, Fig.13 The size of the area of ​​the color label portion 31c in the label image 44 is shown. Fig.13 In the example shown, the size measuring unit 104 measures X μm and Y μm, ie, the size in the longitudinal direction (vertical direction) and the size in the lateral direction (horizontal direction) as the size of the area of ​​the color label portion 31 c in the label image 44 . Fig.13 X μm and Y μm shown are, for example, 57 μm and 83 μm.

[0113] In this way, the size measuring unit 104 can automatically measure the size of the area of ​​the color label portion 31 c of the label image 44 and can measure whether the size of the see-through area is equal to or larger than a predetermined value.

[0114] [1-2-5. DS determination unit 105]

[0115] The DS determination unit 105 determines that the abnormal portion is a penetration defect when the size of the area of ​​the color label portion in the label image is greater than a predetermined value in the measurement performed by the size measurement unit 104. In addition, the DS determination unit 105 can realize the above-mentioned determination function by executing a control program stored in a memory by a processor in a computer realizing the functions of the inspection device 10.

[0116] In this embodiment, the DS determination unit 105, for example, Fig.13 When the size of the region of the color label portion 31 c in the illustrated label image 44 is equal to or larger than a predetermined value, it is determined that the defect mode of the abnormal portion 31 a corresponding to the color label portion 31 c is a penetration defect.

[0117] In this way, the DS determination unit 105 can automatically determine whether the defect mode of the abnormal part corresponding to the color label part is a penetration defect based on the color and size of the area of ​​the color label part of the label image.

[0118] [1-3. Operation of the inspection device 10]

[0119] An example of the operation of the inspection device 10 having the above-described configuration will be described below.

[0120] Fig.14 This is a flowchart showing the operation of the inspection device 10 according to the present embodiment.

[0121] First, the inspection device 10 acquires an abnormal portion image (S11). More specifically, the image acquisition unit 101 acquires an abnormal portion image as a background difference image, which includes an abnormal portion of a pixel region obtained by performing image processing using a background difference method on an inspection image of a pixel region of the organic EL display panel 30. For example, the image acquisition unit 101 acquires Figure 8 The abnormal portion image 43 is shown as an image including the abnormal portion 31b.

[0122] Next, the inspection device 10 generates a label image using the learned generation model (S12). More specifically, the label image generation unit 102 generates a label image from the abnormal part image obtained in step S11 using the learned generation model. The label image is an image in which the region showing the abnormal part is converted into a color corresponding to the defect pattern of the abnormal part. For example, the label image generation unit 102 generates a label image from the abnormal part image obtained in step S11 using the learned generation model. Figure 8 In the abnormal part image 43 shown in FIG. 1 , the region of the abnormal part 31 b is converted into a color label corresponding to the defect mode of the abnormal part 31 b, thereby generating Fig. 9 The label image 44 shown includes a color label portion 31c.

[0123] Next, the inspection device 10 determines whether there is a possibility that the defect mode of the abnormal part is a penetration defect (S13). More specifically, the non-black spot determination unit 103 determines whether there is a possibility that the defect mode of the abnormal part is a penetration defect in which moisture penetrates into the light-emitting layer of the pixel area based on the color of the area in the label image generated in step S12. For example, the non-black spot determination unit 103 may determine whether there is a possibility that the defect mode of the abnormal part is a penetration defect in which moisture penetrates into the light-emitting layer of the pixel area based on the color of the area in the label image generated in step S12. Fig. 9 The color of the color label portion 31 c of the label image 44 shown is used to determine that the defect mode of the abnormal portion 31 b corresponding to the color label portion 31 c is not a black dot defect.

[0124] In step S13, when it is possible that the defect mode of the abnormal part is a penetration defect ("Yes" in S13), the inspection device 10 measures the size of the area showing the abnormal part in the label image generated in step S12 (S14). More specifically, when it is determined in step S13 that the defect mode of the abnormal part is a penetration defect, the size measuring unit 104 measures whether the size of the area of ​​the color label part in the label image is greater than a specified value. For example, according to Fig. 9 When the label image 44 shown in the figure is judged that the defect mode is a black dot defect and is not normal, the size measuring unit 104 measures the size of the label image 44 by image processing. Fig.13 The size of the area of ​​the color label portion 31c in the label image 44 is shown. In step S13, if there is no possibility that the defect mode of the abnormal portion is a penetration defect (No in S13), the inspection device 10 ends the present process, that is, ends the DS inspection.

[0125] Next, the inspection device 10 determines whether the size of the area measured in step S14 is greater than a specified value (S15). More specifically, the DS determination unit 105 determines in step S15 whether the size of the area of ​​the color label portion in the label image measured in step S14 is greater than a specified value. In the present embodiment, the specified value is a value of several tens of microns.

[0126] In step S15, if the size of the area is greater than a specified value ("Yes" in S15), the inspection device 10 determines that the abnormal part of the abnormal part image acquired in step S11 is a penetration defect (S16). More specifically, in step S15, if the size of the area of ​​the color label part in the label image is greater than a specified value, the DS judgment unit 105 determines that the abnormal part is a penetration defect. For example, the DS judgment unit 105 Fig.13 When the size of the region of the color label portion 31 c in the illustrated label image 44 is equal to or larger than a predetermined value, it is determined that the defect mode of the abnormal portion 31 a corresponding to the color label portion 31 c is a penetration defect.

[0127] In addition, in step S15, when the size of the area is not greater than the specified value ("No" in S15), the inspection device 10 determines that the abnormal part of the abnormal part image obtained in step S11 is not a penetration defect (S17). More specifically, in step S15, when the DS judgment unit 105 determines that the size of the area of ​​the color label part in the label image is smaller than the specified value, the DS judgment unit 105 determines that the abnormal part is not a penetration defect. For example, the DS judgment unit 105 Fig.13 When the size of the color label portion 31c in the label image 44 is smaller than the specified value, it is determined that the defect mode of the abnormal portion 31a corresponding to the color label portion 31c is not a penetration defect. Therefore, the DS determination unit 105 determines that the pixel area of ​​the organic EL display panel 30 having the abnormal portion image including the abnormal portion 31a is a good product.

[0128] [1-4. Effects, etc.]

[0129] The inspection device 10 of the present embodiment acquires an abnormal part image, which is an image of an abnormal part of a pixel region obtained by performing image processing using a background difference method on an inspection image of the pixel region of the organic EL display panel 30. In addition, the inspection device 10 of the present embodiment generates a label image from the acquired abnormal part image using a learned generation model, and the label image is an image in which the region showing the abnormal part is converted into a color corresponding to the defect mode of the abnormal part. Therefore, based on the color of the region in the generated label image, it is determined whether there is a possibility that the defect mode of the abnormal part is a penetration defect.

[0130] In this way, the inspection device 10 of the present embodiment, etc., by using the learned generation model, generates a label image in which the defective portion partially blocked by the BM in the abnormal portion is compensated and is distinguished by color according to each defect mode from the abnormal portion image which is the background difference image including the abnormal portion of the pixel area. That is, the inspection device 10 of the present embodiment, etc., can automatically determine the defect mode in the pixel area of ​​the organic EL display panel 30.

[0131] Here, when determining that the defect mode with the abnormal portion is likely to be a penetration defect, the inspection device 10 of the present embodiment measures whether the size of the area of ​​the color label portion in the label image is greater than or equal to a predetermined value.

[0132] Such a size, that is, the size of the color label of the penetration area showing the possibility of a penetration defect after the defect is compensated, can be easily measured, so it is possible to automatically determine whether the abnormal part of the pixel area is a penetration defect with high accuracy and simplicity. In other words, the inspection device 10 of this embodiment, etc., can automatically determine the defect mode in the pixel area of ​​the organic EL display panel 30.

[0133] Therefore, in DS inspection, since the determination of defect mode and the measurement of the size of the penetration area that depend on the judgment of the operator can be automatically performed, the problem of different judgment criteria due to different operators and the fluctuation of the judgment criteria over time due to time differences even for the same operator can be solved. As a result, stable DS inspection can be performed with the same criterion, which can not only greatly improve the inspection efficiency, but also solve the problem of overkill rate and underkill rate.

[0134] In addition, the inspection device 10 of the present embodiment and the like use a method of using a learned generation model such as a generation model formed by a neural network as a method of generating a label image.

[0135] That is, the learned generative model is learned by using the learning abnormal part image and the learning label image prepared as training data. The learning abnormal part image is a background difference image obtained by performing image processing using the background difference method on the inspection image of the pixel area of ​​the organic EL display panel 30. The learning label image is an image obtained by converting the area showing the abnormal part projected in the abnormal part image into a color corresponding to the defect mode of the abnormal part. The defect mode of the abnormal part shows a black spot defect, a penetration defect, or normal. Here, the learned generative model can be a neural network model based on GAN, such as a neural network model of Pix2Pix.

[0136] Based on this, a pair of images of a learning background difference image (learning abnormal part image) and a learning label image (the missing part is compensated and distinguished by color according to each defect mode) showing various defect modes such as penetration defects and black spot defects can be prepared in advance, so that the generation model can be learned in advance. Therefore, a model of a neural network that is good at both recognition and compensation, that is, a generation model, can be used to accurately and automatically generate a label image with a color label part in which the missing part is compensated and the color is distinguished according to each defect mode from the abnormal part image obtained by image processing using the background difference method. Therefore, the label image can be used to accurately and automatically determine the defect mode and measure the size of the penetration area, so that it is possible to accurately and automatically determine whether the abnormal part of the pixel area is a penetration defect.

[0137] In addition, the abnormal part image for learning may be histogram-adjusted in such a manner that the background area except for the area showing the abnormal part is uniformly whitened, that is, in such a manner that it is converted into a uniform white represented by a grayscale value of 255. In this way, only the information required by the generation model when generating the label image can be extracted in advance, so that learning can be performed in such a manner that the generation model can generate the label image with higher accuracy.

[0138] (Variation Example)

[0139] In addition, in the above-mentioned embodiment, a label image is generated from the abnormal part image of the background difference image using the learned generation model, and the generated label image is used to determine whether the defect mode of the abnormal part is a penetration defect. In order to further improve the accuracy of determining whether the defect mode of the abnormal part is a penetration defect, a model different from the generation model, namely a CNN (Convolutional Neural Network) model, can be used to determine the defect mode of the abnormal part, so as to double-check the determination result. The following will focus on the differences from the above-mentioned embodiment to explain this situation.

[0140] [2-1. Inspection device 10A]

[0141] Fig.15 1 is a block diagram showing an example of the functional configuration of an inspection device 10A according to a modified example of the present embodiment. Figure 6 The inspection device 10 shown is different in that the functions of a CNN determination unit 106A and a non-black spot determination unit 103A are added.

[0142] [2-1-1. CNN determination unit 106A]

[0143] The CNN judgment unit 106A can judge the defect mode of the abnormal part using a model different from the generated model. More specifically, the CNN judgment unit 106A obtains the classification result showing the defect mode of the abnormal part from the abnormal part image using the learned CNN model. In addition, the CNN judgment unit 106A realizes the above judgment function by executing the control program stored in the memory by the processor in the computer realizing the function of the inspection device 10A.

[0144] The learned CNN model is learned as follows. That is, first, a plurality of abnormal part images for learning are prepared as training data according to the classification numbers corresponding to the defect patterns, which are background difference images. In addition, the abnormal part image for learning is the abnormal part image for learning in the above-mentioned embodiment, which is a background difference image obtained by performing image processing using the background difference method on the inspection image of the pixel area of ​​the organic EL display panel 30. And the classification number can be determined in a manner such as 0 for normal, 1 for penetration defect, 2 for black spot defect, and 4 for the case where penetration defect and black spot defect are mixed. When penetration defect and black spot defect are mixed, it can also be 1 and 2. Therefore, the CNN model is learned in a manner in which the input is the abnormal part image for learning and the output is the classification number. By utilizing the learned CNN model obtained in this learning manner, the CNN judgment unit 106A can obtain the classification result of the defect pattern showing the abnormal part from the abnormal part image.

[0145] [2-1-2. Non-black spot determination unit 103A]

[0146] In addition to the functions described in the above embodiments, the non-black spot determination unit 103A also has a function of double checking the determination result. In other words, the non-black spot determination unit 103A further determines the possibility that the defect mode of the abnormal part is a penetration defect based on the classification result obtained by the CNN determination unit 106A and the color of the area in the label image. In this way, since the determination result can be automatically double-checked, the determination accuracy of the penetration defect can be further improved. In addition, the non-black spot determination unit 103A can realize the determination function by, for example, executing a control program stored in a memory by a processor in a computer that realizes the function of the inspection device 10A.

[0147] More specifically, when the defect pattern shown in the classification result obtained by the CNN judgment unit 106A is consistent with the defect pattern shown in the color of the area in the label image, the non-black spot judgment unit 103A judges whether there is a possibility that the defect pattern of the abnormal part is a penetration defect. In addition, when the defect pattern shown in the classification result obtained by the CNN judgment unit 106A is inconsistent with the defect pattern shown in the color of the area in the label image, the non-black spot judgment unit 103A notifies the inconsistency. Accordingly, the non-black spot judgment unit 103A enables the operator to judge whether there is a possibility that the defect pattern of the abnormal part is a penetration defect. The operator only needs to judge whether there is a possibility that the defect pattern of the abnormal part is a penetration defect, that is, it only needs to judge whether there is a penetration area in the abnormal part image. Therefore, when there is a possibility that the defect pattern of the abnormal part is a penetration defect, that is, when there is a penetration area in the abnormal part image, it is sufficient to measure the size of the penetration area to judge whether it is above the specified value.

[0148] [2-2. Operation of Inspection Device 10A]

[0149] An example of the operation of the inspection device 10A having the above configuration will be described below.

[0150] Fig.16 1 is a flowchart showing a part of the operation of the inspection device 10A according to the modification of the present embodiment. Fig.14 The difference between the operation of the inspection device 10 shown in FIG. 1 and FIG. 2 is the processing content of step S12. More specifically, in this modification, instead of Fig.14 The processing of step S12 shown in FIG. Fig.16 The process of step S12A is shown.

[0151] First, the inspection device 10A acquires an abnormal portion image (S11). More specifically, the image acquisition unit 101 acquires an abnormal portion image as a background difference image, which includes an abnormal portion of a pixel region obtained by performing image processing using a background difference method on an inspection image of a pixel region of the organic EL display panel 30.

[0152] Next, the inspection device 10A generates a label image using the learned generation model in step S12A (S121). More specifically, the label image generation unit 102 generates a label image by transforming the region showing the abnormal portion into a color corresponding to the defect pattern of the abnormal portion from the abnormal portion image acquired in step S11 using the learned generation model.

[0153] Furthermore, in step S12A, the inspection device 10A obtains a classification result of a defect pattern showing the abnormal portion from the abnormal portion image obtained in step S11 using the learned CNN model (S122). More specifically, the CNN determination unit 106A obtains a classification result of a defect pattern showing the abnormal portion from the abnormal portion image obtained in step S11 using the learned CNN model.

[0154] Next, the inspection device 10A determines in step S12A whether the defect pattern indicated by the classification result acquired in step S122 matches the defect pattern indicated by the color indicating the abnormal portion in the label image generated in step S121 ( S123 ).

[0155] In step S123, if the defect patterns are consistent (“Yes” in S123), the process proceeds to Fig.14 Step S13 is shown. In this way, by double checking the determination of the defect mode of the abnormal part, the determination accuracy of the penetration defect can be further improved.

[0156] In addition, in step S123, if the defect mode is inconsistent ("No" in S123), the operator is notified of the inconsistency of the defect mode (S124). In this way, by double-checking the determination of the defect mode of the abnormal part, the operator is allowed to make a determination in the case of inconsistency, so that the determination can be made without misjudging the penetration defect.

[0157] [2-3. Effects, etc.]

[0158] In this modified example, in order to further improve the judgment accuracy of the inspection device 10 of the embodiment, a CNN model different from the generation model is prepared, and learning is performed using a background difference image that is pre-classified according to each defect mode, that is, a learning abnormal part image. Accordingly, in the learned CNN, when an abnormal part image as a background difference image is input, a classification number corresponding to the defect mode can be output.

[0159] Furthermore, the inspection device 10A of this variant uses the learned generation model to generate a label image from the abnormal part image, and uses the learned CNN to obtain the classification number corresponding to the defect pattern of the abnormal part of the abnormal part image. The inspection device 10A of this variant can automatically double-check the judgment result by comparing the defect pattern judged by the label image generated by the learned generation model with the classification result showing the defect pattern obtained by the learned CNN model. Accordingly, the judgment accuracy of the penetration defect can be further improved. In addition, regarding the result of the double check, when the defect pattern judged by the generation model is inconsistent with the classification result showing the defect pattern, the operator can make the judgment. In the case of inconsistency, the judgment can be made by the operator, so that the penetration defect can be judged without misjudging it.

[0160] (Other Embodiments)

[0161] As mentioned above, the inspection device, the inspection method, and the like according to the present disclosure have been described based on the embodiments and the modified examples, but the present disclosure is not limited to these embodiments and the modified examples.

[0162] Without departing from the gist of the present disclosure, various modifications that can be conceived by those skilled in the art are implemented in the embodiments and modified examples, and methods of combining some of the constituent elements in the embodiments and modified examples are all included in the scope of the present disclosure.

[0163] Furthermore, the following embodiments are also included in the scope of one or more embodiments of the present disclosure.

[0164] (1) A part of the components constituting the above-mentioned inspection device may be a computer system consisting of a microprocessor, ROM, RAM, GPU, hard disk unit, display unit, keyboard, mouse, etc. A computer program is stored in the RAM or hard disk unit. The microprocessor realizes its function by operating according to the computer program. The computer program here is composed of a combination of multiple instruction codes indicating instructions to the computer in order to realize the prescribed function.

[0165] (2) A part of the components constituting the above-mentioned inspection device may be constituted by a system LSI (Large Scale Integration). System LSI is a super-multifunctional LSI manufactured by integrating multiple components on a single chip, and specifically is a computer system composed of a microprocessor, ROM, RAM, GPU, etc. A computer program is stored in the RAM. The microprocessor or the GPU operates according to the computer program, so that the system LSI realizes its functions.

[0166] (3) A part of the components constituting the above-mentioned inspection device may be constituted by an IC card or a single module that can be installed and removed from each device. The IC card or the module is a computer system composed of a microprocessor, ROM, RAM, GPU, etc. The IC card or the module may include the above-mentioned super-multifunctional LSI. The IC card or the module realizes its function by operating the microprocessor or the GPU according to the computer program. The IC card or the module may also be tamper-proof.

[0167] (4) Furthermore, a part of the components constituting the above-mentioned inspection device may be a configuration in which the computer program or the digital signal is recorded on a computer-readable recording medium, such as a floppy disk, a hard disk, a CD-ROM, an MO, a DVD, a DVD-ROM, a DVD-RAM, a BD (Blu-ray (registered trademark) Disc), a semiconductor memory, etc. Furthermore, the digital signal may be recorded on these recording media.

[0168] Furthermore, a part of the components constituting the above-mentioned inspection device may transmit the computer program or the digital signal via an electronic communication line, a wireless or wired communication line, a network represented by the Internet, data broadcasting, or the like.

[0169] (5) The present disclosure may be the above-mentioned methods, a computer program for realizing these methods by a computer, or a digital signal constituted by the computer program.

[0170] (6) Furthermore, the present disclosure may be a computer system including a microprocessor, a GPU, and a memory, wherein the memory stores the above-mentioned computer program, and the microprocessor or the GPU may operate according to the computer program.

[0171] (7) Furthermore, the program or the digital signal may be recorded on the recording medium and transmitted, or the program or the digital signal may be transmitted via the network or the like, thereby being executed as another independent computer system.

[0172] (8) Furthermore, part of the components constituting the above-mentioned inspection device may be executed in the cloud or in a server device.

[0173] (9) The above-mentioned embodiments and modifications may be combined.

[0174] Industrial Applicability

[0175] The present disclosure can be used in an inspection method, an inspection device, a program, etc., which can automatically determine whether there is a black spot-like display defect caused by a defect in a moisture barrier layer during an inspection process for inspecting display defects in a pixel area of ​​a display panel using an organic EL display panel or a quantum dot light-emitting element.

[0176] Description of Reference Numerals

[0177] 10.10A inspection device

[0178] 20 Camera device

[0179] 21 Loading table

[0180] 22 Stage drive unit

[0181] 30Organic EL display panel

[0182] Abnormal parts of 31a, 31b, 61a

[0183] 31c, 62b color label part

[0184] 41 Abnormal pixel images

[0185] 42 normal pixel image

[0186] 43, 61 abnormal part image

[0187] 44, 62 label images

[0188] 91, 92 Enlarged image

[0189] 101 Image Acquisition Department

[0190] 102 label image generation unit

[0191] 103, 103A non-black spot judgment unit

[0192] 104 Size Measurement Department

[0193] 105DS Judgment Department

[0194] 106A CNN Judgment Department

[0195] 311, 317 glass substrate

[0196] 312 Thin Film Transistor Layer

[0197] 313 Luminous Layer

[0198] 314 protective film

[0199] 315 filling material

[0200] 316 color filter layer

[0201] 320 Foreign matter

[0202] 321 Intrusion Path

[0203] 322 Arrow

[0204] 1000 computers

[0205] 1001 Input device

[0206] 1002 output device

[0207] 1003CPU

[0208] 1004 Internal Storage

[0209] 1005RAM

[0210] 1006 GPU

[0211] 1007 reading device

[0212] 1008 transceiver

[0213] 1009 bus

Claims

1. An inspection method is a method for inspecting a display panel by a computer. The inspection method comprises the following steps: an acquisition step of acquiring an abnormal portion image, wherein the abnormal portion image is an image of an abnormal portion of the pixel region obtained by performing image processing using a background difference method on an inspection image of the pixel region of the display panel; A generation step of generating a label image from the abnormal portion image using the learned generation model, wherein the label image is an image obtained by converting a region showing the abnormal portion into a color corresponding to a defect pattern of the abnormal portion; as well as A judging step of judging, based on the color of the area in the label image, whether there is a possibility that the defect mode of the abnormal part is a penetration defect, wherein the penetration defect is a defect caused by the functional layer of the pixel area being degraded and not emitting light, The defect modes include penetration defects and black spot defects that do not emit light due to short circuits or open circuits in pixel areas.

2. The inspection method according to claim 1, The inspection method further includes a CNN judgment step. Before the judgment step, a classification result showing a defect pattern of the abnormal part is obtained from the abnormal part image using a learned CNN model, wherein the CNN refers to a Convolutional Neural Network. In the judging step, a possibility is determined as to whether the defect mode of the abnormal part is a penetration defect based on the classification result obtained in the CNN judging step and the color of the area in the label image.

3. The inspection method according to claim 2, In the judging step, when the defect pattern shown by the classification result obtained by the CNN judging step is consistent with the defect pattern shown by the color of the area in the label image, the computer judges whether there is a possibility that the defect pattern of the abnormal part is a penetration defect, In the case of inconsistency, the inconsistency is notified to the operator so that the operator can judge whether there is a possibility that the defect mode of the abnormal part is a penetration defect, The inspection method further comprises the following steps: a measuring step of measuring whether the size of the area in the label image is greater than a predetermined value when it is determined in the determining step that there is a possibility that the defect mode of the abnormal portion is a penetration defect; and The seepage defect judging step is to judge that the abnormal portion is a seepage defect when the size of the region is greater than the predetermined value in the measuring step.

4. The inspection method according to claim 1, The inspection method comprises the following steps: a measuring step of measuring whether the size of the area in the label image is greater than a predetermined value when it is determined in the determining step that there is a possibility that the defect mode of the abnormal portion is a penetration defect; and The seepage defect judging step is to judge that the abnormal portion is a seepage defect when the size of the region is greater than the predetermined value in the measuring step.

5. The inspection method according to any one of claims 1 to 4, The learned generation model is learned using a learning abnormal part image and a learning label image prepared as training data, wherein the learning abnormal part image is an image obtained by performing image processing using a background difference method on an inspection image of a pixel area of ​​the display panel, and the learning label image is an image obtained by converting a region showing an abnormal part reflected in the abnormal part image into a color corresponding to a defect pattern of the abnormal part. The defect pattern of the abnormal portion shows a black spot defect, a penetration defect, or normal.

6. The inspection method according to claim 5, The learned generative model is a neural network model based on GAN, where GAN refers to Generative Adversarial Networks, i.e., generative adversarial networks.

7. The inspection method according to claim 5, The learned generative model is a neural network model of Pix2Pix.

8. The inspection method according to claim 5, The abnormal portion image for learning is subjected to histogram adjustment so that the background region other than the region showing the abnormal portion becomes uniformly white.

9. An inspection device, which is an inspection device for display panels performed by a computer, The inspection device comprises: an image acquisition unit that acquires an abnormal portion image, wherein the abnormal portion image is an image including an abnormal portion of the pixel region obtained by performing image processing using a background difference method on an inspection image of the pixel region of the display panel; a label image generating unit that generates a label image from the abnormal portion image using the learned generation model, wherein the label image is an image obtained by converting a region showing the abnormal portion into a color corresponding to a defect pattern of the abnormal portion; as well as The non-black spot determination unit determines, based on the color of the area in the label image, whether there is a possibility that the defect mode of the abnormal part is a penetration defect, wherein the penetration defect is a defect caused by the functional layer of the pixel area being degraded and not emitting light, The defect modes include penetration defects and black spot defects that do not emit light due to short circuits or open circuits in pixel areas.

10. A program for causing a computer to execute a method for inspecting a display panel, The program causes the computer to perform the following steps: an acquisition step of acquiring an abnormal portion image, wherein the abnormal portion image is an image of an abnormal portion of the pixel region obtained by performing image processing using a background difference method on an inspection image of the pixel region of the display panel; A generation step of generating a label image from the abnormal portion image using the learned generation model, wherein the label image is an image obtained by converting a region showing the abnormal portion into a color corresponding to a defect pattern of the abnormal portion; as well as A judging step of judging, based on the color of the area in the label image, whether there is a possibility that the defect mode of the abnormal part is a penetration defect, wherein the penetration defect is a defect caused by the functional layer of the pixel area being degraded and not emitting light, The defect modes include penetration defects and black spot defects that do not emit light due to short circuits or open circuits in pixel areas.

Images