Intelligent visual detection method and device for panel and computer equipment

Abstract

The invention discloses an intelligent visual detection method and device for a panel and computer equipment. The method comprises the steps of obtaining all types of appearance defect template images of a panel, performing appearance defect type classification on all types of panel appearance defect template images, and performing deep learning network training according to all types of panel appearance defect template images and the appearance defect type classification, and performing training to obtain image models of all types of appearance defects of the panel, and performing defect detection on the appearance of the to-be-detected panel according to the image models of all types of appearance defects of the panel. By means of the mode, the efficiency and accuracy of panel appearance defect detection can be improved.

Description

Technical Field

[0001] The present invention relates to the field of panel detection technology, and in particular to a panel intelligent visual detection method, device and computer equipment. Background Art

[0002] In the field of panel inspection, appearance defect detection is an important part, and the appearance affects the quality of the panel to a certain extent.

[0003] However, existing panel inspection solutions generally rely on manual inspection of panel appearance defects during appearance inspection, resulting in low efficiency and low accuracy in panel appearance defect detection. Summary of the invention

[0004] In view of this, an object of the present invention is to provide a panel intelligent visual inspection method, device and computer equipment, which can improve the efficiency and accuracy of panel appearance defect detection.

[0005] According to one aspect of the present invention, there is provided a method for intelligent visual inspection of panels, comprising: obtaining all types of panel appearance defect sample images, and classifying the appearance defect types of the panel appearance defect sample images of all types; performing deep learning network training based on the panel appearance defect sample images of all types and the appearance defect type classification, and obtaining an image model of all types of appearance defects of the panel through training; and performing defect inspection on the appearance of the panel to be inspected based on the image model of all types of appearance defects of the panel.

[0006] Among them, the method of acquiring all types of appearance defect sample images of panels and classifying the appearance defect types of the appearance defect sample images of all types of panels includes: acquiring all types of appearance defect sample images of panels, dividing all types of appearance defect sample images into a preset number of regional image blocks, establishing two-dimensional coordinates under the same coordinate standard with the center point of each regional image block, extracting image features under each dimensional coordinate under the two-dimensional coordinates, fusing the image features with all types of appearance defect sample images of the panels to generate image fusion features, and classifying the appearance defect types according to the image integration features.

[0007] Among them, the deep learning network training is performed according to all types of appearance defect sample images of the panel and the appearance defect type classification to obtain the image model of all types of appearance defects of the panel through training, including: graying the appearance defect sample images under each classification item according to all types of appearance defect sample images of the panel and the appearance defect type classification, performing deep learning network training on the grayed appearance defect sample images using Gaussian filtering to train the frequency domain information of the grayed appearance defect sample images, locating the defect area from the frequency domain information to form a defect area set under each classification item, extracting the defect set of the defect area set under each classification item, and training the image model of all types of appearance defects of the panel according to the defect set.

[0008] Among them, the defect detection is performed on the appearance of the panel to be tested based on the image model of all types of appearance defects of the panel, including: based on the image model of all types of appearance defects of the panel, defect detection is performed by mapping the appearance image of the appearance of the panel to be tested and mapping it to the corresponding defect name in the image model.

[0009] Wherein, after performing defect detection on the appearance of the panel to be tested according to the image model of all types of appearance defects of the panel, the method further includes: marking the appearance defects of the appearance of the panel according to the detection result of the defect detection on the appearance of the panel to be tested.

[0010] According to another aspect of the present invention, there is provided an intelligent visual inspection device for a panel, comprising: a classification module, a training module and a detection module; the classification module is used to obtain all types of appearance defect sample images of the panel, and classify the appearance defect types of all types of panel appearance defect sample images; the training module is used to perform deep learning network training based on all types of appearance defect sample images of the panel and the appearance defect type classification, and obtain an image model of all types of appearance defects of the panel through training; the detection module is used to perform defect detection on the appearance of the panel to be tested based on the image model of all types of appearance defects of the panel.

[0011] Among them, the classification module is specifically used to: obtain all types of appearance defect sample images of the panel, divide all types of appearance defect sample images into a preset number of regional image blocks, establish two-dimensional coordinates under the same coordinate standard with the center point of each regional image block, extract image features under each dimensional coordinate under the two-dimensional coordinates, fuse the image features with all types of appearance defect sample images of the panel, generate image fusion features, and classify the appearance defect types according to the image integration features.

[0012] Among them, the training module is specifically used to: according to all types of appearance defect sample images of the panel and the appearance defect type classification, grayscale the appearance defect sample images under each classification item, use Gaussian filtering to perform deep learning network training on the grayscale appearance defect sample images, train the frequency domain information of the grayscale appearance defect sample images, locate the defect area from the frequency domain information to form a defect area set under each classification item, extract the defect set of the defect area set under each classification item, and train the image model of all types of appearance defects of the panel according to the defect set.

[0013] The detection module is specifically used to: perform defect detection on the appearance image of the panel to be tested by mapping the appearance image to the corresponding defect name in the image model according to the image model of all types of appearance defects of the panel.

[0014] The panel intelligent visual inspection device further comprises a marking module for marking appearance defects of the panel according to the inspection result of defect inspection on the appearance of the panel to be inspected.

[0015] According to another aspect of the present invention, a computer device is provided, comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the panel intelligent visual inspection method as described in any one of the above items.

[0016] According to another aspect of the present invention, a computer-readable storage medium is provided, storing a computer program, wherein when the computer program is executed by a processor, the intelligent visual inspection method for a panel as described in any one of the above is implemented.

[0017] It can be found that the above scheme can obtain all types of appearance defect sample images of the panel, classify the appearance defect types of all types of panel appearance defect sample images, and perform deep learning network training based on all types of appearance defect sample images of the panel and the appearance defect type classification to obtain an image model of all types of appearance defects of the panel through training, and can perform defect detection on the appearance of the panel to be tested based on the image model of all types of appearance defects of the panel, thereby improving the efficiency and accuracy of panel appearance defect detection.

[0018] Furthermore, the above scheme can obtain all types of appearance defect sample images of the panel, divide all types of appearance defect sample images into a preset number of regional image blocks, establish two-dimensional coordinates under the same coordinate standard with the center point of each regional image block, extract the image features under each dimension of the two-dimensional coordinates, fuse the image features with all types of appearance defect sample images of the panel, generate image fusion features, and classify the appearance defect types according to the image integration features. The advantage of this is that by fusing the image features with all types of appearance defect sample images of the panel, rich image fusion features can be generated, and the subtle differences in panel appearance defects can be fully captured and analyzed, thereby improving the efficiency and accuracy of appearance defect type classification.

[0019] Furthermore, the above scheme can grayscale the appearance defect sample images under each classification item according to all types of appearance defect sample images of the panel and the classification of the appearance defect types, and use Gaussian filtering to perform deep learning network training on the grayscale appearance defect sample images to train the frequency domain information of the grayscale appearance defect sample images, locate the defect area from the frequency domain information to form a defect area set under each classification item, extract the defect set of the defect area set under each classification item, and train the image model of all types of appearance defects of the panel based on the defect set. The advantage of this is that by grayscale processing and Gaussian filtering of all types of appearance defect sample images of the panel, and using the deep learning network to train these processed images to extract frequency domain information, the defect area under each classification item can be accurately located and extracted, and then a comprehensive defect set can be constructed, which can significantly improve the efficiency and accuracy of training the image model of all types of appearance defects of the panel.

[0020] Furthermore, the above scheme can perform defect detection on the appearance image of the appearance of the panel to be tested by mapping it to the corresponding defect names in the image model based on the image model of all types of appearance defects of the panel. The advantage of this is that by performing defect detection on the appearance image of the appearance of the panel to be tested by mapping it to the corresponding defect names in the image model, various subtle appearance defects of the panel can be quickly identified and classified, the possibility of human intervention and misjudgment can be reduced, and the efficiency and accuracy of panel appearance defect detection can be improved.

[0021] Furthermore, the above scheme can mark the appearance defects of the panel according to the detection results of the defect detection on the appearance of the panel to be tested. The advantage of this is that the appearance defects of the panel can be marked, and the location, type and severity of the panel appearance defects can be clearly indicated, providing clear guidance for subsequent repair work of the panel appearance defects. At the same time, it also helps to accumulate panel appearance defect data, optimize the panel appearance defect detection algorithm, and further improve the efficiency and accuracy of panel appearance defect detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0023] Figure 1 It is a schematic diagram of the process of an embodiment of the panel intelligent visual inspection method of the present invention;

[0024] Figure 2 It is a flow chart of another embodiment of the panel intelligent visual inspection method of the present invention;

[0025] Figure 3 It is a structural schematic diagram of an embodiment of a panel intelligent visual inspection device of the present invention;

[0026] Figure 4 It is a structural schematic diagram of another embodiment of the panel intelligent visual inspection device of the present invention;

[0027] Figure 5 It is a structural schematic diagram of an embodiment of a computer device of the present invention. DETAILED DESCRIPTION

[0028] The present invention will be further described in detail below in conjunction with the accompanying drawings and examples. It is particularly noted that the following examples are only used to illustrate the present invention, but are not intended to limit the scope of the present invention. Similarly, the following examples are only partial embodiments of the present invention rather than all embodiments, and all other embodiments obtained by those of ordinary skill in the art without creative work are within the scope of protection of the present invention.

[0029] The present invention provides a panel intelligent visual inspection method, which can improve the efficiency and accuracy of panel appearance defect detection.

[0030] See also Figure 1 , Figure 1FIG. 1 is a flow chart of an embodiment of the panel intelligent visual inspection method of the present invention. It should be noted that if there is substantially the same result, the method of the present invention is not limited to Figure 1 The process sequence shown is limited. Figure 1 As shown, the method comprises the following steps:

[0031] S101: Acquire all types of panel appearance defect sample images, and classify all types of panel appearance defect sample images into appearance defect types.

[0032] The acquiring of all types of panel appearance defect sample images and classification of all types of panel appearance defect sample images by appearance defect type may include:

[0033] Acquire all types of appearance defect sample images of the panel, divide all types of appearance defect sample images into a preset number of regional image blocks, establish two-dimensional coordinates under the same coordinate standard with the center point of each regional image block, extract image features under each dimensional coordinate of the two-dimensional coordinate, fuse the image features with all types of appearance defect sample images of the panel to generate image fusion features, and classify the appearance defect types according to the image integration features. The advantage of this is that by fusing the image features with all types of appearance defect sample images of the panel, rich image fusion features can be generated, and the subtle differences in panel appearance defects can be fully captured and analyzed, thereby improving the efficiency and accuracy of appearance defect type classification.

[0034] S102: Perform deep learning network training based on sample images of all types of appearance defects of the panel and the classification of the appearance defect types, and obtain image models of all types of appearance defects of the panel through training.

[0035] Among them, the deep learning network training is performed according to the sample images of all types of appearance defects of the panel and the classification of the appearance defect types, and the image model of all types of appearance defects of the panel is obtained by training, which may include:

[0036] According to the appearance defect sample images of all types of the panel and the classification of the appearance defect types, the appearance defect sample images under each classification item are grayed, and the grayed appearance defect sample images are subjected to deep learning network training using Gaussian filtering to train the frequency domain information of the grayed appearance defect sample images, and the defect area is located from the frequency domain information to form a defect area set under each classification item, and the defect set of the defect area set under each classification item is extracted. Based on the defect set, an image model of all types of appearance defects of the panel is trained. The advantage of this is that by graying and Gaussian filtering all types of appearance defect sample images of the panel, and using the deep learning network to train these processed images to extract frequency domain information, the defect area under each classification item can be accurately located and extracted, thereby constructing a comprehensive defect set, which can significantly improve the efficiency and accuracy of the image model of all types of appearance defects of the panel trained.

[0037] S103: performing defect detection on the appearance of the panel to be tested according to the image model of all types of appearance defects of the panel.

[0038] Among them, the defect detection of the appearance of the panel to be tested based on the image model of all types of appearance defects of the panel may include:

[0039] Based on the image model of all types of appearance defects of the panel, defect detection is performed by mapping the appearance image of the appearance of the panel to be tested to the corresponding defect names in the image model. The advantage of this is that defect detection is performed by mapping the appearance image of the appearance of the panel to be tested to the corresponding defect names in the image model, so that various subtle appearance defects of the panel can be quickly identified and classified, the possibility of human intervention and misjudgment can be reduced, and the efficiency and accuracy of panel appearance defect detection can be improved.

[0040] After performing defect detection on the appearance of the panel to be tested according to the image model of all types of appearance defects of the panel, the method may further include:

[0041] According to the test results of defect detection on the appearance of the panel to be tested, the appearance defects of the panel are marked. The advantage of this is that the appearance defects of the panel can be marked, and the location, type and severity of the panel appearance defects can be clearly indicated, providing clear guidance for subsequent repair work of the panel appearance defects. It also helps to accumulate panel appearance defect data, optimize the panel appearance defect detection algorithm, and further improve the efficiency and accuracy of panel appearance defect detection.

[0042] It can be found that in this embodiment, all types of appearance defect sample images of the panel can be obtained, and the appearance defect types of all types of panel appearance defect sample images can be classified. Deep learning network training can be performed based on all types of appearance defect sample images of the panel and the appearance defect type classification to obtain an image model of all types of appearance defects of the panel. Moreover, based on the image model of all types of appearance defects of the panel, defect detection can be performed on the appearance of the panel to be tested, thereby improving the efficiency and accuracy of panel appearance defect detection.

[0043] Furthermore, in the present embodiment, all types of appearance defect sample images of the panel can be obtained, all types of appearance defect sample images can be divided into a preset number of regional image blocks, two-dimensional coordinates under the same coordinate standard are established with the center point of each regional image block, the image features under each dimension of the two-dimensional coordinates are extracted, the image features are fused with all types of appearance defect sample images of the panel to generate image fusion features, and the appearance defect types are classified according to the image integration features. The advantage of this is that by fusing the image features with all types of appearance defect sample images of the panel, rich image fusion features can be generated, and the subtle differences in the appearance defects of the panel can be fully captured and analyzed, thereby improving the efficiency and accuracy of the classification of appearance defect types.

[0044] Furthermore, in the present embodiment, the appearance defect sample images of all types of the panel and the classification of the appearance defect types can be grayed out for the appearance defect sample images under each classification item, and the grayed appearance defect sample images can be subjected to deep learning network training using Gaussian filtering to train the frequency domain information of the grayed appearance defect sample images, and the defect areas can be located from the frequency domain information to form a defect area set under each classification item, and the defect set of the defect area set under each classification item can be extracted. Based on the defect set, an image model of all types of appearance defects of the panel can be trained. The advantage of this is that by graying out and Gaussian filtering the appearance defect sample images of all types of the panel, and using the deep learning network to train these processed images to extract frequency domain information, the defect areas under each classification item can be accurately located and extracted, thereby constructing a comprehensive defect set, which can significantly improve the efficiency and accuracy of training the image model of all types of appearance defects of the panel.

[0045] Furthermore, in the present embodiment, defect detection can be performed by mapping the appearance image of the appearance of the panel to be tested to the corresponding defect names in the image model based on the image model of all types of appearance defects of the panel. The advantage of this is that defect detection can be performed by mapping the appearance image of the appearance of the panel to be tested to the corresponding defect names in the image model, so that various subtle appearance defects of the panel can be quickly identified and classified, reducing the possibility of human intervention and misjudgment, and improving the efficiency and accuracy of panel appearance defect detection.

[0046] See also Figure 2 , Figure 2 1 is a flow chart of another embodiment of the panel intelligent visual inspection method of the present invention. In this embodiment, the method includes the following steps:

[0047] S201: Acquire all types of panel appearance defect sample images, and classify all types of panel appearance defect sample images into appearance defect types.

[0048] This can be as described in S101 above and will not be elaborated here.

[0049] S202: Perform deep learning network training based on sample images of all types of appearance defects of the panel and the classification of the appearance defect types, and obtain image models of all types of appearance defects of the panel through training.

[0050] This can be as described in S102 above, and will not be elaborated here.

[0051] S203: performing defect detection on the appearance of the panel to be tested according to the image model of all types of appearance defects of the panel.

[0052] This can be as described in S103 above, and will not be elaborated here.

[0053] S204: marking appearance defects of the panel according to the result of the defect detection on the appearance of the panel to be tested.

[0054] It can be found that in this embodiment, the appearance defects of the panel can be marked according to the detection results of the defect detection on the appearance of the panel to be tested. The advantage of this is that the appearance defects of the panel can be marked, and the location, type and severity of the panel appearance defects can be clearly indicated, providing clear guidance for subsequent repair work of the panel appearance defects. At the same time, it also helps to accumulate panel appearance defect data, optimize the panel appearance defect detection algorithm, and further improve the efficiency and accuracy of panel appearance defect detection.

[0055] The present invention also provides a panel intelligent visual inspection device, which can improve the efficiency and accuracy of panel appearance defect detection.

[0056] See also Figure 3 , Figure 3 1 is a schematic diagram of the structure of an embodiment of the panel intelligent visual inspection device of the present invention. In this embodiment, the panel intelligent visual inspection device 30 includes a classification module 31, a training module 32 and a detection module 33.

[0057] The classification module 31 is used to obtain all types of panel appearance defect sample images and classify the appearance defect types of all types of panel appearance defect sample images.

[0058] The training module 32 is used to perform deep learning network training based on the sample images of all types of appearance defects of the panel and the classification of the appearance defect types, so as to obtain image models of all types of appearance defects of the panel through training.

[0059] The detection module 33 is used to perform defect detection on the appearance of the panel to be tested according to the image model of all types of appearance defects of the panel.

[0060] Optionally, the classification module 31 may be specifically used for:

[0061] Acquire all types of appearance defect sample images of the panel, divide all types of appearance defect sample images into a preset number of regional image blocks, establish two-dimensional coordinates under the same coordinate standard with the center point of each regional image block, extract the image features under each dimensional coordinate of the two-dimensional coordinate, fuse the image features with all types of appearance defect sample images of the panel, generate image fusion features, and classify the appearance defect types according to the image integration features.

[0062] Optionally, the training module 32 may be specifically used for:

[0063] According to all types of appearance defect sample images of the panel and the classification of the appearance defect types, the appearance defect sample images under each classification item are grayscaled, and the grayscaled appearance defect sample images are subjected to deep learning network training using a Gaussian filter method to train the frequency domain information of the grayscaled appearance defect sample images, and the defect areas are located from the frequency domain information to form a defect area set under each classification item, and a defect set of the defect area set under each classification item is extracted. Based on the defect set, an image model of all types of appearance defects of the panel is trained.

[0064] Optionally, the detection module 33 may be specifically used for:

[0065] According to the image model of all types of appearance defects of the panel, defect detection is performed by mapping the appearance image of the appearance of the panel to be tested to the corresponding defect name in the image model.

[0066] See also Figure 4 , Figure 4 4 is a schematic diagram of the structure of another embodiment of the panel intelligent visual inspection device of the present invention. Different from the previous embodiment, the panel intelligent visual inspection device 40 of this embodiment further includes a marking module 41.

[0067] The marking module 41 is used to mark appearance defects of the panel according to the inspection result of the defect inspection on the appearance of the panel to be inspected.

[0068] Each unit module of the panel intelligent visual inspection device 30 / 40 can respectively execute the corresponding steps in the above method embodiment, so each unit module will not be described in detail here. Please refer to the description of the above corresponding steps for details.

[0069] The present invention further provides a computer device, such as Figure 5 As shown, it includes: at least one processor 51; and a memory 52 that is communicatively connected to the at least one processor 51; wherein the memory 52 stores instructions that can be executed by the at least one processor 51, and the instructions are executed by the at least one processor 51 so that the at least one processor 51 can execute the above-mentioned panel intelligent visual detection method.

[0070] The memory 52 and the processor 51 are connected in a bus manner, and the bus may include any number of interconnected buses and bridges, and the bus connects various circuits of one or more processors 51 and the memory 52 together. The bus can also connect various other circuits such as peripheral devices, voltage regulators, and power management circuits, which are well known in the art and are therefore not further described herein. The bus interface provides an interface between the bus and the transceiver. The transceiver can be one element or multiple elements, such as multiple receivers and transmitters, providing a unit for communicating with various other devices on a transmission medium. The data processed by the processor 51 is transmitted on a wireless medium via an antenna, and further, the antenna also receives data and transmits the data to the processor 51.

[0071] The processor 51 is responsible for managing the bus and general processing, and can also provide various functions, including timing, peripheral interfaces, voltage regulation, power management and other control functions. The memory 52 can be used to store data used by the processor 51 when performing operations.

[0072] The present invention further provides a computer-readable storage medium storing a computer program, which implements the above method embodiment when executed by a processor.

[0073] It can be found that the above scheme can obtain all types of appearance defect sample images of the panel, classify the appearance defect types of all types of panel appearance defect sample images, and perform deep learning network training based on all types of appearance defect sample images of the panel and the appearance defect type classification to obtain an image model of all types of appearance defects of the panel through training, and can perform defect detection on the appearance of the panel to be tested based on the image model of all types of appearance defects of the panel, thereby improving the efficiency and accuracy of panel appearance defect detection.

[0074] Furthermore, the above scheme can obtain all types of appearance defect sample images of the panel, divide all types of appearance defect sample images into a preset number of regional image blocks, establish two-dimensional coordinates under the same coordinate standard with the center point of each regional image block, extract the image features under each dimension of the two-dimensional coordinates, fuse the image features with all types of appearance defect sample images of the panel, generate image fusion features, and classify the appearance defect types according to the image integration features. The advantage of this is that by fusing the image features with all types of appearance defect sample images of the panel, rich image fusion features can be generated, and the subtle differences in panel appearance defects can be fully captured and analyzed, thereby improving the efficiency and accuracy of appearance defect type classification.

[0075] Furthermore, the above scheme can grayscale the appearance defect sample images under each classification item according to all types of appearance defect sample images of the panel and the classification of the appearance defect types, and use Gaussian filtering to perform deep learning network training on the grayscale appearance defect sample images to train the frequency domain information of the grayscale appearance defect sample images, locate the defect area from the frequency domain information to form a defect area set under each classification item, extract the defect set of the defect area set under each classification item, and train the image model of all types of appearance defects of the panel based on the defect set. The advantage of this is that by grayscale processing and Gaussian filtering of all types of appearance defect sample images of the panel, and using the deep learning network to train these processed images to extract frequency domain information, the defect area under each classification item can be accurately located and extracted, and then a comprehensive defect set can be constructed, which can significantly improve the efficiency and accuracy of training the image model of all types of appearance defects of the panel.

[0076] Furthermore, the above scheme can perform defect detection on the appearance image of the appearance of the panel to be tested by mapping it to the corresponding defect names in the image model based on the image model of all types of appearance defects of the panel. The advantage of this is that by performing defect detection on the appearance image of the appearance of the panel to be tested by mapping it to the corresponding defect names in the image model, various subtle appearance defects of the panel can be quickly identified and classified, the possibility of human intervention and misjudgment can be reduced, and the efficiency and accuracy of panel appearance defect detection can be improved.

[0077] Furthermore, the above scheme can mark the appearance defects of the panel according to the detection results of the defect detection on the appearance of the panel to be tested. The advantage of this is that the appearance defects of the panel can be marked, and the location, type and severity of the panel appearance defects can be clearly indicated, providing clear guidance for subsequent repair work of the panel appearance defects. At the same time, it also helps to accumulate panel appearance defect data, optimize the panel appearance defect detection algorithm, and further improve the efficiency and accuracy of panel appearance defect detection.

[0078] In the several embodiments provided by the present invention, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device implementation described above is only schematic, for example, the division of modules or units is only a logical function division, and there may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be an indirect coupling or communication connection through some interfaces, devices or units, which can be electrical, mechanical or other forms.

[0079] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the present embodiment.

[0080] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0081] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions for a computer device (which can be a personal computer, server, or network device, etc.) or a processor (processor) to perform all or part of the steps of each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk and other media that can store program codes.

[0082] The above descriptions are only some embodiments of the present invention, and are not intended to limit the protection scope of the present invention. Any equivalent device or equivalent process transformation made using the contents of the present invention specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A panel intelligent visual inspection method, characterized in that: include: Acquire all types of panel appearance defect sample images, and classify the appearance defect types of all types of panel appearance defect sample images; Performing deep learning network training based on all types of appearance defect sample images of the panel and the appearance defect type classification to obtain image models of all types of appearance defects of the panel; Defect detection is performed on the appearance of the panel to be tested according to the image models of all types of appearance defects of the panel.

2. The panel intelligent visual inspection method according to claim 1, characterized in that: The acquiring of all types of panel appearance defect sample images and classifying the appearance defect types of all types of panel appearance defect sample images include: Acquire all types of appearance defect sample images of the panel, divide all types of appearance defect sample images into a preset number of regional image blocks, establish two-dimensional coordinates under the same coordinate standard with the center point of each regional image block, extract image features under each dimensional coordinate under the two-dimensional coordinates, fuse the image features with all types of appearance defect sample images of the panel, generate image fusion features, and classify the appearance defect types according to the image integration features.

3. The panel intelligent visual inspection method according to claim 1, characterized in that: The deep learning network training is performed according to the sample images of all types of appearance defects of the panel and the classification of the appearance defect types, and the image model of all types of appearance defects of the panel is obtained through training, including: According to the appearance defect sample images of all types of the panel and the appearance defect type classification, the appearance defect sample images under each classification item are grayed, and the grayed appearance defect sample images are trained with a deep learning network using a Gaussian filter method to train the frequency domain information of the grayed appearance defect sample images, and the defect areas are located from the frequency domain information to form a defect area set under each classification item, and a defect set of the defect area set under each classification item is extracted, and based on the defect set, an image model of all types of appearance defects of the panel is trained.

4. The panel intelligent visual inspection method according to claim 1, characterized in that: The performing defect detection on the appearance of the panel to be tested according to the image model of all types of appearance defects of the panel comprises: According to the image model of all types of appearance defects of the panel, defect detection is performed by mapping the appearance image of the appearance of the panel to be tested to the corresponding defect name in the image model.

5. The panel intelligent visual inspection method according to claim 1, characterized in that: After performing defect detection on the appearance of the panel to be tested according to the image model of all types of appearance defects of the panel, the method further includes: According to the inspection result of the defect inspection on the appearance of the panel to be inspected, the appearance defects of the appearance of the panel are marked.

6. A panel intelligent visual inspection device, characterized in that: include: Classification module, training module and detection module; The classification module is used to obtain all types of panel appearance defect sample images and classify the appearance defect types of all types of panel appearance defect sample images; The training module is used to perform deep learning network training based on the sample images of all types of appearance defects of the panel and the classification of the appearance defect types, so as to obtain image models of all types of appearance defects of the panel through training; The detection module is used to perform defect detection on the appearance of the panel to be tested according to the image model of all types of appearance defects of the panel.

7. The panel intelligent visual inspection device according to claim 6, characterized in that: The classification module is specifically used for: Acquire all types of appearance defect sample images of the panel, divide all types of appearance defect sample images into a preset number of regional image blocks, establish two-dimensional coordinates under the same coordinate standard with the center point of each regional image block, extract image features under each dimensional coordinate under the two-dimensional coordinates, fuse the image features with all types of appearance defect sample images of the panel, generate image fusion features, and classify the appearance defect types according to the image integration features.

8. The panel intelligent visual inspection device according to claim 6, characterized in that: The training module is specifically used for: According to the appearance defect sample images of all types of the panel and the appearance defect type classification, the appearance defect sample images under each classification item are grayed, and the grayed appearance defect sample images are trained with a deep learning network using a Gaussian filter method to train the frequency domain information of the grayed appearance defect sample images, and the defect areas are located from the frequency domain information to form a defect area set under each classification item, and a defect set of the defect area set under each classification item is extracted, and based on the defect set, an image model of all types of appearance defects of the panel is trained.

9. A computer device, characterized in that: include: at least one processor; And, a memory communicatively connected to at least one processor; wherein the memory stores instructions executable by at least one processor, and the instructions are executed by at least one processor so that at least one processor can execute the panel intelligent visual inspection method as described in any one of claims 1 to 5.

10. A computer-readable storage medium, characterized in that: A computer program is stored, and when the computer program is executed by a processor, the panel intelligent visual inspection method according to any one of claims 1 to 5 is implemented.

Images