Image detection method and device, equipment and medium
By training the feature values of the display panel image using a deep learning model, the problem of image defocusing in automated optical inspection equipment was solved, improving the efficiency and accuracy of defect detection.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- YUNGU GUAN TECH CO LTD
- Filing Date
- 2026-01-30
- Publication Date
- 2026-05-15
AI Technical Summary
Existing automated optical inspection equipment is prone to image defocusing in display panel defect detection, affecting inspection efficiency and accuracy.
A deep learning model is used to train feature values of known out-of-focus and non-out-of-focus images. By extracting multi-dimensional feature values of the training images, such as frequency domain grayscale feature values, Laplacian values, and color temperature values, it is determined whether the target image is out of focus, and a reshoot is performed when out-of-focus is detected.
It improves the efficiency and accuracy of defect detection and reduces economic losses caused by image defocusing.
Smart Images
Figure CN122048879A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of display technology, and in particular to an image detection method, apparatus, device, and medium. Background Technology
[0002] OLED (Organic Light-Emitting Diode) is an active-matrix light-emitting device with a sandwich structure consisting of multiple organic layers and electrodes on both sides. Currently, AMOLED (Active-Matrix Organic Light-Emitting Diode) based display panels have been commercialized in fields such as smartphones, watches, and laptops.
[0003] After the display panel is manufactured, defect detection is usually required. Currently, Automated Optical Inspection (AOI) equipment is used to obtain optical images of the display panel, and defects are detected based on these images. However, the images obtained by AOI equipment can be out of focus, which affects the efficiency and accuracy of defect detection. Summary of the Invention
[0004] In view of this, the purpose of this application is to propose an image detection method, apparatus, equipment and medium that can accurately determine whether an image is out of focus, thereby improving the efficiency and accuracy of defect detection.
[0005] To achieve the above objectives, this application provides an image detection method, which includes: Acquire training images, which include known out-of-focus images and known non-out-of-focus images; Extract the feature values of the training images; The deep learning model is trained by taking the feature values as input and whether the corresponding image is out of focus as output. A trained deep learning model is used to detect whether a target image is out of focus.
[0006] In one implementation, the training images are uncompressed raw images; Preferably, the original image is obtained by an automated optical inspection device; Preferably, the known out-of-focus images and known non-out-of-focus images in the original image are determined by manual identification.
[0007] In one embodiment, extracting the feature values of the training image includes: Extract feature values from multiple different dimensions of the training image; Preferably, the feature values include one or more of the following: grayscale feature values of the training image in the frequency domain, the Laplacian value of the training image, and the color temperature value of the training image.
[0008] In one embodiment, the grayscale feature values of the training image in the frequency domain are obtained by performing a Fourier transform on the training image. Preferably, the grayscale feature values of the training image in the frequency domain include one or more of the following: maximum value, minimum value, range, median, mode, and standard deviation.
[0009] In one embodiment, the Laplacian value of the training image is calculated using the Laplacian operator; Preferably, the step of calculating the Laplacian value of the training image using the Laplacian operator includes: Convert the training images into grayscale images; Iterate through each pixel of the grayscale image. For each pixel, obtain the grayscale value of the pixel and its neighboring pixels according to the discretized Laplacian operator, and calculate the Laplacian value of each pixel based on the grayscale value.
[0010] In one embodiment, the calculation process of the color temperature value of the training image includes: Calculate the normalized RGB value of each pixel in the training image; The color coordinates of each pixel in the training image are calculated based on the normalized RGB values; Calculate the chromaticity coordinates of each pixel in the training image based on the chromaticity coordinates; The color temperature value of each pixel in the training image is calculated based on the chromaticity coordinates of each pixel in the training image; The average color temperature value is calculated based on the color temperature value of each pixel in the training image, and is used as the color temperature value of the training image.
[0011] In one embodiment, the step of detecting whether the target image is out of focus using a trained deep learning model further includes: When the target image is detected to be out of focus, a re-shot of the out-of-focus target image is performed; Preferably, both the training image and the target image are display panel images. The step of detecting whether the target image is out of focus using a trained deep learning model further includes: When the target image is detected to be in focus, defect detection is performed on the display panel based on the in-focus target image.
[0012] Based on the same inventive concept, this application also provides an image detection device, which includes: The image acquisition module is used to acquire training images, which include known out-of-focus images and known non-out-of-focus images; The feature extraction module is used to extract feature values from the training images; The model training module is used to train the deep learning model by taking the feature values as input and whether the corresponding image is out of focus as output. The detection module is used to detect whether the target image is out of focus using a trained deep learning model.
[0013] Based on the same inventive concept, this application also provides an electronic device, which includes a processor, a memory, and a program or instructions stored in the memory and executable on the processor, wherein the program or instructions, when executed by the processor, implement the steps of the image detection method described above.
[0014] Based on the same inventive concept, this application also provides a computer-readable storage medium storing a program or instructions that, when executed by a processor, implement the steps of the image detection method described above.
[0015] Compared with existing technologies, the image detection method, apparatus, device and medium provided in this application use known out-of-focus images and known non-out-of-focus images as training images, extract feature values from the training images, use the feature values as input and the corresponding image out of focus as output to train a deep learning model, and use the trained deep learning model to detect whether the target image is out of focus. This can accurately determine whether the image is out of focus, thereby improving the efficiency and accuracy of defect detection. Attached Figure Description
[0016] To more clearly illustrate the technical solutions in this application or related technologies, the drawings used in the description of the embodiments or related technologies will be briefly introduced below. Obviously, the drawings described below are only embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0017] Figure 1 This is a schematic diagram of an image being out of focus. Figure 2 This is a schematic diagram showing that the image is not out of focus; Figure 3 A flowchart illustrating an image detection method provided in one embodiment of this application; Figure 4 A flowchart of an image detection method provided in another embodiment of this application; Figure 5 A flowchart of an image detection method provided in another embodiment of this application; Figure 6 A flowchart of an image detection method provided in another embodiment of this application; Figure 7 A schematic diagram of an image detection device provided in another embodiment of this application; Figure 8 This is a schematic diagram of a more specific electronic device hardware structure provided for another embodiment of this application. Detailed Implementation
[0018] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with specific embodiments and the accompanying drawings.
[0019] It should be noted that, unless otherwise defined, the technical or scientific terms used in the embodiments of this application should have the ordinary meaning understood by one of ordinary skill in the art to which this application pertains. The terms "first," "second," and similar terms used in the embodiments of this application do not indicate any order, quantity, or importance, but are merely used to distinguish different components. Terms such as "comprising" or "including" mean that the element or object preceding the word encompasses the elements or objects listed after the word and their equivalents, without excluding other elements or objects. Terms such as "connected" or "linked" are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect. Terms such as "upper," "lower," "left," and "right" are only used to indicate relative positional relationships; when the absolute position of the described object changes, the relative positional relationship may also change accordingly.
[0020] Currently, display panel manufacturing involves long processes, numerous production sites, and complex techniques, resulting in hundreds of types of defects. Before proceeding to the next stage, products must undergo defect detection to discard substandard items and repairable defects. If product defects are not detected and flow into the next stage, it can lead to mass scrapping and even greater economic losses.
[0021] However, as the PPI (Pixels Per Inch) of products continues to increase, images acquired by automated optical inspection equipment may become out of focus due to hardware differences and other reasons, thus affecting the efficiency and accuracy of defect detection. (Refer to...) Figure 1 As shown, this is an example of an automated optical inspection device acquiring a defocused image. (Refer to...) Figure 2 As shown, this is an image obtained by an automated optical inspection device that is not out of focus.
[0022] Based on this, this application provides an image detection scheme to solve the above problems, as detailed in the following embodiments.
[0023] Reference Figure 3 As shown, one embodiment of this application discloses an image detection method, which includes the following steps: Step S10: Obtain training images, which include known out-of-focus images and known focused images. Optionally, the training images are images of the display panel, and the presence of defects in the display panel can be determined by examining the images of the focused display panel.
[0024] Step S20: Extract feature values from the training images. The known out-of-focus images and the known uncluttered images have different feature values or feature values within different ranges.
[0025] Step S30: Train the deep learning model using the feature values as input and the corresponding image focus level as output. Deep learning models are a type of machine learning model based on artificial neural networks. By constructing a multi-layered network structure, they automatically learn features and patterns from large amounts of data to perform tasks such as data classification, prediction, and generation. Common deep learning models include feedforward neural networks, convolutional neural networks, recurrent neural networks, and generative adversarial networks.
[0026] Step S40: Detect whether the target image is out of focus using the trained deep learning model. Specifically, extract the feature values of the target image and input them into the trained deep learning model to output whether the target image is out of focus.
[0027] The image detection method provided in this embodiment uses known out-of-focus images and known non-out-of-focus images as training images, extracts feature values from the training images, uses the feature values as input and the corresponding image out of focus as output to train a deep learning model, and uses the trained deep learning model to detect whether the target image is out of focus. This can accurately determine whether the image is out of focus, thereby improving the efficiency and accuracy of defect detection.
[0028] In one embodiment, in step S10, the training image is an uncompressed original image. The uncompressed original image differs from the compressed image; the uncompressed original image has higher precision and accuracy, which is beneficial for improving detection accuracy.
[0029] Preferably, the original image is obtained by taking pictures using an automated optical inspection device to ensure the accuracy of the original image.
[0030] Preferably, the known out-of-focus images and known non-out-of-focus images in the original image are identified manually, which ensures the accuracy of the identification and thus improves the precision of the detection.
[0031] In one embodiment, step S20, extracting feature values from the training images, includes extracting feature values from multiple different dimensions of the training images. By using feature values from multiple different dimensions to train the deep learning model, the deep learning model can learn more general image features and patterns, rather than simply memorizing specific images from the training data. This allows the deep learning model to accurately judge and classify new, unseen images based on its learned multi-dimensional feature knowledge, reducing overfitting and improving the generalization performance of the deep learning model.
[0032] Preferably, the feature values include one or more of the following: grayscale feature values of the training image in the frequency domain, the Laplacian value of the training image, and the color temperature value of the training image. For example, the feature values include grayscale feature values of the training image in the frequency domain and the Laplacian value of the training image; or, the feature values include grayscale feature values of the training image in the frequency domain and the color temperature value of the training image; or, the feature values include grayscale feature values of the training image in the frequency domain, the Laplacian value of the training image, and the color temperature value of the training image.
[0033] In one embodiment, the grayscale feature values of the training image in the frequency domain are obtained by performing a Fourier transform on the training image. The Fourier transform is based on the principle of Fourier series, which states that any periodic function can be represented as a linear combination of a series of sine and cosine functions. For an image, the changes in its pixel values can be viewed as a two-dimensional signal, which is decomposed into a superposition of sine and cosine functions of different frequencies through the Fourier transform. In the frequency domain, each frequency component has a corresponding amplitude and phase, and the grayscale feature values are mainly related to the amplitude; the magnitude of the amplitude reflects the degree of contribution of that frequency component to the original image.
[0034] Specifically, the formula for the two-dimensional Fourier transform is: ; Where f(x,y) is the image grayscale value function; F(u,v) is the transformed frequency domain function; u and v are frequency domain coordinates, and j is the imaginary unit.
[0035] In digital image processing, the Discrete Fourier Transform (DFT) is commonly used as an approximation. For an M*N image, the formula for the Discrete Fourier Transform is: ; In practice, the Fast Fourier Transform (FFT) algorithm is often used to efficiently compute the DFT.
[0036] Preferably, the grayscale feature values of the training image in the frequency domain include one or more of the following: maximum value, minimum value, range, median, mode, and standard deviation. Having multiple grayscale feature values can better train the deep learning model and improve the accuracy of image detection.
[0037] For example, the grayscale feature values of the training image in the frequency domain include the maximum value, minimum value, and range; or, the grayscale feature values of the training image in the frequency domain include the maximum value, minimum value, range, median, and mode; or, the grayscale feature values of the training image in the frequency domain include the maximum value, minimum value, range, median, mode, and standard deviation.
[0038] In one embodiment, the Laplacian value of the training image is calculated using the Laplacian operator. Specifically, the second derivative property of the Laplacian operator is used to detect abrupt changes in grayscale values in the image, thereby identifying image edges. Since the rate of grayscale change is larger at edges, their Laplacian values will change significantly. By thresholding the Laplacian values, edge information of the image can be extracted.
[0039] Reference Figure 4 As shown, preferably, the above-mentioned calculation of the Laplacian value of the training image using the Laplacian operator includes the following steps: Step S21: Convert the training images into grayscale images. Color images typically contain multiple channels (e.g., an RGB image has red, green, and blue channels), each with an independent pixel value. After conversion to grayscale, the image is represented by only one channel, with each pixel having only one grayscale value. This significantly reduces computational complexity, as the Laplacian operator only needs to operate on the pixel value of a single channel, eliminating the need to process multiple channels separately, thus improving computational efficiency and reducing computational resource consumption.
[0040] Step S22: Traverse each pixel of the grayscale image. For each pixel, obtain the grayscale value of the pixel and its neighboring pixels according to the discretized Laplacian operator, and calculate the Laplacian value of each pixel based on the grayscale value.
[0041] Specifically, the Laplace operator is defined as follows: ; in, Representing the Laplacian operator, f(x,y) is the grayscale value function of the image. and Let represent the second-order partial derivatives of the image in the x and y directions, respectively. By calculating the sum of these two second-order partial derivatives, we can identify edge and texture features in the image. Intuitively, the sum of the second-order partial derivatives will be larger in areas of the image where grayscale values change drastically (such as edges).
[0042] In digital image processing, since images are composed of discrete pixels, the Laplacian operator needs to be discretized and approximated by convolution operations. Its discrete form is: ; This formula demonstrates the use of convolution operations on images to extract edge information. The standard convolution template for the Laplacian operator is: ; In practical applications, this convolution kernel is applied to an image to detect edges by calculating the difference between the center pixel and its neighboring pixels.
[0043] Reference Figure 5 As shown, in one embodiment, the calculation process of the color temperature value of the training image includes the following steps: Step S23: Calculate the normalized RGB value of each pixel in the training image.
[0044] The RGB values represent the intensity of each pixel across the three color channels: R (red), G (green), and B (blue).
[0045] Specifically, the normalized RGB values are r, g, and b, and their calculation formulas are as follows: r=R / (R+B+G); g=G / (R+B+G); b=B / (R+B+G); Step S24: Calculate the color coordinates of each pixel in the training image based on the normalized RGB values.
[0046] Optionally, the formula for calculating the color coordinates (x, y) is: ; ; Step S25: Calculate the chromaticity coordinates of each pixel in the training image based on the chromaticity coordinates.
[0047] Optionally, the formula for calculating the chromaticity coordinates (u,v) is: ; ; Step S26: Calculate the color temperature value of each pixel in the training image based on the chromaticity coordinates of each pixel in the training image.
[0048] Optionally, the chromaticity coordinates can be transformed using the following formula: ; ; Then, calculate the color temperature value t using the following formula: ; Step S27: Calculate the average color temperature value based on the color temperature value of each pixel in the training image, and use it as the color temperature value of the training image.
[0049] Reference Figure 6 As shown, in one embodiment, step S40, which uses a trained deep learning model to detect whether the target image is out of focus, further includes the following steps: Step S50: When the target image is detected to be out of focus, a re-image of the out-of-focus target image is taken to obtain a focused image for subsequent defect detection.
[0050] Preferably, both the training image and the target image are display panel images; step S40, using the trained deep learning model to detect whether the target image is out of focus, further includes the following steps: Step S60: When the target image is detected to be in focus, perform defect detection on the display panel based on the in-focus target image.
[0051] Furthermore, defect detection can identify defects in display panels, allowing for the repair or rejection of defective panels. This prevents defective panels from flowing into the next process, avoids mass scrapping, and reduces economic losses.
[0052] It should be noted that the method in this embodiment can be executed by a single device, such as a computer or server. The method can also be applied in a distributed scenario, where multiple devices cooperate to complete the task. In such a distributed scenario, one of these devices may execute only one or more steps of the method in this embodiment, and the multiple devices will interact with each other to complete the method described.
[0053] It should be noted that the above description describes some embodiments of this application. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recorded in the claims can be performed in a different order than that shown in the above embodiments and still achieve the desired result. Furthermore, the processes depicted in the drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.
[0054] Reference Figure 7 As shown, based on the same inventive concept, and corresponding to any of the above embodiments, this application also provides an image detection device, which includes the following modules: The image acquisition module is used to acquire training images, which include known out-of-focus images and known non-out-of-focus images; The feature extraction module is used to extract feature values from the training images; The model training module is used to train the deep learning model by taking the feature values as input and whether the corresponding image is out of focus as output. The detection module is used to detect whether the target image is out of focus using a trained deep learning model.
[0055] The image detection device provided in this embodiment uses known out-of-focus images and known non-out-of-focus images as training images, extracts feature values from the training images, uses the feature values as input and the corresponding image out of focus as output to train a deep learning model, and uses the trained deep learning model to detect whether the target image is out of focus. This can accurately determine whether the image is out of focus, thereby improving the efficiency and accuracy of defect detection.
[0056] In one embodiment, the feature extraction module includes: The grayscale feature value calculation module is used to calculate the grayscale feature values of the training image in the frequency domain. The Laplacian value calculation module is used to calculate the Laplacian value of the training image; The color temperature calculation module is used to calculate the color temperature value of the training image.
[0057] Of course, when implementing this application, the functions of each module can be implemented in one or more software and / or hardware.
[0058] The apparatus of the above embodiments is used to implement the corresponding image detection method in any of the foregoing embodiments, and has the beneficial effects of the corresponding method embodiments, which will not be repeated here.
[0059] Based on the same inventive concept, corresponding to any of the above embodiments, this application also provides an electronic device, which includes a processor, a memory, and a program or instructions stored in the memory and executable on the processor, wherein the program or instructions, when executed by the processor, implement the steps of any of the above-described image detection methods.
[0060] Figure 8 This embodiment illustrates a more specific hardware structure of an electronic device, which may include a processor 1010, a memory 1020, an input / output interface 1030, a communication interface 1040, and a bus 1050. The processor 1010, memory 1020, input / output interface 1030, and communication interface 1040 are interconnected internally via the bus 1050.
[0061] The processor 1010 can be implemented using a general-purpose CPU (Central Processing Unit), microprocessor, application-specific integrated circuit (ASIC), or one or more integrated circuits, and is used to execute relevant programs to implement the technical solutions provided in the embodiments of this specification.
[0062] The memory 1020 can be implemented in the form of ROM (Read Only Memory), RAM (Random Access Memory), static storage device, dynamic storage device, etc. The memory 1020 can store the operating system and other applications. When the technical solutions provided in the embodiments of this specification are implemented by software or firmware, the relevant program code is stored in the memory 1020 and is called and executed by the processor 1010.
[0063] The input / output interface 1030 is used to connect input / output modules to realize information input and output. The input / output modules can be configured as components in the device (not shown in the figure) or externally connected to the device to provide corresponding functions. Input devices may include keyboards, mice, touch screens, microphones, various sensors, etc., and output devices may include displays, speakers, vibrators, indicator lights, etc.
[0064] The communication interface 1040 is used to connect a communication module (not shown in the figure) to enable communication between this device and other devices. The communication module can communicate via wired means (such as USB, Ethernet cable, etc.) or wireless means (such as mobile network, WIFI, Bluetooth, etc.).
[0065] Bus 1050 includes a pathway for transmitting information between various components of the device, such as processor 1010, memory 1020, input / output interface 1030, and communication interface 1040.
[0066] It should be noted that although the above-described device only shows the processor 1010, memory 1020, input / output interface 1030, communication interface 1040, and bus 1050, in specific implementations, the device may also include other components necessary for normal operation. Furthermore, those skilled in the art will understand that the above-described device may only include the components necessary for implementing the embodiments of this specification, and not necessarily all the components shown in the figures.
[0067] The electronic devices described above are used to implement the corresponding image detection methods in any of the foregoing embodiments and have the beneficial effects of the corresponding method embodiments, which will not be repeated here.
[0068] Based on the same inventive concept, corresponding to any of the above embodiments, this application also provides a computer-readable storage medium storing a program or instructions that, when executed by a processor, implement the steps of the image detection method described in any of the above claims.
[0069] The computer-readable medium of this embodiment includes permanent and non-permanent, removable and non-removable media, and information storage can be implemented by any method or technology. Information can be computer-readable instructions, data structures, program modules, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic magnetic disk storage or other magnetic storage devices, or any other non-transfer medium that can be used to store information accessible by a computing device.
[0070] The computer instructions stored in the storage medium of the above embodiments are used to cause the computer to execute the image detection method as described in any of the above embodiments, and have the beneficial effects of the corresponding method embodiments, which will not be repeated here.
[0071] Although this application has been described in conjunction with specific embodiments thereof, many substitutions, modifications and variations of these embodiments will be apparent to those skilled in the art from the foregoing description.
[0072] The embodiments of this application are intended to cover all such substitutions, modifications, and variations that fall within the broad scope of the appended claims. Therefore, any omissions, modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the embodiments of this application should be included within the protection scope of this application.
[0073] It should be noted that the above description describes some embodiments of this application. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recorded in the claims can be performed in a different order than that shown in the above embodiments and still achieve the desired result. Furthermore, the processes depicted in the drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.
[0074] The embodiments of this application are intended to cover all such substitutions, modifications, and variations that fall within the broad scope of the appended claims. Therefore, any omissions, modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the embodiments of this application should be included within the protection scope of this application.
Claims
1. An image detection method, characterized in that, include: Acquire training images, which include known out-of-focus images and known non-out-of-focus images; Extract the feature values of the training images; The deep learning model is trained by taking the feature values as input and whether the corresponding image is out of focus as output. A trained deep learning model is used to detect whether a target image is out of focus.
2. The image detection method as described in claim 1, characterized in that, The training images are uncompressed raw images; Preferably, the original image is obtained by an automated optical inspection device; Preferably, the known out-of-focus images and known non-out-of-focus images in the original image are determined by manual identification.
3. The image detection method as described in claim 1, characterized in that, The extraction of feature values from the training image includes: Extract feature values from multiple different dimensions of the training image; Preferably, the feature values include one or more of the following: grayscale feature values of the training image in the frequency domain, the Laplacian value of the training image, and the color temperature value of the training image.
4. The image detection method as described in claim 3, characterized in that, The grayscale feature values of the training image in the frequency domain are obtained by performing a Fourier transform on the training image. Preferably, the grayscale feature values of the training image in the frequency domain include one or more of the following: maximum value, minimum value, range, median, mode, and standard deviation.
5. The image detection method as described in claim 3, characterized in that, The Laplacian value of the training image is calculated using the Laplacian operator; Preferably, the step of calculating the Laplacian value of the training image using the Laplacian operator includes: Convert the training images into grayscale images; Iterate through each pixel of the grayscale image. For each pixel, obtain the grayscale value of the pixel and its neighboring pixels according to the discretized Laplacian operator, and calculate the Laplacian value of each pixel based on the grayscale value.
6. The image detection method as described in claim 3, characterized in that, The calculation process for the color temperature value of the training image includes: Calculate the normalized RGB value of each pixel in the training image; The color coordinates of each pixel in the training image are calculated based on the normalized RGB values; Calculate the chromaticity coordinates of each pixel in the training image based on the chromaticity coordinates; The color temperature value of each pixel in the training image is calculated based on the chromaticity coordinates of each pixel in the training image; The average color temperature value is calculated based on the color temperature value of each pixel in the training image, and is used as the color temperature value of the training image.
7. The image detection method as described in claim 1, characterized in that, The step of detecting whether the target image is out of focus using a trained deep learning model also includes: When the target image is detected to be out of focus, a re-shot of the out-of-focus target image is performed; Preferably, both the training image and the target image are display panel images. The step of detecting whether the target image is out of focus using a trained deep learning model further includes: When the target image is detected to be in focus, defect detection is performed on the display panel based on the in-focus target image.
8. An image detection device, characterized in that, include: The image acquisition module is used to acquire training images, which include known out-of-focus images and known non-out-of-focus images; The feature extraction module is used to extract feature values from the training images; The model training module is used to train the deep learning model by taking the feature values as input and whether the corresponding image is out of focus as output. The detection module is used to detect whether the target image is out of focus using a trained deep learning model.
9. An electronic device, characterized in that, It includes a processor, a memory, and a program or instructions stored in the memory and executable on the processor, wherein the program or instructions, when executed by the processor, implement the steps of the image detection method as described in any one of claims 1 to 7.
10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a program or instructions that, when executed by a processor, implement the steps of the image detection method as described in any one of claims 1 to 7.