WO2020007118A1 - 显示屏外围电路检测方法、装置、电子设备及存储介质 - Google Patents

显示屏外围电路检测方法、装置、电子设备及存储介质 Download PDF

Info

Publication number
WO2020007118A1
WO2020007118A1 PCT/CN2019/085912 CN2019085912W WO2020007118A1 WO 2020007118 A1 WO2020007118 A1 WO 2020007118A1 CN 2019085912 W CN2019085912 W CN 2019085912W WO 2020007118 A1 WO2020007118 A1 WO 2020007118A1
Authority
WO
WIPO (PCT)
Prior art keywords
peripheral circuit
defect detection
image
display
defect
Prior art date
Application number
PCT/CN2019/085912
Other languages
English (en)
French (fr)
Inventor
文亚伟
冷家冰
刘明浩
徐玉林
郭江亮
李旭
Original Assignee
北京百度网讯科技有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 北京百度网讯科技有限公司 filed Critical 北京百度网讯科技有限公司
Priority to JP2019563795A priority Critical patent/JP7025452B2/ja
Priority to KR1020197034316A priority patent/KR102320371B1/ko
Publication of WO2020007118A1 publication Critical patent/WO2020007118A1/zh
Priority to US16/995,898 priority patent/US20200380899A1/en

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F3/00—Input arrangements for transferring data to be processed into a form capable of being handled by the computer; Output arrangements for transferring data from processing unit to output unit, e.g. interface arrangements
    • G06F3/14—Digital output to display device ; Cooperation and interconnection of the display device with other functional units
    • G06F3/147—Digital output to display device ; Cooperation and interconnection of the display device with other functional units using display panels
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
    • G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
    • G01N21/84—Systems specially adapted for particular applications
    • G01N21/88—Investigating the presence of flaws or contamination
    • G01N21/8851—Scan or image signal processing specially adapted therefor, e.g. for scan signal adjustment, for detecting different kinds of defects, for compensating for structures, markings, edges
    • G—PHYSICS
    • G09—EDUCATION; CRYPTOGRAPHY; DISPLAY; ADVERTISING; SEALS
    • G09G—ARRANGEMENTS OR CIRCUITS FOR CONTROL OF INDICATING DEVICES USING STATIC MEANS TO PRESENT VARIABLE INFORMATION
    • G09G3/00—Control arrangements or circuits, of interest only in connection with visual indicators other than cathode-ray tubes
    • G09G3/006—Electronic inspection or testing of displays and display drivers, e.g. of LED or LCD displays
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
    • G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
    • G01N21/84—Systems specially adapted for particular applications
    • G01N21/88—Investigating the presence of flaws or contamination
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
    • G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
    • G01N21/84—Systems specially adapted for particular applications
    • G01N21/88—Investigating the presence of flaws or contamination
    • G01N21/95—Investigating the presence of flaws or contamination characterised by the material or shape of the object to be examined
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
    • G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
    • G01N21/84—Systems specially adapted for particular applications
    • G01N21/88—Investigating the presence of flaws or contamination
    • G01N21/95—Investigating the presence of flaws or contamination characterised by the material or shape of the object to be examined
    • G01N21/956—Inspecting patterns on the surface of objects
    • G01N21/95607—Inspecting patterns on the surface of objects using a comparative method
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
    • G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
    • G01R31/28—Testing of electronic circuits, e.g. by signal tracer
    • G—PHYSICS
    • G02—OPTICS
    • G02F—OPTICAL DEVICES OR ARRANGEMENTS FOR THE CONTROL OF LIGHT BY MODIFICATION OF THE OPTICAL PROPERTIES OF THE MEDIA OF THE ELEMENTS INVOLVED THEREIN; NON-LINEAR OPTICS; FREQUENCY-CHANGING OF LIGHT; OPTICAL LOGIC ELEMENTS; OPTICAL ANALOGUE/DIGITAL CONVERTERS
    • G02F1/00—Devices or arrangements for the control of the intensity, colour, phase, polarisation or direction of light arriving from an independent light source, e.g. switching, gating or modulating; Non-linear optics
    • G02F1/01—Devices or arrangements for the control of the intensity, colour, phase, polarisation or direction of light arriving from an independent light source, e.g. switching, gating or modulating; Non-linear optics for the control of the intensity, phase, polarisation or colour 
    • G02F1/13—Devices or arrangements for the control of the intensity, colour, phase, polarisation or direction of light arriving from an independent light source, e.g. switching, gating or modulating; Non-linear optics for the control of the intensity, phase, polarisation or colour  based on liquid crystals, e.g. single liquid crystal display cells
    • G02F1/1306—Details
    • G02F1/1309—Repairing; Testing
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/0002—Inspection of images, e.g. flaw detection
    • G06T7/0004—Industrial image inspection
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/0002—Inspection of images, e.g. flaw detection
    • G06T7/0004—Industrial image inspection
    • G06T7/0008—Industrial image inspection checking presence/absence
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
    • G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
    • G01N21/84—Systems specially adapted for particular applications
    • G01N21/88—Investigating the presence of flaws or contamination
    • G01N21/8851—Scan or image signal processing specially adapted therefor, e.g. for scan signal adjustment, for detecting different kinds of defects, for compensating for structures, markings, edges
    • G01N2021/8883—Scan or image signal processing specially adapted therefor, e.g. for scan signal adjustment, for detecting different kinds of defects, for compensating for structures, markings, edges involving the calculation of gauges, generating models
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
    • G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
    • G01N21/84—Systems specially adapted for particular applications
    • G01N21/88—Investigating the presence of flaws or contamination
    • G01N21/95—Investigating the presence of flaws or contamination characterised by the material or shape of the object to be examined
    • G01N2021/9513—Liquid crystal panels
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
    • G01N2201/00—Features of devices classified in G01N21/00
    • G01N2201/12—Circuits of general importance; Signal processing
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/045—Combinations of networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/20—Special algorithmic details
    • G06T2207/20081—Training; Learning
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/20—Special algorithmic details
    • G06T2207/20084—Artificial neural networks [ANN]
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/30—Subject of image; Context of image processing
    • G06T2207/30108—Industrial image inspection
    • G06T2207/30121—CRT, LCD or plasma display

Definitions

  • the invention relates to a defect detection technology, in particular to a method, a device, an electronic device and a storage medium for detecting a peripheral circuit of a display screen.
  • the detection of the peripheral circuits of the display screen mainly uses manual detection or machine-assisted manual detection methods.
  • the manual inspection method refers to relying on the naked eyes of industry experts to observe the pictures collected from the production environment to give a judgment;
  • the machine-assisted manual inspection method refers to the use of a quality inspection system that has solidified the experience of industry experts to first inspect the periphery of the display screen. The circuit image is inspected, and pictures that are suspected of defects are preliminarily screened. Then, industry experts perform manual detection and judgment on pictures that are suspected of defects.
  • the invention provides a method, a device, an electronic device and a storage medium for detecting a peripheral circuit of a display screen, so as to overcome the existing subjective influencing factors of the display circuit peripheral circuit defect detection method, resulting in low detection accuracy and system performance. Poor and low business expansion capabilities.
  • a method for detecting a peripheral circuit of a display screen including:
  • the quality inspection request including a display peripheral circuit image collected by an image acquisition device on the display peripheral circuit production line;
  • the defect detection model is an instance segmentation Mask RCNN algorithm using a historical defect display peripheral circuit image. Obtained by training
  • the quality of the display peripheral circuit corresponding to the display peripheral circuit image is determined according to the defect detection result.
  • the method before the inputting the peripheral circuit image of the display screen into a defect detection model to obtain a defect detection result, the method further includes:
  • the method before the enlarging or reducing the image of the peripheral circuit of the display screen, the method further includes:
  • image preprocessing on the display peripheral circuit image, wherein the image preprocessing includes one or more of the following processing:
  • the step of inputting the image to be tested into a defect detection model to obtain a defect detection result includes:
  • the image to be tested is input to the defect detection model running on the detection model server to obtain a defect detection result.
  • the defect detection result includes: a category of each defect, and / or a contour position of each defect;
  • the determining the quality of the display peripheral circuit corresponding to the display peripheral circuit image according to the defect detection result includes:
  • the quality of the display peripheral circuit corresponding to the display peripheral circuit image is determined.
  • the display peripheral circuit image and the defect detection result are input into the defect detection model in order to optimize the defect detection model.
  • a second aspect of the present application provides a display circuit peripheral circuit detection device, including:
  • a receiving module configured to receive a quality detection request sent by a console deployed on a display peripheral circuit production line, where the quality detection request includes a display peripheral circuit image collected by an image acquisition device on the display peripheral circuit production line;
  • a pre-processing module is used to enlarge or reduce the image of the peripheral circuit of the display screen to obtain an image to be tested whose size is consistent with the input size requirement of the defect detection model, wherein the defect detection model is a historical defect display peripheral circuit image Instance segmentation Mask RCNN algorithm training;
  • a processing module configured to input the image to be tested into a defect detection model to obtain a defect detection result
  • a determining module configured to determine, according to the defect detection result, the quality of the display peripheral circuit corresponding to the display peripheral circuit image.
  • the processing module is further configured to: prior to the input of the display peripheral circuit image into the defect detection model to obtain a defect detection result, use a historical defect
  • the actual pixel category of the display peripheral circuit image is used for the Mask RCNN algorithm training on the defect detection model, so that the defect detection model can predict the pixel category of the historical defect display peripheral circuit image output from the actual pixel classification.
  • the loss value between pixel categories is lower than a preset loss threshold.
  • the preprocessing module is further configured to perform an operation on the display peripheral circuit image before the display peripheral circuit image is enlarged or reduced.
  • Image preprocessing includes one or more of the following processes: trimming, cutting, and rotating.
  • the processing module is specifically configured to determine a detection model server that bears processing resources according to a load balancing policy; and input the image to be tested to the A defect detection result is obtained in the defect detection model on the detection model server.
  • the defect detection result includes: a category of each defect, and / or a contour position of each defect;
  • the determining module is specifically configured to determine the quality of the display peripheral circuit corresponding to the display peripheral circuit image according to the production stage information and the defect detection result.
  • the processing module is further configured to determine a quality of a display peripheral circuit corresponding to the display peripheral circuit image according to the defect detection result. After good or bad, if it is determined that the display peripheral circuit is a damaged circuit, perform one or more of the following operations:
  • the display peripheral circuit image and the defect detection result are input into the defect detection model in order to optimize the defect detection model.
  • a third aspect of the present application provides an electronic device including a processor, a memory, and a computer program stored on the memory and executable on the processor.
  • the processor executes the program, the first aspect as described above and The method according to any one of the various possible implementations of the first aspect.
  • a fourth aspect of the present application provides a storage medium, where the storage medium stores instructions, and when it runs on a computer, causes the computer to execute as described in the first aspect and any one of the various possible implementation manners of the first aspect Methods.
  • the display peripheral circuit detection method, device, electronic equipment and storage medium provided by the present invention receive a quality detection request sent by a console deployed on a display peripheral circuit production line, and the quality detection request includes the display peripheral A peripheral circuit image of a display screen collected by an image acquisition device on a circuit production line; the peripheral circuit image of the display screen is enlarged or reduced to obtain an image to be tested whose size is consistent with the input size requirement of the defect detection model, wherein the defect detection model It is obtained by performing an example segmentation algorithm MASK and RCNN on the historical defect display peripheral circuit image; inputting the image to be tested into a defect detection model to obtain a defect detection result; and determining the display peripheral circuit image according to the defect detection result
  • the quality of the corresponding display peripheral circuits is good or bad.
  • the defect detection model is obtained by performing MASK and RCNN training on the peripheral circuit image of the historical defect display, the defect detection results obtained by using the defect detection model have high classification accuracy, strong intelligence, system performance, and business.
  • the scalability is high, which solves the problems of low detection accuracy, poor system performance, and low business expansion ability due to the large subjective influence factors in the existing method for detecting defects in peripheral circuits of display screens.
  • FIG. 1 is a schematic structural diagram of a display circuit peripheral circuit detection system according to an embodiment of the present invention.
  • FIG. 2 is a schematic flowchart of a first embodiment of a method for detecting a peripheral circuit of a display screen according to an embodiment of the present application
  • FIG. 3 is a schematic flowchart of a second embodiment of a method for detecting a peripheral circuit of a display screen according to an embodiment of the present application
  • FIG. 4 is a schematic structural diagram of an embodiment of a display circuit peripheral circuit detecting device according to an embodiment of the present application.
  • FIG. 5 is a schematic structural diagram of an embodiment of an electronic device according to an embodiment of the present invention.
  • the size of the sequence number of each process does not mean the order of execution.
  • the execution order of each process should be determined by its function and internal logic.
  • the implementation process constitutes any limitation.
  • B corresponding to A means that B is associated with A, and B can be determined according to A. Determining B according to A does not mean determining B only based on A, but also determining B based on A and / or other information. The matching between A and B is that the similarity between A and B is greater than or equal to a preset threshold.
  • the overall intelligent automation level of the 3C industry (3C industry refers to the information appliance industry that integrates the application of the three major technology products of computers, communications, and consumer electronics) is low. Analysis shows that most manufacturers use two types of detection methods for mobile phone screens: manual detection methods and machine-assisted manual detection methods.
  • the manual detection method refers to relying on the naked eyes of industry experts to observe the images collected from the production environment for judgment. This method is subject to human subjective influence factors, has low detection efficiency, and has a large damage to the human eye. Because the generating circuit of the display peripheral circuit is generally a dust-free environment, workers need to prepare for cleaning before entering, and wear dust-free clothes, which may also adversely affect the health and safety of workers.
  • the machine-assisted manual detection method can also be referred to as the detection method based on the liquid crystal module detection equipment.
  • the specific principle is: first, the non-defective image is filtered by a quality inspection system with certain judgment ability, and then the industry experts will The image is detected.
  • quality inspection systems are mostly developed for expert systems and feature engineering systems, which means that experts have solidified their experience in the quality inspection system to make them have certain automation capabilities. Therefore, the machine-assisted manual detection method not only has low accuracy and poor system performance, and cannot cover all the detection standards of the manufacturer. Moreover, this method is also inefficient, and it is easy to miss and misjudge, and it is difficult to reuse the image data after detection. Dig.
  • the characteristics and determination rules are solidified into the machine based on the experience of industry experts, and it is difficult to iterate with the development of the business. As a result of the development of the production process, the detection accuracy of the quality inspection system is getting more and more Low, and may even drop to a completely unusable state.
  • the characteristics of the quality inspection system are pre-cured in the hardware by third-party suppliers. When upgrading, not only the production line needs to be significantly modified, but it is also expensive, and it has obvious shortcomings in terms of security, standardization, and scalability. , It is not conducive to the optimization and upgrade of the display peripheral circuit production line, and the business expansion capability is low.
  • both the manual detection method and the machine-assisted manual detection method have the following disadvantages: Not only are they inefficient and prone to misjudgment, but the industrial data generated by these two methods are not easy to store, manage, and reuse for secondary mining.
  • the embodiment of the present application develops an automatic, high-precision, adaptively modified and upgraded display peripheral circuit detection method, which uses image acquisition equipment to collect real-time data on the display peripheral circuit production line. Display peripheral circuit image, real-time detection and judgment of the surface quality of the display peripheral circuit. If a defect is detected in the display peripheral circuit collected by the current image acquisition device, the position of each defect in the picture and the Category, the embodiment of the present application distinguishes defective individuals from similar types of defects.
  • the defects described in the embodiments of the present application may include, but are not limited to, different types of defect problems including point defects, foreign object defects, and scratch defects. Not one by one here.
  • the example segmentation Mask RCNN algorithm is a two-stage framework.
  • the first stage scans the image and generates proposals (proposals, that is, regions that may contain a target).
  • the second stage classifies proposals and generates them. Bounding box and mask.
  • Mask R-CNN is an extension of Faster R-CNN and was proposed by the same author last year.
  • Faster RCNN is a popular object detection framework, and Mask RCNN extends it into an instance segmentation framework.
  • Mask RCNN is a new convolutional network based on Faster RCNN architecture. It completes instance segmentation in one fell swoop. This method completes high-quality instance segmentation while effectively targeting.
  • the Mask RCNN algorithm is mainly to extend the original Faster-RCNN, add a branch to use the existing detection to perform parallel prediction on the target.
  • this network structure is relatively easy to implement and train, and can be easily applied to other fields, such as object detection, segmentation, and keypoint detection of people.
  • FIG. 1 is a schematic structural diagram of a display circuit peripheral circuit detection system according to an embodiment of the present invention.
  • the method for detecting a peripheral circuit of a display screen provided by the present invention is used to perform defect detection on the peripheral circuit of a display screen.
  • the display peripheral circuit detection system mainly includes a console 12, a server group 13, a controller 14, a database 15, a trainer 16, and an image acquisition device 11 deployed on a display peripheral circuit production line.
  • the image acquisition device 11 collects the display peripheral circuit images on the display peripheral circuit production line
  • the console 12 receives the display peripheral circuit images collected by the image acquisition device 11 and sends the display peripheral circuit images to the server group 13
  • the detection model server 130 inputs the received display peripheral circuit image into the defect detection model running itself to obtain the defect detection result
  • the controller 14 receives the defect detection result of the detection model server 130 and combines it with the production stage
  • the information gives a business response
  • the controller 14 may also store the defect detection result in the database 15 as a log.
  • the display peripheral circuit images collected by the image acquisition device 11 can also be directly stored in the database 15 as raw data for training the defect detection model.
  • the trainer 16 extracts the historical defect display peripheral circuit images in the database and trains the defect detection model based on the Mask RCNN algorithm.
  • the above database 15 may include a production database 151 and a training database 152.
  • the production database 151 may receive and save the defect detection results sent by the controller 14 and the display screen peripheral circuit images collected by the image acquisition device 11.
  • the training database 152 may The historical defect display peripheral circuit image and the corresponding original display peripheral circuit image extracted from the production database 151 are stored, so that the trainer 16 trains to obtain a defect detection model with high detection accuracy.
  • the trainer 16 in the embodiment of the present application may be a training engine implemented by hardware and / or software functions, as a training tool for a defect detection model.
  • the display peripheral circuit detection system of the embodiment of the present application may further include other physical modules such as a processor, a memory, and the embodiment is not limited thereto.
  • FIG. 2 is a schematic flowchart of a first embodiment of a method for detecting a peripheral circuit of a display screen according to an embodiment of the present application.
  • the method shown in FIG. 2 may be implemented by a software device, a hardware device, or a combination of software and hardware. installation. Including steps S101 to S104, the details are as follows:
  • S101 Receive a quality detection request sent by a console deployed on a display peripheral circuit production line, where the quality detection request includes a display peripheral circuit image collected by an image acquisition device on the display peripheral circuit production line.
  • a plurality of different devices such as an image acquisition device, a console, a server group, a controller, and a database are deployed on the display peripheral circuit production line.
  • the image acquisition device can be a high-precision image acquisition camera.
  • multiple images can be collected during the production process. Display peripheral circuit image corresponding to the display peripheral circuit.
  • the console deployed on the display peripheral circuit production line can send quality to the server group with the defect detection model deployed on the display peripheral circuit production line.
  • the quality detection request includes a display peripheral circuit image collected by the image acquisition device, so that a server in the server group that receives the quality detection request processes the received display peripheral circuit image.
  • S102 Enlarge or reduce the image of the peripheral circuit of the display screen to obtain an image to be tested whose size is consistent with the input size requirement of the defect detection model.
  • the defect detection model is obtained by training an instance segmentation Mask RCNN algorithm using a historical defect display peripheral circuit image.
  • the defect detection model trained by instance segmentation Mask RCNN has a size requirement for the input image. Once the size of the input image does not match the size of the model input requirement, the defect detection model will not be able to process it.
  • the peripheral circuit graphics of the display screen are first performed before entering the defect detection model. Scaling processing so that the size of the image to be tested is consistent with the input size requirement of the defect detection model.
  • Enlarging or reducing the image of the peripheral circuit of the display screen can be understood as the constant enlargement or reduction of pixels, or it can be understood as the reduction or enlargement of pixel reduction. Pixels that are too high may exceed the processing capability of the defect detection model. Therefore, when the pixels of the display peripheral circuit image are too high, you can also perform pixel reduction processing on the display peripheral circuit image first, which is not limited here.
  • the server receiving the quality inspection request obtains the display peripheral circuit image in the quality inspection request, and performs pre-processing of enlargement or reduction to obtain an image to be tested whose size is consistent with the input size requirement of the defect detection model. Then the image to be tested is input into a defect detection model running on the server, and the defect detection model performs defect detection, and then the defect detection result is obtained.
  • image preprocessing may be performed on the image of the display peripheral circuit, wherein the image preprocessing includes the following processing: One or more of: trimming, cutting, rotating.
  • the image acquisition device deployed on the display peripheral circuit production line is generally a high-precision camera. Therefore, the image of the display peripheral circuit captured by the image acquisition device may be large in size, high in pixels, or in different positions. Suitable etc. Therefore, after receiving the display peripheral circuit image included in the quality detection request sent by the console, it is necessary to preprocess the display peripheral circuit image according to the actual situation. For example, if the peripheral area of the peripheral circuit image of the display screen is large, at this time, the peripheral circuit image of the display screen may be trimmed to retain a useful part of the peripheral circuit image of the display screen.
  • the defect detection model running on the server is obtained by training the Segment RCNN algorithm on the peripheral circuit image of the historical defect display.
  • this embodiment uses the Mask RCNN algorithm to perform instance segmentation.
  • Instance segmentation refers to letting the computer perform segmentation based on the individual instances of the image, that is, to distinguish each defect and identify the type of each defect.
  • the defect detection model adopts a MASK RCNN structure.
  • the display peripheral circuit image on the display peripheral circuit production line is used as the input of the defect detection model.
  • the MASK RCNN structure of the defect detection model is used to identify the characteristics of each pixel in the display peripheral circuit image, and the display peripheral circuit is obtained. Which pixels in the image are normal pixels, which pixels are defective pixels, and which type of defect are the defective pixels.
  • a model training process may be further included before the image of the display peripheral circuit is input into a defect detection model to obtain a defect detection result.
  • the MASK RCNN algorithm training may be performed on the defect detection model with the actual pixel category of the peripheral circuit image of the historical defect display screen, so that the defect detection model outputs The loss value between the predicted pixel category and the actual pixel category is lower than a preset loss threshold.
  • the loss value can be understood as the total loss value
  • the defect detection model is a combined training of the candidate area loss value, area category loss value, area boundary loss value, and pixel instance loss value of the historical defect display screen image, so that A result that the total loss value of the candidate region loss value, the region category loss value, the region boundary loss value, and the pixel instance loss value meets a preset loss threshold.
  • the candidate area loss value refers to a loss value between a selected defect area and an actual defect area in the historical defect display screen image
  • the area category loss value refers to a predicted defect category and an actual defect area in the selected defect area.
  • Loss value between defect categories the region boundary loss value refers to the loss value between the predicted defect boundary and the actual defect boundary in the selected defect area
  • the pixel instance loss value refers to the historical defect display screen image The loss value between the predicted pixel instance and the actual pixel instance.
  • the embodiment of the present application can use the MASK RCNN model to have high robustness to the deformation, blurring, and lighting changes of the display peripheral circuit image collected by the image acquisition device on the display peripheral circuit production line, and it is useful for classification tasks. Higher generalizability.
  • the organization of the MASK RCNN model required for training the defect detection model may be different, which can be performed according to the actual situation It is determined that this embodiment does not limit it.
  • the quality of the display peripheral circuit corresponding to the display peripheral circuit image may be determined according to the defect detection result.
  • the defect detection result may include: a category of each defect, and / or a contour position of each defect.
  • the defect detection result that can be obtained by the defect detection model can include the defect category (there are several types of defects on the display peripheral circuit), the defect location (the specificity of each defect) Pixel position), the contour of the defect (the contour shape of each defect).
  • the presentation of the defect detection results can be understood as: the defect detection model outputs a segmentation map, which is identified by the first color as normal pixels, the second color identifies the first defect, the second color identifies the second defect, and the first defect And the second defect can be the same type of defect, or it can be a different type of defect.
  • the defect detection result of detecting two types of defects for example, it can be a segmentation map with white as the background color and containing blue and green patches, where white represents pixels in a normal area, and blue represents a point-like defect. Pixels of the area, green represents pixels of another point-like defect.
  • the MASK RCNN model is pixel recognition, so various types of defect patterns can be obtained from the defect detection results, which can be understood as the contour shapes of various types of defects and their pixel locations in the display peripheral circuit image.
  • S104 (determining the quality of the display peripheral circuit corresponding to the display peripheral circuit image according to the defect detection result) may be replaced by: determining the display screen according to the production stage information and the defect detection result.
  • the quality of the peripheral circuit of the display screen corresponding to the peripheral circuit image is good or bad.
  • a variety of different production stage information such as the manufacturer, production environment, and type of display peripheral circuits may obtain different defect detection results during the display peripheral circuit detection process.
  • the production stages they undergo are different. Therefore, when analyzing the defect detection results obtained above, it is necessary to combine the production stage information of each display peripheral circuit to determine the quality of the display peripheral circuits. Good or bad.
  • the defect detection model in the embodiment of the present application can detect several types of defect types in the display peripheral circuit image and the specific number of each type of defect, that is, the defect detection model obtained by using the MASK RCNN algorithm can Distinguish between different defective individuals belonging to the same category.
  • the method for detecting a peripheral circuit of a display screen receives a quality detection request sent by a console deployed on a display peripheral circuit production line, and the quality detection request includes data collected by an image acquisition device on the display peripheral circuit production line.
  • the display peripheral circuit image is input to the defect detection model to obtain a defect detection result, and the quality of the display peripheral circuit corresponding to the display peripheral circuit image is determined according to the defect detection result. Since the above defect detection model is obtained by training the MASK and RCNN algorithm on the peripheral circuit image of the historical defect display, the defect detection results obtained by using the defect detection model have high classification accuracy, strong intelligence, and improved system performance. High business scalability.
  • FIG. 3 it is a schematic flowchart of a second embodiment of a method for detecting a peripheral circuit of a display screen according to an embodiment of the present application.
  • the above S104 inputting the image to be tested into a defect detection model to obtain a defect detection result
  • steps S301-S302 can be implemented through steps S301-S302, as follows:
  • S301 Determine a detection model server that carries processing resources according to a load balancing policy.
  • a server group is deployed on the display peripheral circuit production line.
  • the number of servers in the server group may be multiple, and each server runs a defect detection model.
  • the defect detection model running on each server is the same. Therefore, each server can receive the quality inspection request sent by the console, and then can use the defect detection model carried by itself to process the image of the display peripheral circuit. Quality Inspection.
  • the console can also send a quality detection request to any server in the server group in real time.
  • the load balancing can be performed according to a preset load Strategy, determine a server from the server group as a detection model server that carries processing resources, that is, load balancing and scheduling in real time according to the deployment of the defect detection model on the display peripheral circuit production line.
  • the above-mentioned peripheral circuit image of the display screen may be input into the defect detection model running on the detection model server,
  • the defect detection model is used to detect defects in the peripheral circuit image of the display screen, and then the defect detection results are obtained.
  • the defect detection model is obtained by a training module training a preset pixel category and an actual pixel category in a peripheral circuit image of a historical defect display screen.
  • the method for detecting a peripheral circuit of a display screen determines a detection model server carrying processing resources according to a load balancing policy, and inputs the image to be tested into a defect detection model running on the detection model server to obtain a defect.
  • the detection result can achieve load balancing on the server, improve the detection efficiency of the display peripheral circuit image, and improve the performance of the display peripheral circuit detection system.
  • the method may further include the following steps:
  • the display peripheral circuit image and the defect detection result are input into the defect detection model in order to optimize the defect detection model.
  • the tester may preset a solution when the display peripheral circuit is determined to be a bad screen according to the production scenario and production stage information of the display peripheral circuit, for example, the controller sends the The manager sends an alarm message, and / or stores the above-mentioned defect detection result as a log in the production database through the controller, and / or, sends a production control instruction to the console through the controller to eliminate the defect, and / or,
  • the display peripheral circuit image and the above-mentioned defect detection result are input into the above-mentioned defect detection model in order to optimize the above-mentioned defect detection model and the like.
  • an alarm message may be issued to enable production management.
  • the developer locates the category and location of the defect in time, and gives a solution.
  • the above-mentioned defect detection result may be stored in the production database as a log by the controller, that is, the category of each defect of the peripheral circuit of the display screen , And / or the contour position of each defect is stored as a log in the production database, which can be filtered into the training database, and the training module (which can be a software program such as a training engine) updates the above according to the defective display peripheral circuit image Defect detection model.
  • a production control instruction may also be sent to the console through the controller to eliminate the defect. That is, the inspection model server that carries the defect inspection model can determine the cause of the defect through the controller, and then adjust the production process accordingly, that is, the inspection model server sends the production control instruction to the console through the controller to eliminate the periphery of the display screen. Defects in the circuit to reduce the probability of damage to the circuit.
  • the image of the peripheral circuit of the display screen and the defect detection result may be directly input into the defect detection model in order to optimize the defect detection model. That is, the image of the display peripheral circuit corresponding to the damaged circuit is directly used as the training set of the defect detection model to optimize the defect detection model, thereby improving the detection accuracy of the defect detection model.
  • the embodiments of the present application are not limited to the above-mentioned one or more operations that can be performed by the detection model server when it is determined that the peripheral circuit of the display screen is a damaged circuit, which can be determined according to the actual situation, and will not be repeated here.
  • the operation steps corresponding to the display peripheral circuit detection method may also be distributed to the above.
  • Multiple different devices to perform For example, the image acquisition device collects the display peripheral circuit image, and the console sends the display peripheral circuit image collected by the image acquisition device to the detection model server in the server group according to the load balancing strategy, and the defect detection running on the detection model server is performed.
  • the model performs preset preprocessing on the peripheral circuit images of the display screen, and then performs defect detection, and gives the defect detection results.
  • the detection model server can send the defect detection results to the controller.
  • the controller combines the actual business scenario and responds to the requirements of the above-mentioned defect detection results according to business requirements, such as alarms, storage logs, and control. Production control instructions, etc.
  • the controller can also store the defect detection results and the above-mentioned response processing behavior as logs in the production database, so that the training module updates according to the display peripheral circuit image and defect detection results in the training database.
  • the training database stores data such as the peripheral circuit image of the display screen with defects and corresponding defect detection results, which are selected from the production database.
  • the defect detection model running on the server can be gradually replaced by a small-traffic online method, so as to achieve the purpose of dynamically expanding and generalizing the defect detection model with business scene and production stage information.
  • the display peripheral circuit detection method in the embodiment of the present application runs for a period of time on the display peripheral circuit production line, the accuracy of the above defect detection and defect location can be reviewed manually through the information in the production database, and then the above training database is updated. Retrain the defect detection model to improve the accuracy of defect detection.
  • the display circuit peripheral circuit detecting device provided in the embodiment of the present application may mainly include a receiving module 41, a pre-processing module 42, a processing module 43, and a determining module 44.
  • the receiving module 41 is configured to receive a quality detection request sent by a console deployed on a display peripheral circuit production line, and the quality detection request includes a display peripheral collected by an image acquisition device on the display peripheral circuit production line. Circuit image.
  • the pre-processing module 42 is configured to enlarge or reduce the image of the peripheral circuit of the display screen to obtain an image to be tested whose size is consistent with the input size requirement of the defect detection model.
  • the defect detection model is a historical defect display peripheral circuit.
  • the image is obtained by training the MASK RCNN algorithm on instance segmentation.
  • the processing module 43 is configured to input the image to be tested into a defect detection model to obtain a defect detection result.
  • a determining module 44 is configured to determine, according to the defect detection result, the quality of a display screen peripheral circuit corresponding to the display screen peripheral circuit image.
  • the device for detecting peripheral circuits of the display screen in the embodiment shown in FIG. 4 may be correspondingly used to execute the steps in the method embodiment shown in FIG. 2.
  • the implementation principles and technical effects are similar, and will not be repeated here.
  • the processing module 43 is further configured to, before the defect detection result is obtained by inputting the image of the display peripheral circuit into the defect detection model, compare the actual pixel category of the image of the peripheral circuit of the historical defect display with the actual pixel type.
  • the defect detection model is trained by the MASK RCNN algorithm, so that the loss value between the predicted pixel type of the defect detection model for the historical defect display peripheral circuit image output and the actual pixel type is lower than a preset Loss threshold.
  • the preprocessing module 42 is further configured to perform image preprocessing on the display peripheral circuit image before zooming in or out of the display peripheral circuit image, where the image preprocessing includes One or more of the following: trimming, cutting, rotating.
  • the processing module 43 is specifically configured to determine a detection model server carrying processing resources according to a load balancing policy; and input the image to be tested into the defect detection model running on the detection model server Get defect detection results.
  • the defect detection result includes: a category of each defect, and / or a contour position of each defect.
  • the determining module 44 is specifically configured to determine the quality of the display peripheral circuit corresponding to the display peripheral circuit image according to the production stage information and the defect detection result.
  • the processing module 43 is further configured to, after determining the quality of the display peripheral circuit corresponding to the display peripheral circuit image according to the defect detection result, determine the display peripheral circuit To damage the circuit, do one or more of the following:
  • the display peripheral circuit image and the defect detection result are input into the defect detection model in order to optimize the defect detection model.
  • the device for detecting peripheral circuits of a display screen of the foregoing device embodiment may be used to execute the implementation solutions of the method embodiments shown in FIG. 2 to FIG.
  • the electronic device includes a processor 51, a memory 52, and a computer program.
  • the memory 52 is configured to store the computer program, and the memory may also be a flash memory.
  • the computer program is, for example, an application program, a functional module, and the like that implement the above method.
  • the processor 51 is configured to execute a computer program stored in the memory to implement each step performed by the electronic device in the foregoing method. For details, refer to related descriptions in the foregoing method embodiments.
  • the memory 52 may be independent or integrated with the processor 51.
  • the electronic device may further include:
  • the bus 53 is configured to connect the memory 52 and the processor 51.
  • the present application also provides a storage medium.
  • the storage medium has instructions stored therein, which when run on a computer, cause the computer to execute the method in the method embodiments shown in FIG. 2 to FIG. 3.
  • the storage medium may be a computer storage medium or a communication medium.
  • Communication media include any medium that facilitates transfer of a computer program from one place to another.
  • Computer storage media can be any available media that can be accessed by a general purpose or special purpose computer.
  • a storage medium is coupled to the processor such that the processor can read information from, and write information to, the readable storage medium.
  • the storage medium may also be an integral part of the processor.
  • the processor and the storage medium may be located in application specific integrated circuits (Application Specific Integrated Circuits, ASIC for short).
  • the ASIC may reside in a user equipment.
  • the processor and the storage medium may also exist as discrete components in a communication device.
  • the present application also provides a program product, which includes a computer program stored in a storage medium.
  • At least one processor of the display peripheral circuit detection device may read the computer program from a storage medium, and the at least one processor executes the computer program to cause the display peripheral circuit detection device to execute the method in the method embodiments shown in FIGS.
  • the processor may be a central processing unit (English: Central Processing Unit, CPU for short), or other general-purpose processors, digital signal processors (English: Digital Signal Processor, Abbreviation: DSP), Application Specific Integrated Circuit (English: Application Specific Integrated Circuit, Abbreviation: ASIC), etc.
  • a general-purpose processor may be a microprocessor or the processor may be any conventional processor or the like. The steps combined with the method disclosed in the present application can be directly embodied as being executed by a hardware processor, or executed and completed by a combination of hardware and software modules in the processor.

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Chemical & Material Sciences (AREA)
  • Health & Medical Sciences (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • General Health & Medical Sciences (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Immunology (AREA)
  • Biochemistry (AREA)
  • Pathology (AREA)
  • Analytical Chemistry (AREA)
  • Quality & Reliability (AREA)
  • General Engineering & Computer Science (AREA)
  • Nonlinear Science (AREA)
  • Optics & Photonics (AREA)
  • Signal Processing (AREA)
  • Crystallography & Structural Chemistry (AREA)
  • Biomedical Technology (AREA)
  • Molecular Biology (AREA)
  • Computer Hardware Design (AREA)
  • Biophysics (AREA)
  • Computational Linguistics (AREA)
  • Data Mining & Analysis (AREA)
  • Evolutionary Computation (AREA)
  • Artificial Intelligence (AREA)
  • Computing Systems (AREA)
  • Mathematical Physics (AREA)
  • Software Systems (AREA)
  • Human Computer Interaction (AREA)
  • Image Processing (AREA)
  • Image Analysis (AREA)
  • Controls And Circuits For Display Device (AREA)
  • Investigating Materials By The Use Of Optical Means Adapted For Particular Applications (AREA)

Abstract

本发明提供一种显示屏外围电路检测方法、装置、电子设备及存储介质，通过接收部署在显示屏外围电路生产线上的控制台发送的质量检测请求，所述质量检测请求中包含所述显示屏外围电路生产线上的图像采集设备采集的显示屏外围电路图像；将所述显示屏外围电路图像放大或缩小，得到大小与缺陷检测模型的输入大小要求一致的待测图像，其中，所述缺陷检测模型是对历史缺陷显示屏外围电路图像进行实例分割算法MASK RCNN训练得到的；将所述待测图像输入到缺陷检测模型中得到缺陷检测结果；根据所述缺陷检测结果确定所述显示屏外围电路图像对应的显示屏外围电路的质量好坏。该技术方案的缺陷检测准确度高、系统性能好，业务扩展能力高。

Description

显示屏外围电路检测方法、装置、电子设备及存储介质
本申请要求于2018年07月02日提交中国专利局、申请号为201810709836.7、申请人为北京百度网讯科技有限公司、发明名称为“显示屏外围电路检测方法、装置、电子设备及存储介质”的中国专利申请的优先权，其全部内容通过引用结合在本申请中。
技术领域
本发明涉及缺陷检测技术，尤其涉及一种显示屏外围电路检测方法、装置、电子设备及存储介质。
背景技术
随着科技的发展，信息显示技术在人们生活中的作用与日俱增，显示屏也因其体积小、重量轻、功率低、分辨率高、亮度高和无几何变形诸多特点被广泛应用。但在显示屏的生产过程中，由于工艺及环境的原因可能导致显示屏外围电路存在缺陷，例如，点类缺陷、异物类缺陷和划痕类缺陷等。因而，显示屏外围电路检测是生产过程中的重要环节。
现有技术中，显示屏外围电路检测主要采用人工检测或机器辅助的人工检测方法。具体的，人工检测方法是指依赖行业专家肉眼观察从生产环境中采集到的图片给出判断；机器辅助的人工检测方法是指首先利用固化有行业专家经验的质检系统对待检测的显示屏外围电路图像进行检测，初步筛选出疑似存在缺陷的图片，再由行业专家对疑似存在缺陷的图片进行人工检测判断。
然而，不管是人工检测方法，还是机器辅助的人工检测方法均受人的主观影响因素较大，检测准确度低、系统性能差，业务扩展能力低。
发明内容
本发明提供一种显示屏外围电路检测方法、装置、电子设备及存储介质，以克服现有显示屏外围电路缺陷检测方法中由于受人的主观影响因素较大，致使检测准确度低、系统性能差、业务扩展能力低的问题。
根据本发明的第一方面，提供一种显示屏外围电路检测方法，包括：
接收部署在显示屏外围电路生产线上的控制台发送的质量检测请求，所述质量检测请求中包含所述显示屏外围电路生产线上的图像采集设备采集的显示屏外围电路图像；
将所述显示屏外围电路图像放大或缩小，得到大小与缺陷检测模型的输入大小要求一致的待测图像，其中，所述缺陷检测模型是用历史缺陷显示屏外围电路图像进行实例分割Mask RCNN算法训练得到的；
将所述待测图像输入到缺陷检测模型中得到缺陷检测结果；
根据所述缺陷检测结果确定所述显示屏外围电路图像对应的显示屏外围电路的质量好坏。
可选地，在第一方面的一种可能实现方式中，在所述将所述显示屏外围电路图像输入到缺陷检测模型中得到缺陷检测结果之前，还包括：
以历史缺陷显示屏外围电路图像的实际像素类别对所述缺陷检测模型进行所述Mask RCNN算法训练，以使所述缺陷检测模型对所述历史缺陷显示屏外围电路图像输出的预测像素类别，与所述实际像素类别之间的损失值低于预设损失阈值。
可选地，在第一方面的另一种可能实现方式中，在所述将所述显示屏外围电路图像放大或缩小之前，还包括：
对所述显示屏外围电路图像进行图像预处理，其中，所述图像预处理包括下述处理中的一项或多项：
裁边、剪切、旋转。
可选地，在第一方面的再一种可能实现方式中，所述将所述待测图像输入到缺陷检测模型中得到缺陷检测结果，包括：
根据负载均衡策略，确定承载处理资源的检测模型服务器；
将所述待测图像输入到运行在所述检测模型服务器上的所述缺陷检测模型中得到缺陷检测结果。
可选地，在第一方面的又一种可能实现方式中，所述缺陷检测结果，包括：每个缺陷的类别，和/或各缺陷的轮廓位置；
所述根据所述缺陷检测结果确定所述显示屏外围电路图像对应的显示屏外围电路的质量好坏，包括：
根据生产阶段信息以及所述缺陷检测结果，确定所述显示屏外围电路图像对应的显示屏外围电路的质量好坏。
可选地，在第一方面的又一种可能实现方式中，在所述根据所述缺陷检测结果确定所述显示屏外围电路图像对应的显示屏外围电路的质量好坏之后，还包括：
若确定所述显示屏外围电路为损坏电路，则执行以下一项或多项操作：
通过控制器向生产管理者发送报警信息；
通过控制器将所述缺陷检测结果作为日志存储到生产数据库中；
通过控制器向所述控制台发送生产控制指令以便消除缺陷；
将所述显示屏外围电路图像和所述缺陷检测结果输入到所述缺陷检测模型中以便优化所述缺陷检测模型。
本申请第二方面提供一种显示屏外围电路检测装置，包括：
接收模块，用于接收部署在显示屏外围电路生产线上的控制台发送的质量检测请求，所述质量检测请求中包含所述显示屏外围电路生产线上的图像采集设备采集的显示屏外围电路图像；
预处理模块，用于将所述显示屏外围电路图像放大或缩小，得到大小与缺陷检测模型的输入大小要求一致的待测图像，其中，所述缺陷检测模型是用历史缺陷显示屏外围电路图像进行实例分割Mask RCNN算法训练得到的；
处理模块，用于将所述待测图像输入到缺陷检测模型中得到缺陷检测结果；
确定模块，用于根据所述缺陷检测结果确定所述显示屏外围电路图像对应的显示屏外围电路的质量好坏。
可选地，在第二方面的一种可能实现方式中，所述处理模块，还用于在所述将所述显示屏外围电路图像输入到缺陷检测模型中得到缺陷检测结果之前，以历史缺陷显示屏外围电路图像的实际像素类别对所述缺陷检测模型进行所述Mask RCNN算法训练，以使所述缺陷检测模型对所述历史缺陷显示屏外围电路图像输出的预测像素类别，与所述实际像素类别之间的损失值低于预设损失阈值。
可选地，在第二方面的另一种可能实现方式中，所述预处理模块，还用于在所述将所述显示屏外围电路图像放大或缩小之前对所述显示屏外围电路图像进行图像预处理，其中，所述图像预处理包括下述处理中的一项或多项：裁边、剪切、旋转。
可选地，在第二方面的再一种可能实现方式中，所述处理模块，具体用于根据负载均衡策略，确定承载处理资源的检测模型服务器；将所述待测图像输入到运行在所述检测模型服务器上的所述缺陷检测模型中得到缺陷检测结果。
可选地，在第二方面的又一种可能实现方式中，所述缺陷检测结果，包括：每个缺陷的类别，和/或各缺陷的轮廓位置；
所述确定模块，具体用于根据生产阶段信息以及所述缺陷检测结果，确定所述显示屏外围电路图像对应的显示屏外围电路的质量好坏。
可选地，在第二方面的又一种可能实现方式中，所述处理模块，还用于在所述根据所述缺陷检测结果确定所述显示屏外围电路图像对应的显示屏外围电路的质量好坏之后，若确定所述显示屏外围电路为损坏电路，则执行以下一项或多项操作：
通过控制器向生产管理者发送报警信息；
通过控制器将所述缺陷检测结果作为日志存储到生产数据库中；
通过控制器向所述控制台发送生产控制指令以便消除缺陷；
将所述显示屏外围电路图像和所述缺陷检测结果输入到所述缺陷检测模型中以便优化所述缺陷检测模型。
本申请第三方面提供一种电子设备，包括处理器、存储器及存储在所述存储器上并可在处理器上运行的计算机程序，所述处理器执行所述程序时实现如上述第一方面以及第一方面各种可能实现方式中任一项所述的方法。
本申请第四方面提供一种存储介质，所述存储介质中存储有指令，当其在计算机上运行时，使得计算机执行如第一方面以及第一方面各种可能实现方式中任一项所述的方法。
本发明提供的显示屏外围电路检测方法、装置、电子设备及存储介质，通过接收部署在显示屏外围电路生产线上的控制台发送的质量检测请求，所述质量检测请求中包含所述显示屏外围电路生产线上的图像采集设备采集的显示屏外围电路图像；将所述显示屏外围电路图像放大或缩小，得到大小与缺陷检测模型的输入大小要求一致的待测图像，其中，所述缺陷检测模型是对历史缺陷显示屏外围电路图像进行实例分割算法MASK RCNN训练得到的；将所述待测图像输入到缺陷检测模型中得到缺陷检测结果；根据所述缺陷检测结果确定所述显示屏外围电路图像对应的显示屏外围电路的质量好坏。由于上述缺陷检测模型是对历史缺陷显示屏外围电路图像进行MASK RCNN训练得到的，因而，利用该缺陷检测模型得到的缺陷检测结果的分类精度高，智能化能力强，系统性能有所提高，业务可扩展能力高，解决了现有显示屏外围电路缺陷检测方法中由于受人的主观影响因素较大，致使检测准确度低、系统性能差、业务扩展能力低的问题。
附图说明
图1为本发明实施例提供的一种显示屏外围电路检测系统的结构示意图；
图2为本申请实施例提供的显示屏外围电路检测方法实施例一的流程示意图；
图3为本申请实施例提供的显示屏外围电路检测方法实施例二的流程示意图；
图4为本申请实施例提供的显示屏外围电路检测装置实施例的结构示意图；
图5为本发明实施例提供的一种电子设备实施例的结构示意图。
具体实施方式
为使本发明实施例的目的、技术方案和优点更加清楚，下面将结合本发明实施例中的附图，对本发明实施例中的技术方案进行清楚、完整地描述，显然，所描述的实施例仅仅是本发明一部分实施例，而不是全部的实施例。基于本发明中的实施例，本领域普通技术人员在没有做出创造性劳动前提下所获得的所有其他实施例，都属于本发明保护的范围。
应当理解，在本申请的各种实施例中，各过程的序号的大小并不意味着执行顺序的先后，各过程的执行顺序应以其功能和内在逻辑确定，而不应对本申请实施例的实施过程构成任何限定。
应当理解，在本申请中，“包括”和“具有”以及他们的任何变形，意图在于覆盖不排他的包含，例如，包含了一系列步骤或单元的过程、方法、系统、产品或设备不必限于清楚地列出的那些步骤或单元，而是可包括没有清楚地列出的或对于这些过程、方法、产品或设备固有的其它步骤或单元。
应当理解，在本申请中，“多个”是指两个或两个以上。“和/或”仅仅是一种描述关联对象的关联关系，表示可以存在三种关系，例如，和/或B，可以表示：单独存在A，同时存在A和B，单独存在B这三种情况。字符“/”一般表示前后关联对象是一种“或”的关系。
应当理解，在本申请中，“与A相对应的B”、“A与B相对应”或者“B与A相对应”，表示B与A相关联，根据A可以确定B。根据A确定B并不意味着仅仅根据A确定B，还可以根据A和/或其他信息确定B。A与B的匹配，是A与B的相似度大于或等于预设的阈值。
取决于语境，如在此所使用的“若”可以被解释成为“在……时”或“当……时”或“响应于确定”或“响应于检测”。
现阶段，3C产业(3C产业是指结合电脑、通讯、和消费性电子三大科技产品整合应用的资讯家电产业)的整体智能自动化程度较低，通过对手机屏等显示屏外围电路行业的调研分析可知，大部分生产厂家对手机屏采用的检测方式可以分为两种，即：人工检测方法和机器辅助的人工检测方法。
其中，人工检测方法是指依赖于行业专家肉眼观察从生产环境中采集到的图像进行判断，该方法受人的主观影响因素较大、检测效率较低，且对人眼的伤害较大，此外，由于 显示屏外围电路的生成车间一般为无尘环境，工作人员进去前需要进行清洁准备，穿戴无尘衣服，其还可能对工作人员的健康和安全会产生不利影响。
机器辅助的人工检测方法也可以称为基于液晶模组检测设备检测方法，具体原理为：首先由具有一定判断能力的质检系统过滤掉不存在缺陷的图像，再由行业专家对疑似存在缺陷的图像进行检测判断。在机器辅助的人工检测方法中，质检系统多为专家系统和特征工程系统发展而来，是指专家将经验固化在质检系统中，使其具有一定的自动化能力。因此，机器辅助的人工检测方法不仅准确率低，系统性能差，无法覆盖厂商所有的检测标准，而且这种方法还效率低，容易漏判误判，检测后的图像数据很难进行二次利用挖掘。此外，在上述质检系统中，特征和判定规则都是基于行业专家的经验固化到机器中的，难以随业务的发展迭代，导致随着生产工艺的发展，质检系统的检测精度越来越低，甚至可能降低到完全不可用的状态。另外，质检系统的特征都由第三方供应商预先固化在硬件中，升级时不仅需要对生产线进行重大改造，而且价格昂贵，其在安全性、规范化、可扩展性等方面都存在着明显不足，不利于显示屏外围电路生产线的优化升级，业务扩展能力低。
综上所述，人工检测方法和机器辅助的人工检测方法均存在如下缺点：不仅效率低下、容易出现误判，而且这两种方法产生的工业数据不易存储、管理和二次挖掘再利用。
本申请实施例基于人工智能技术在计算机视觉中的最新发展，研发一种自动化、高精度、自适应修正升级的显示屏外围电路检测方法，利用图像采集设备在显示屏外围电路生产线上实时采集的显示屏外围电路图像，实时对显示屏外围电路的表面质量进行检测判断，如果检测到当前图像采集设备采集到的显示屏外围电路存在缺陷，则确定出每个缺陷在图片中的位置及缺陷的类别，本申请实施例对同类型缺陷区分缺陷个体。
可选地，本申请实施例中所述的缺陷可以包括，但是不局限于包括点类缺陷、异物类缺陷和划痕类缺陷等不同类别的缺陷问题。此处不进行一一介绍。
应当理解，在本申请中，实例分割Mask RCNN算法，是一个两阶段的框架，第一个阶段扫描图像并生成提议(proposals，即有可能包含一个目标的区域)，第二阶段分类提议并生成边界框和掩码。Mask R-CNN扩展自Faster R-CNN，由同一作者在去年提出。Faster RCNN是一个流行的目标检测框架，Mask RCNN将其扩展为实例分割框架。Mask RCNN是基于faster RCNN架构提出的新的卷积网络，一举完成了实例分割，该方法在有效地目标的同时完成了高质量的实例分割。Mask RCNN算法主要是把原有的Faster-RCNN进行扩展，添加一个分支使用现有的检测对目标进行并行预测。同时，这个网络结构比较容易实现和训练，可以很方便的应用到其他的领域，例如目标检测，分割，和人物关键点检测等。
下面以具体地实施例对本发明的技术方案进行详细说明。下面这几个具体的实施例可以相互结合，对于相同或相似的概念或过程可能在某些实施例不再赘述。
下面首先针对本申请实施例所适用的应用场景进行简要说明。参见图1，为本发明实施例提供的一种显示屏外围电路检测系统的结构示意图。在图1所示的系统中应用了本发明提供的显示屏外围电路检测方法对显示屏外围电路进行缺陷检测。如图1所示，该显示屏外围电路检测系统主要包括：控制台12、服务器组13、控制器14、数据库15、训练器16和部署在显示屏外围电路生产线上的图像采集设备11。
其中，图像采集设备11采集显示屏外围电路生产线上的显示屏外围电路图像，控制台12接收图像采集设备11采集的显示屏外围电路图像，并将该显示屏外围电路图像发送给服务器组13中的检测模型服务器130，检测模型服务器130将接收到的显示屏外围电路图像输入到本身运行的缺陷检测模型中得到缺陷检测结果，控制器14接收检测模型服务器130的缺陷检测结果，并结合生产阶段信息给出业务响应，控制器14还可以将缺陷检测结果作为日志存储到数据库15中。此外，图像采集设备11采集到的显示屏外围电路图像还可以直接存储到数据库15中，作为缺陷检测模型训练的原始数据。训练器16提取数据库中的历史缺陷显示屏外围电路图像基于Mask RCNN算法训练得到缺陷检测模型。
可选地，上述数据库15可以包括生产数据库151和训练数据库152，生产数据库151可以接收并保存控制器14发送的缺陷检测结果以及图像采集设备11采集到的显示屏外围电路图像，训练数据库152可以存储从生产数据库151提取的历史缺陷显示屏外围电路图像和对应的原始显示屏外围电路图像，以使训练器16训练得到检测准确率高的缺陷检测模型。
可选地，本申请实施例中的训练器16可以是由硬件和/或软件功能实现的训练引擎，作为缺陷检测模型的训练工具。本申请实施例的显示屏外围电路检测系统中还可以包括处理器、存储器等其他实体模块，本实施例不限于此。
参见图2，为本申请实施例提供的显示屏外围电路检测方法实施例一的流程示意图，图2所示方法的执行主体可以是软件装置，也可以是硬件装置，或者是软件与硬件相结合的装置。包括步骤S101至步骤S104，具体如下：
S101，接收部署在显示屏外围电路生产线上的控制台发送的质量检测请求，所述质量检测请求中包含所述显示屏外围电路生产线上的图像采集设备采集的显示屏外围电路图像。
可选地，在本申请实施例中，显示屏外围电路生产线上部署有图像采集设备、控制台、 服务器组、控制器、数据库等多个不同的设备。图像采集设备可以是高精度图像采集摄像头，在显示屏外围电路的生产过程中，通过调整图像采集设备的角度、光线、滤镜、倍镜、聚焦等，可以采集到多张正处于生产过程中的显示屏外围电路对应的显示屏外围电路图像。
当显示屏外围电路生产线上的图像采集设备采集到显示屏外围电路图像之后，部署在显示屏外围电路生产线上的控制台则可以向显示屏外围电路生产线上部署有缺陷检测模型的服务器组发送质量检测请求，该质量检测请求中包含上述图像采集设备采集的显示屏外围电路图像，以使服务器组中接收到该质量检测请求的服务器对接收到的显示屏外围电路图像进行处理。
S102，将所述显示屏外围电路图像放大或缩小，得到大小与缺陷检测模型的输入大小要求一致的待测图像。
S103，将所述待测图像输入到缺陷检测模型中得到缺陷检测结果。
其中，所述缺陷检测模型是用历史缺陷显示屏外围电路图像进行实例分割Mask RCNN算法训练得到的。由实例分割Mask RCNN训练得到的缺陷检测模型对输入的图像有大小要求，一旦输入图像的大小与模型输入要求的大小不一致，缺陷检测模型将无法对其进行处理。对显示屏外围电路进行检测时，全局图像所示意的线路走向、绕线形状更能表现其可能存在的缺陷问题，因此本实施例中在输入缺陷检测模型之前，先对显示屏外围电路图形进行缩放处理，以使得待测图像的大小与缺陷检测模型的输入大小要求一致。
对显示屏外围电路图像放大或缩小，可以理解为像素不变的放大或缩小，也可以理解为像素降低的放大或缩小。像素过高可能超出缺陷检测模型的处理能力，因此在显示屏外围电路图像像素过高的情况下，也可以先对显示屏外围电路图像进行降像素处理，在此不做限定。
可选地，接收到质量检测请求的服务器将质量检测请求中的显示屏外围电路图像获取出来，并进行放大或缩小的预处理，得到大小与缺陷检测模型的输入大小要求一致的待测图像。然后将待测图像输入到服务器上运行着的缺陷检测模型中，由缺陷检测模型执行缺陷检测，进而得到缺陷检测结果。
在一种实现方式中，在所述将所述显示屏外围电路图像放大或缩小之前，还可以对所述显示屏外围电路图像进行图像预处理，其中，所述图像预处理包括下述处理中的一项或多项：裁边、剪切、旋转。可以理解为，部署在显示屏外围电路生产线上的图像采集设备一般是高精度摄像头，因而，利用该图像采集设备采集到的显示屏外围电路图像可能尺寸较大、或者像素较高、或者位置不合适等。因而，当接收到控制台发送的包含在质量检测 请求中的显示屏外围电路图像之后，需要根据实际情况对显示屏外围电路图像进行预处理。例如，若显示屏外围电路图像的边缘区域较大，此时，可以对显示屏外围电路图像进行裁边处理，保留显示屏外围电路图像的有用部分。
值得说明的是，服务器上运行的缺陷检测模型是对历史缺陷显示屏外围电路图像进行实例分割Mask RCNN算法训练得到的。具体的，本实施例是用Mask RCNN算法来进行实例分割。实例分割是指让计算机根据图像的实例个体来进行分割，即对每个缺陷进行区分识别，并识别出各缺陷的类型。在本申请实施例中，缺陷检测模型采用MASK RCNN结构。具体的，显示屏外围电路生产线上的显示屏外围电路图像作为缺陷检测模型的输入，利用缺陷检测模型的MASK RCNN结构识别显示屏外围电路图像中每个像素点的特征，即得到显示屏外围电路图像中哪些像素点是正常像素点，哪些像素点是缺陷像素点，且缺陷像素点具体是哪个类型的缺陷。
作为一种示例，在所述将所述显示屏外围电路图像输入到缺陷检测模型中得到缺陷检测结果之前，还可以包括模型训练过程。具体地，可以是以历史缺陷显示屏外围电路图像的实际像素类别对所述缺陷检测模型进行所述MASK RCNN算法训练，以使所述缺陷检测模型对所述历史缺陷显示屏外围电路图像输出的预测像素类别，与所述实际像素类别之间的损失值低于预设损失阈值。
损失值可以理解为是总损失值，所述缺陷检测模型是对所述历史缺陷显示屏图像的候选区域损失值、区域类别损失值、区域边界损失值和像素实例损失值进行组合训练，以使所述候选区域损失值、所述区域类别损失值、所述区域边界损失值和所述像素实例损失值的总损失值满足预设损失阈值的结果。其中，所述候选区域损失值指所述历史缺陷显示屏图像中选定缺陷区域与实际缺陷区域之间的损失值，所述区域类别损失值指所述选定缺陷区域中预测缺陷类别与实际缺陷类别之间的损失值，所述区域边界损失值指所述选定缺陷区域中预测缺陷边界与实际缺陷边界之间的损失值，所述像素实例损失值指所述历史缺陷显示屏图像中预测像素实例与实际像素实例之间的损失值。
本申请实施例可以利用MASK RCNN模型，对显示屏外围电路生产线上由图像采集设备采集到的显示屏外围电路图像的变形、模糊、光照变化等特征具有较高的鲁棒性，对于分类任务具有更高的可泛化性。
值得说明的是，在本申请实施例中，对于不同的生产场景和显示屏外围电路图像的特点，训练上述缺陷检测模型所需要的MASK RCNN模型的组织方式均可能不同，其可以根据实际情况进行确定，本实施例并不对其进行限定。
S104，根据所述缺陷检测结果确定所述显示屏外围电路图像对应的显示屏外围电路的质量好坏。
可选地，在本申请的实施例中，当根据缺陷检测模型得出缺陷检测结果之后，可以依据该缺陷检测结果确定上述显示屏外围电路图像对应的显示屏外围电路的质量好坏。
可选地，在本申请的一实施例中，上述缺陷检测结果，可以包括：每个缺陷的类别，和/或各缺陷的轮廓位置。例如，当显示屏外围电路图像中存在缺陷时，该缺陷检测模型可以得出的缺陷检测结果中可以包括缺陷类别(显示屏外围电路上共存在几类缺陷)、缺陷位置(每个缺陷的具体像素位置)、缺陷的轮廓(每个缺陷的轮廓形状)。缺陷检测结果的呈现方式可以理解为，缺陷检测模型输出分割图，分割图以第一颜色标识为正常像素、第二颜色标识第一个缺陷、第二颜色标识第二个缺陷，第一个缺陷和第二个缺陷可以是同类型缺陷，也可以是不同类型缺陷。在检测到两类缺陷的缺陷检测结果中，例如可以是一张白色为底色，且包含蓝色色块和绿色色块的分割图，其中白色代表正常区域的像素，蓝色代表一个点类缺陷区域的像素，绿色代表另一个点类缺陷的像素。MASK RCNN模型是像素点识别，因此从缺陷检测结果可以得到各类缺陷图形，可以理解为是各类缺陷的轮廓形状，以及其在显示屏外围电路图像中的像素点位置。
相应的，S104(根据所述缺陷检测结果确定所述显示屏外围电路图像对应的显示屏外围电路的质量好坏)可以替换为：根据生产阶段信息以及所述缺陷检测结果，确定所述显示屏外围电路图像对应的显示屏外围电路的质量好坏。
具体的，显示屏外围电路的生产厂家、生产环境、以及类型等多种不同的生产阶段信息均可能在显示屏外围电路检测过程中得到不同的缺陷检测结果。对于不同种类的显示屏外围电路，其所经历的生产阶段不同，因而，对上述得到的缺陷检测结果进行分析时，需要结合各显示屏外围电路的生产阶段信息进行来确定显示屏外围电路的质量好坏。
值得说明的是，本申请实施例的缺陷检测模型可以检测出显示屏外围电路图像中存在几类缺陷类型，以及每类缺陷的具体个数，也即，采用MASK RCNN算法得到的缺陷检测模型能够区分属于相同类别的不同缺陷个体。
本申请实施例提供的显示屏外围电路检测方法，通过接收部署在显示屏外围电路生产线上的控制台发送的质量检测请求，该质量检测请求中包含显示屏外围电路生产线上的图像采集设备采集的显示屏外围电路图像，将显示屏外围电路图像输入到缺陷检测模型中得到缺陷检测结果，并根据该缺陷检测结果确定显示屏外围电路图像对应的显示屏外围电路的质量好坏。由于上述缺陷检测模型是对历史缺陷显示屏外围电路图像进行MASK RCNN算 法训练得到的，因而，利用该缺陷检测模型得到的缺陷检测结果的分类精度高，智能化能力强，系统性能有所提高，业务可扩展能力高。
参见图3，为本申请实施例提供的显示屏外围电路检测方法实施例二的流程示意图。在上述实施例的基础上，图3所示实施例中，上述S104(将所述待测图像输入到缺陷检测模型中得到缺陷检测结果)可以通过步骤S301-S302实现，具体如下：
S301，根据负载均衡策略，确定承载处理资源的检测模型服务器。
可选地，在本申请的实施例中，显示屏外围电路生产线上部署有一个服务器组，该服务器组中的服务器数量可以为多个，每个服务器上均运行着缺陷检测模型。可选地，每个服务器上运行的缺陷检测模型均是相同的，因此，每个服务器均可接收控制台发送的质量检测请求，进而可以利用自身承载的缺陷检测模型对显示屏外围电路图像进行质量检测。
作为一种示例，由于部署在显示屏外围电路生产线上的图像采集设备实时采集显示屏外围电路图像，因而，控制台也可以实时向服务器组中的任一服务器发送质量检测请求。
可选地，由于服务器组中每个服务器上运行的缺陷检测模型是相同的，因而，为了提高服务器上的缺陷检测模型的检测效率，保证缺陷检测模型的负载均衡，可以根据预先设置的负载均衡策略，从服务器组中确定一个服务器作为承载处理资源的检测模型服务器，即根据显示屏外围电路生产线上缺陷检测模型的部署情况实时进行负载均衡和调度。
S302，将所述待测图像输入到运行在所述检测模型服务器上的所述缺陷检测模型中得到缺陷检测结果。
可选地，在本申请实施例中，当从服务器组中确定出承载处理资源的检测模型服务器之后，便可以将上述显示屏外围电路图像输入到该检测模型服务器上运行的缺陷检测模型中，利用该缺陷检测模型对显示屏外围电路图像的缺陷进行检测，进而得到缺陷检测结果。可选地，该缺陷检测模型是由训练模块对历史缺陷显示屏外围电路图像中的预设像素类别和实际像素类别进行训练得到的。
本申请实施例提供的显示屏外围电路检测方法，通过根据负载均衡策略，确定承载处理资源的检测模型服务器，并且将上述待测图像输入到运行在上述检测模型服务器上的缺陷检测模型中得到缺陷检测结果，能够实现服务器上的负载均衡，提高显示屏外围电路图像的检测效率，提升显示屏外围电路检测系统的性能。
在一种实现方式中，在上述步骤S302(将所述待测图像输入到运行在所述检测模型服务器上的所述缺陷检测模型中得到缺陷检测结果)之后，还可以包括如下步骤：
若确定所述显示屏外围电路为损坏电路，则执行以下一项或多项操作：
通过控制器向生产管理者发送报警信息；
通过控制器将所述缺陷检测结果作为日志存储到生产数据库中；
通过控制器向所述控制台发送生产控制指令以便消除缺陷；
将所述显示屏外围电路图像和所述缺陷检测结果输入到所述缺陷检测模型中以便优化所述缺陷检测模型。
可选地，在本申请实施例中，测试人员可以根据显示屏外围电路的生产场景和生产阶段信息，预先设置当确定显示屏外围电路为坏屏时的解决方案，比如，通过控制器向生产管理者发送报警信息，和/或，通过控制器将上述缺陷检测结果作为日志存储到生产数据库中，和/或，通过控制器向控制台发送生产控制指令以便消除缺陷，和/或，将上述显示屏外围电路图像和上述缺陷检测结果输入到上述缺陷检测模型中以便优化上述缺陷检测模型等。
具体的，作为一种示例，当根据上述缺陷检测结果确定出显示屏外围电路图像对应的显示屏外围电路是损坏电路，即显示屏外围电路中存在缺陷时，可以发出报警信息，以使生产管理者及时定位缺陷的类别和位置，并且给出解决方案。
作为另一种示例，当根据上述缺陷检测结果确定显示屏外围电路中存在缺陷时，可以通过控制器将上述缺陷检测结果作为日志存储到生产数据库中，即将显示屏外围电路的每个缺陷的类别，和/或各缺陷的轮廓位置作为日志存储到生产数据库中，进而可以将其筛选到训练数据库中，由训练模块(可以是训练引擎等软件程序)根据存在缺陷的显示屏外围电路图像更新上述缺陷检测模型。
作为再一种示例，当根据上述缺陷检测结果确定显示屏外围电路中存在缺陷时，还可以通过控制器向控制台发送生产控制指令以便消除缺陷。即，承载缺陷检测模型的检测模型服务器可以通过控制器确定出缺陷出现的原因，进而根据相应的调整生产流程，也即，检测模型服务器通过控制器向控制台发送生产控制指令以消除显示屏外围电路上出现的缺陷，以减少损坏电路出现的概率。
作为又一种示例，当根据上述缺陷检测结果确定显示屏外围电路中存在缺陷时，也可以直接将上述显示屏外围电路图像和上述缺陷检测结果输入到上述缺陷检测模型中以便优化上述缺陷检测模型，即直接将损坏电路对应的显示屏外围电路图像作为缺陷检测模型的训练集，以优化该缺陷检测模型，进而提高缺陷检测模型的检测准确度。
值得说明的是，本申请实施例并不限定在确定显示屏外围电路为损坏电路时检测模型服务器可执行的上述一项或多项操作，其可根据实际情况进行确定，此处不再赘述。
可选地，对于显示屏外围电路生产线上部署的图像采集设备、控制台、服务器组、控制器、数据库等多个不同的设备，也可以将显示屏外围电路检测方法对应的操作步骤分散到上述多个不同的设备来执行。例如，图像采集设备采集显示屏外围电路图像，控制台根据负载均衡策略，将图像采集设备采集到的显示屏外围电路图像发送给服务器组中的检测模型服务器，由检测模型服务器上运行的缺陷检测模型对显示屏外围电路图像进行预设的预处理之后进行缺陷检测，并给出缺陷检测结果。检测模型服务器可以将缺陷检测结果发送给控制器，一方面由控制器结合实际业务场景，并根据业务需求，根据上述缺陷检测结果做出符合实际业务场景要求的响应，如报警、存储日志、控制生产控制指令等，另一方面，控制器还可以将缺陷检测结果及上述响应的处理行为作为日志存储到生产数据库中，以使训练模块根据训练数据库中的显示屏外围电路图像和缺陷检测结果更新上述得到的缺陷检测模型，该训练数据库中存储的是从生产数据库中筛选的具有缺陷的显示屏外围电路图像和对应的缺陷检测结果等数据。
值得说明的是，对于每一次优化的缺陷检测模型可通过小流量上线的方式逐步取代正在服务器上运行的缺陷检测模型，以达到缺陷检测模型随业务场景和生产阶段信息动态扩展泛化的目的。当本申请实施例中显示屏外围电路检测方法在显示屏外围电路生产线上运行一段时间后，可以人工通过生产数据库中的信息，复查上述缺陷检测和缺陷定位的准确率，然后更新上述训练数据库，重新训练缺陷检测模型，以提高缺陷检测准确率。
下述为本申请装置实施例，可以用于执行本申请方法实施例。对于本申请装置实施例中未披露的细节，请参照本申请方法实施例。
参见图4，为本申请实施例提供的显示屏外围电路检测装置实施例的结构示意图。如图4所示，本申请实施例提供的显示屏外围电路检测装置可以主要包括：接收模块41、预处理模块42、处理模块43和确定模块44。
其中，接收模块41，用于接收部署在显示屏外围电路生产线上的控制台发送的质量检测请求，所述质量检测请求中包含所述显示屏外围电路生产线上的图像采集设备采集的显示屏外围电路图像。
预处理模块42，用于将所述显示屏外围电路图像放大或缩小，得到大小与缺陷检测模型的输入大小要求一致的待测图像，其中，所述缺陷检测模型是用历史缺陷显示屏外围电路图像进行实例分割MASK RCNN算法训练得到的。
处理模块43，用于将所述待测图像输入到缺陷检测模型中得到缺陷检测结果。
确定模块44，用于根据所述缺陷检测结果确定所述显示屏外围电路图像对应的显示 屏外围电路的质量好坏。
图4所示实施例的显示屏外围电路检测装置对应地可用于执行图2所示方法实施例中的步骤，其实现原理和技术效果类似，此处不再赘述。
可选地，所述处理模块43，还用于在所述将所述显示屏外围电路图像输入到缺陷检测模型中得到缺陷检测结果之前，以历史缺陷显示屏外围电路图像的实际像素类别对所述缺陷检测模型进行所述MASK RCNN算法训练，以使所述缺陷检测模型对所述历史缺陷显示屏外围电路图像输出的预测像素类别，与所述实际像素类别之间的损失值低于预设损失阈值。
可选地，所述预处理模块42，还用于在所述将所述显示屏外围电路图像放大或缩小之前对所述显示屏外围电路图像进行图像预处理，其中，所述图像预处理包括下述处理中的一项或多项：裁边、剪切、旋转。
可选地，所述处理模块43，具体用于根据负载均衡策略，确定承载处理资源的检测模型服务器；将所述待测图像输入到运行在所述检测模型服务器上的所述缺陷检测模型中得到缺陷检测结果。
可选地，所述缺陷检测结果，包括：每个缺陷的类别，和/或各缺陷的轮廓位置。
所述确定模块44，具体用于根据生产阶段信息以及所述缺陷检测结果，确定所述显示屏外围电路图像对应的显示屏外围电路的质量好坏。
可选地，所述处理模块43，还用于在所述根据所述缺陷检测结果确定所述显示屏外围电路图像对应的显示屏外围电路的质量好坏之后，若确定所述显示屏外围电路为损坏电路，则执行以下一项或多项操作：
通过控制器向生产管理者发送报警信息；
通过控制器将所述缺陷检测结果作为日志存储到生产数据库中；
通过控制器向所述控制台发送生产控制指令以便消除缺陷；
将所述显示屏外围电路图像和所述缺陷检测结果输入到所述缺陷检测模型中以便优化所述缺陷检测模型。
上述装置实施例的显示屏外围电路检测装置可用于执行图2至图3所示方法实施例的实现方案，具体实现方式和技术效果类似，这里不再赘述。
参见图5，为本发明实施例提供的一种电子设备实施例的结构示意图，该电子设备包括：处理器51、存储器52和计算机程序；其中
存储器52，用于存储所述计算机程序，该存储器还可以是闪存(flash)。所述 计算机程序例如是实现上述方法的应用程序、功能模块等。
处理器51，用于执行所述存储器存储的计算机程序，以实现上述方法中电子设备执行的各个步骤。具体可以参见前面方法实施例中的相关描述。
可选地，存储器52既可以是独立的，也可以跟处理器51集成在一起。
当所述存储器52是独立于处理器51之外的器件时，所述电子设备还可以包括：
总线53，用于连接所述存储器52和处理器51。
本申请还提供一种存储介质，所述存储介质中存储有指令，当其在计算机上运行时，使得计算机执行如图2至图3所示方法实施例的方法。
其中，存储介质可以是计算机存储介质，也可以是通信介质。通信介质包括便于从一个地方向另一个地方传送计算机程序的任何介质。计算机存储介质可以是通用或专用计算机能够存取的任何可用介质。例如，存储介质耦合至处理器，从而使处理器能够从该可读存储介质读取信息，且可向该存储介质写入信息。当然，存储介质也可以是处理器的组成部分。处理器和存储介质可以位于专用集成电路(Application Specific Integrated Circuits，简称：ASIC)中。另外，该ASIC可以位于用户设备中。当然，处理器和存储介质也可以作为分立组件存在于通信设备中。
本申请还提供一种程序产品，该程序产品包括计算机程序，该计算机程序存储在存储介质中。显示屏外围电路检测装置的至少一个处理器可以从存储介质读取该计算机程序，至少一个处理器执行该计算机程序使得显示屏外围电路检测装置执行图2至图3所示方法实施例的方法。
在上述电子设备的实施例中，应理解，处理器可以是中央处理单元(英文：Central Processing Unit，简称：CPU)，还可以是其他通用处理器、数字信号处理器(英文：Digital Signal Processor，简称：DSP)、专用集成电路(英文：Application Specific Integrated Circuit，简称：ASIC)等。通用处理器可以是微处理器或者该处理器也可以是任何常规的处理器等。结合本申请所公开的方法的步骤可以直接体现为硬件处理器执行完成，或者用处理器中的硬件及软件模块组合执行完成。
最后应说明的是：以上各实施例仅用以说明本发明的技术方案，而非对其限制；尽管参照前述各实施例对本发明进行了详细的说明，本领域的普通技术人员应当理解：其依然可以对前述各实施例所记载的技术方案进行修改，或者对其中部分或者全部技术特征进行等同替换；而这些修改或者替换，并不使相应技术方案的本质脱离本发明各实施例技术方案的范围。

Claims (14)

  1. 一种显示屏外围电路检测方法，其特征在于，包括：
    接收部署在显示屏外围电路生产线上的控制台发送的质量检测请求，所述质量检测请求中包含所述显示屏外围电路生产线上的图像采集设备采集的显示屏外围电路图像；
    将所述显示屏外围电路图像放大或缩小，得到大小与缺陷检测模型的输入大小要求一致的待测图像，其中，所述缺陷检测模型是用历史缺陷显示屏外围电路图像进行实例分割Mask RCNN算法训练得到的；
    将所述待测图像输入到缺陷检测模型中得到缺陷检测结果；
    根据所述缺陷检测结果确定所述显示屏外围电路图像对应的显示屏外围电路的质量好坏。
  2. 根据权利要求1所述的方法，其特征在于，在所述将所述显示屏外围电路图像输入到缺陷检测模型中得到缺陷检测结果之前，还包括：
    以历史缺陷显示屏外围电路图像的实际像素类别对所述缺陷检测模型进行所述Mask RCNN算法训练，以使所述缺陷检测模型对所述历史缺陷显示屏外围电路图像输出的预测像素类别，与所述实际像素类别之间的损失值低于预设损失阈值。
  3. 根据权利要求1所述的方法，其特征在于，在所述将所述显示屏外围电路图像放大或缩小之前，还包括：
    对所述显示屏外围电路图像进行图像预处理，其中，所述图像预处理包括下述处理中的一项或多项：
    裁边、剪切、旋转。
  4. 根据权利要求1-3任一项所述的方法，其特征在于，所述将所述待测图像输入到缺陷检测模型中得到缺陷检测结果，包括：
    根据负载均衡策略，确定承载处理资源的检测模型服务器；
    将所述待测图像输入到运行在所述检测模型服务器上的所述缺陷检测模型中得到缺陷检测结果。
  5. 根据权利要求1-3任一项所述的方法，其特征在于，所述缺陷检测结果，包括：每个缺陷的类别，和/或各缺陷的轮廓位置；
    所述根据所述缺陷检测结果确定所述显示屏外围电路图像对应的显示屏外围电路的质量好坏，包括：
    根据生产阶段信息以及所述缺陷检测结果，确定所述显示屏外围电路图像对应的显示 屏外围电路的质量好坏。
  6. 根据权利要求1-3任一项所述的方法，其特征在于，在所述根据所述缺陷检测结果确定所述显示屏外围电路图像对应的显示屏外围电路的质量好坏之后，还包括：
    若确定所述显示屏外围电路为损坏电路，则执行以下一项或多项操作：
    通过控制器向生产管理者发送报警信息；
    通过控制器将所述缺陷检测结果作为日志存储到生产数据库中；
    通过控制器向所述控制台发送生产控制指令以便消除缺陷；
    将所述显示屏外围电路图像和所述缺陷检测结果输入到所述缺陷检测模型中以便优化所述缺陷检测模型。
  7. 一种显示屏外围电路检测装置，其特征在于，包括：
    接收模块，用于接收部署在显示屏外围电路生产线上的控制台发送的质量检测请求，所述质量检测请求中包含所述显示屏外围电路生产线上的图像采集设备采集的显示屏外围电路图像；
    预处理模块，用于将所述显示屏外围电路图像放大或缩小，得到大小与缺陷检测模型的输入大小要求一致的待测图像，其中，所述缺陷检测模型是用历史缺陷显示屏外围电路图像进行实例分割Mask RCNN算法训练得到的；
    处理模块，用于将所述待测图像输入到缺陷检测模型中得到缺陷检测结果；
    确定模块，用于根据所述缺陷检测结果确定所述显示屏外围电路图像对应的显示屏外围电路的质量好坏。
  8. 根据权利要求7所述的装置，其特征在于，
    所述处理模块，还用于在所述将所述显示屏外围电路图像输入到缺陷检测模型中得到缺陷检测结果之前，以历史缺陷显示屏外围电路图像的实际像素类别对所述缺陷检测模型进行所述Mask RCNN算法训练，以使所述缺陷检测模型对所述历史缺陷显示屏外围电路图像输出的预测像素类别，与所述实际像素类别之间的损失值低于预设损失阈值。
  9. 根据权利要求7所述的装置，其特征在于，
    所述预处理模块，还用于在所述将所述显示屏外围电路图像放大或缩小之前对所述显示屏外围电路图像进行图像预处理，其中，所述图像预处理包括下述处理中的一项或多项：裁边、剪切、旋转。
  10. 根据权利要求7-9任一项所述的装置，其特征在于，
    所述处理模块，具体用于根据负载均衡策略，确定承载处理资源的检测模型服务器； 将所述待测图像输入到运行在所述检测模型服务器上的所述缺陷检测模型中得到缺陷检测结果。
  11. 根据权利要求7-9任一项所述的装置，其特征在于，所述缺陷检测结果，包括：每个缺陷的类别，和/或各缺陷的轮廓位置；
    所述确定模块，具体用于根据生产阶段信息以及所述缺陷检测结果，确定所述显示屏外围电路图像对应的显示屏外围电路的质量好坏。
  12. 根据权利要求7-9任一项所述的装置，其特征在于，所述处理模块，还用于在所述根据所述缺陷检测结果确定所述显示屏外围电路图像对应的显示屏外围电路的质量好坏之后，若确定所述显示屏外围电路为损坏电路，则执行以下一项或多项操作：
    通过控制器向生产管理者发送报警信息；
    通过控制器将所述缺陷检测结果作为日志存储到生产数据库中；
    通过控制器向所述控制台发送生产控制指令以便消除缺陷；
    将所述显示屏外围电路图像和所述缺陷检测结果输入到所述缺陷检测模型中以便优化所述缺陷检测模型。
  13. 一种电子设备，包括处理器、存储器及存储在所述存储器上并可在处理器上运行的计算机程序，其特征在于，所述处理器执行所述程序时实现如上述权利要求1-6任一项所述的方法。
  14. 一种存储介质，其特征在于，所述存储介质中存储有指令，当其在计算机上运行时，使得计算机执行如权利要求1-6任一项所述的方法。
PCT/CN2019/085912 2018-07-02 2019-05-08 显示屏外围电路检测方法、装置、电子设备及存储介质 WO2020007118A1 (zh)

Priority Applications (3)

Application Number Priority Date Filing Date Title
JP2019563795A JP7025452B2 (ja) 2018-07-02 2019-05-08 ディスプレイスクリーン周辺回路の検出方法、ディスプレイスクリーン周辺回路の検出装置、電子機器及び記憶媒体
KR1020197034316A KR102320371B1 (ko) 2018-07-02 2019-05-08 디스플레이 스크린 주변 회로 검출 방법, 장치, 전자기기 및 저장매체
US16/995,898 US20200380899A1 (en) 2018-07-02 2020-08-18 Method and apparatus for detecting peripheral circuit of display screen, electronic device, and storage medium

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201810709836.7A CN109085174A (zh) 2018-07-02 2018-07-02 显示屏外围电路检测方法、装置、电子设备及存储介质
CN201810709836.7 2018-07-02

Related Child Applications (1)

Application Number Title Priority Date Filing Date
US16/995,898 Continuation US20200380899A1 (en) 2018-07-02 2020-08-18 Method and apparatus for detecting peripheral circuit of display screen, electronic device, and storage medium

Publications (1)

Publication Number Publication Date
WO2020007118A1 true WO2020007118A1 (zh) 2020-01-09

Family

ID=64836902

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2019/085912 WO2020007118A1 (zh) 2018-07-02 2019-05-08 显示屏外围电路检测方法、装置、电子设备及存储介质

Country Status (5)

Country Link
US (1) US20200380899A1 (zh)
JP (1) JP7025452B2 (zh)
KR (1) KR102320371B1 (zh)
CN (1) CN109085174A (zh)
WO (1) WO2020007118A1 (zh)

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111402250A (zh) * 2020-03-26 2020-07-10 中国联合网络通信集团有限公司 基于边缘计算的机器视觉缺陷检测方法和平台
CN113256570A (zh) * 2021-05-10 2021-08-13 郑州轻工业大学 基于人工智能的视觉信息处理方法、装置、设备及介质
CN113256623A (zh) * 2021-06-29 2021-08-13 南昌工程学院 一种基于改进mask rcnn的fpc缺陷检测方法
WO2022019110A1 (ja) * 2020-07-21 2022-01-27 株式会社シバサキ プログラム、情報処理装置、情報処理方法及びモデル生成方法

Families Citing this family (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109085174A (zh) * 2018-07-02 2018-12-25 北京百度网讯科技有限公司 显示屏外围电路检测方法、装置、电子设备及存储介质
CN111402214A (zh) * 2020-03-07 2020-07-10 西南交通大学 基于神经网络的接触网吊弦载流环断裂缺陷自动检测方法
US11435298B2 (en) * 2020-07-24 2022-09-06 Hewlett Packard Enterprise Development Lp Circuit board anomaly indication
CN112712504B (zh) * 2020-12-30 2023-08-15 广东粤云工业互联网创新科技有限公司 基于云端的工件检测方法及系统、计算机可读存储介质
CN112816496B (zh) * 2021-01-05 2022-09-23 广州市华颉电子科技有限公司 一种汽车域控制器的接口装配质量自动光学检测方法及装置
CN114943855A (zh) * 2021-02-09 2022-08-26 富泰华工业（深圳）有限公司 图像分类标注方法、装置、电子设备及存储介质
US20220318667A1 (en) * 2021-03-30 2022-10-06 Accenture Global Solutions Limited Intelligent real-time defect prediction, detection, and ai driven automated correction solution
CN116229856B (zh) * 2023-05-10 2023-07-28 山西晋聚轩科技有限公司 一种计算机用自动控制的屏幕检测系统及方法

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107886500A (zh) * 2017-10-13 2018-04-06 北京邮电大学 一种基于机器视觉和机器学习的产品生产监控方法及系统
CN108090897A (zh) * 2017-12-18 2018-05-29 川亿电脑(深圳)有限公司 印刷线路板缺陷的检测方法、检测装置及存储介质
CN108230317A (zh) * 2018-01-09 2018-06-29 北京百度网讯科技有限公司 钢板缺陷检测分类方法、装置、设备及计算机可读介质
CN109085174A (zh) * 2018-07-02 2018-12-25 北京百度网讯科技有限公司 显示屏外围电路检测方法、装置、电子设备及存储介质

Family Cites Families (14)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP3920003B2 (ja) * 2000-04-25 2007-05-30 株式会社ルネサステクノロジ 検査データ処理方法およびその装置
KR20090131000A (ko) * 2008-06-17 2009-12-28 아주하이텍(주) 플렉시블 인쇄회로기판의 통합 검사 시스템 및 그 방법
CN101726493B (zh) * 2009-12-02 2012-07-18 中国建筑材料科学研究总院 一种水泥基材料收缩与开裂性能检测方法及装置
US8995747B2 (en) * 2010-07-29 2015-03-31 Sharp Laboratories Of America, Inc. Methods, systems and apparatus for defect detection and classification
WO2016004063A1 (en) * 2014-06-30 2016-01-07 Amazon Technologies, Inc. Feature processing recipes for machine learning
US10650508B2 (en) * 2014-12-03 2020-05-12 Kla-Tencor Corporation Automatic defect classification without sampling and feature selection
JP2018005640A (ja) * 2016-07-04 2018-01-11 タカノ株式会社 分類器生成装置、画像検査装置、及び、プログラム
JP2018045673A (ja) * 2016-09-09 2018-03-22 株式会社Ｓｃｒｅｅｎホールディングス 分類器構築方法、画像分類方法、分類器構築装置および画像分類装置
CN108073932A (zh) * 2016-11-16 2018-05-25 中国科学院沈阳计算技术研究所有限公司 一种基于Gabor滤波的工件图像特征提取与识别方法
US10685295B1 (en) * 2016-12-29 2020-06-16 X Development Llc Allocating resources for a machine learning model
WO2018191698A1 (en) * 2017-04-13 2018-10-18 Instrumental, Inc. Method for predicting defects in assembly units
CN107561738B (zh) * 2017-08-30 2020-06-12 湖南理工学院 基于fcn的tft-lcd表面缺陷快速检测方法
CN107729882A (zh) * 2017-11-19 2018-02-23 济源维恩科技开发有限公司 基于图像识别的情绪识别判定方法
CN108089753B (zh) * 2017-12-28 2021-03-09 安徽慧视金瞳科技有限公司 一种利用Faster-RCNN对指尖位置进行预测的定位方法

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107886500A (zh) * 2017-10-13 2018-04-06 北京邮电大学 一种基于机器视觉和机器学习的产品生产监控方法及系统
CN108090897A (zh) * 2017-12-18 2018-05-29 川亿电脑(深圳)有限公司 印刷线路板缺陷的检测方法、检测装置及存储介质
CN108230317A (zh) * 2018-01-09 2018-06-29 北京百度网讯科技有限公司 钢板缺陷检测分类方法、装置、设备及计算机可读介质
CN109085174A (zh) * 2018-07-02 2018-12-25 北京百度网讯科技有限公司 显示屏外围电路检测方法、装置、电子设备及存储介质

Cited By (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111402250A (zh) * 2020-03-26 2020-07-10 中国联合网络通信集团有限公司 基于边缘计算的机器视觉缺陷检测方法和平台
WO2022019110A1 (ja) * 2020-07-21 2022-01-27 株式会社シバサキ プログラム、情報処理装置、情報処理方法及びモデル生成方法
JP7391285B2 (ja) 2020-07-21 2023-12-05 株式会社シバサキ プログラム、情報処理装置、情報処理方法及びモデル生成方法
CN113256570A (zh) * 2021-05-10 2021-08-13 郑州轻工业大学 基于人工智能的视觉信息处理方法、装置、设备及介质
CN113256623A (zh) * 2021-06-29 2021-08-13 南昌工程学院 一种基于改进mask rcnn的fpc缺陷检测方法
CN113256623B (zh) * 2021-06-29 2021-10-26 南昌工程学院 一种基于改进mask rcnn的fpc缺陷检测方法

Also Published As

Publication number Publication date
JP2020530126A (ja) 2020-10-15
US20200380899A1 (en) 2020-12-03
KR20200004822A (ko) 2020-01-14
JP7025452B2 (ja) 2022-02-24
CN109085174A (zh) 2018-12-25
KR102320371B1 (ko) 2021-11-01

Similar Documents

Publication Publication Date Title
WO2020007118A1 (zh) 显示屏外围电路检测方法、装置、电子设备及存储介质
US11380232B2 (en) Display screen quality detection method, apparatus, electronic device and storage medium
US11488294B2 (en) Method for detecting display screen quality, apparatus, electronic device and storage medium
WO2020007119A1 (zh) 显示屏外围电路检测方法、装置、电子设备及存储介质
EP3937128A2 (en) Image defect detection method and apparatus, electronic device, storage medium and product
CN110060237B (zh) 一种故障检测方法、装置、设备及系统
CN111507958B (zh) 目标检测方法、检测模型的训练方法及电子设备
CN109064446A (zh) 显示屏质量检测方法、装置、电子设备及存储介质
CN109087281A (zh) 显示屏外围电路检测方法、装置、电子设备及存储介质
CN108961239A (zh) 连铸坯质量检测方法、装置、电子设备及存储介质
CN115131283A (zh) 目标对象的缺陷检测、模型训练方法、装置、设备及介质
CN116385430A (zh) 一种机器视觉瑕疵检测方法、装置、介质及设备
JP7446060B2 (ja) 情報処理装置、プログラム及び情報処理方法
CN113537037A (zh) 路面病害识别方法、系统、电子设备及存储介质
CN117351472A (zh) 烟叶信息检测方法、装置及电子设备
KR20220151130A (ko) 영상 처리 방법, 장치, 전자 기기, 저장 매체 및 컴퓨터 프로그램
CN109087282A (zh) 显示屏外围电路检测方法、装置、电子设备及存储介质
CN112001336A (zh) 行人越界报警方法、装置、设备及系统
CN117349734B (zh) 水表设备识别方法、装置、电子设备及存储介质
CN115542100B (zh) 绝缘子故障检测方法、装置、设备及介质
CN116259092A (zh) 无人机航拍人脸识别方法、系统、设备以及可读存储介质
CN117789109A (zh) 一种工业场景下无人值守的异常行为检测方法和系统
CN114842073A (zh) 图像数据扩增方法、装置、设备、介质和计算机程序产品
CN116123040A (zh) 一种基于多模态数据融合的风机叶片状态检测方法及系统

Legal Events

Date Code Title Description
ENP Entry into the national phase

Ref document number: 2019563795

Country of ref document: JP

Kind code of ref document: A

ENP Entry into the national phase

Ref document number: 20197034316

Country of ref document: KR

Kind code of ref document: A

121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 19831442

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 19831442

Country of ref document: EP

Kind code of ref document: A1