Machine vision detection method and platform for backlight mura defect

Abstract

The application provides a machine vision detection method and platform for backlight Mura defects, and relates to the technical field of machine vision detection. The method comprises the following steps: receiving a display task of a liquid crystal display; mining control parameters of a backlight source according to real-time data of a display environment of the liquid crystal display and display requirement information; controlling the liquid crystal display to execute the display task according to a backlight control strategy; performing multi-level Mura defect detection on a display image according to a plurality of machine vision models to obtain a Mura defect detection atlas; performing correlation tracing on the backlight source according to the Mura defect detection atlas to obtain a defect backlight tracing atlas; and optimizing and adjusting the backlight control strategy according to the defect backlight tracing atlas. The application can solve the technical problem of low machine vision detection accuracy of backlight Mura defects in the prior art, and achieve the technical effect of improving machine vision detection accuracy.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of machine vision detection, and particularly relates to a machine vision detection method and platform for backlight Mura defects. BACKGROUND

[0002] With the rapid development of liquid crystal display technology and its wide application in the fields of television, notebook computer, vehicle terminal, smart wearable device, etc., the brightness uniformity, color consistency and image stability of the display panel have become key indicators for measuring product performance and user experience.

[0003] At present, the existing Mura defect detection and backlight control method mainly relies on manual inspection or a single image analysis model to identify and determine the defects of the display panel, and such method has problems of low detection precision, high false positive rate and poor adaptability to complex defect patterns. Meanwhile, the traditional technology usually only focuses on the quality analysis of the display image itself, lacks real-time monitoring and multi-dimensional correlation analysis of the running state of the backlight, and cannot effectively identify the root cause of the defects.

[0004] In summary, in the prior art, due to the lack of real-time monitoring and analysis of the multi-dimensional running state of the backlight of the liquid crystal display, the Mura defects cannot be detected and traced in time and accurately, and the display quality is difficult to be effectively guaranteed. SUMMARY

[0005] The purpose of the present application is to provide a machine vision detection method and platform for backlight Mura defects, so as to solve the technical problem in the prior art that due to the lack of real-time monitoring and analysis of the multi-dimensional running state of the backlight of the liquid crystal display, the Mura defects cannot be detected and traced in time and accurately, and the display quality is difficult to be effectively guaranteed.

[0006] In view of the above problems, the present application provides a machine vision detection method and platform for backlight Mura defects.

[0007] In a first aspect, the present application provides a machine vision detection method for backlight Mura defects, which is realized by a machine vision detection platform for backlight Mura defects and includes the following steps: receiving a display task of a liquid crystal display, wherein the liquid crystal display includes a backlight and a liquid crystal panel, and the display task includes display requirement information; performing control parameter mining on the backlight according to real-time data of a display environment of the liquid crystal display and the display requirement information, and determining a backlight control strategy; controlling the liquid crystal display to execute the display task according to the backlight control strategy, and obtaining a display image; performing multi-level Mura defect detection on the display image according to a plurality of machine vision models, and obtaining a Mura defect detection atlas; performing correlation tracing on the backlight according to the Mura defect detection atlas, and obtaining a defect backlight tracing atlas; and optimizing and adjusting the backlight control strategy according to the defect backlight tracing atlas.

[0008] Preferably, the machine vision detection method for backlight Mura defects further includes the following steps: interconnecting backlight of the same type, and obtaining a backlight interconnected group; constructing a display task feature matrix according to the real-time data of the display environment and the display requirement information; performing control parameter historical retrieval on the backlight interconnected group according to the display task feature matrix, and obtaining a backlight control history set; performing display quality fitting analysis according to the backlight control history set, and obtaining a display quality fitting sequence; and performing optimization identification on the backlight control history set according to the display quality fitting sequence, and generating the backlight control strategy.

[0009] Preferably, the machine vision detection method for backlight Mura defects further includes the following steps: extracting a first control history scheme according to the backlight control history set; performing display quality evaluation retrieval according to the first control history scheme, and obtaining a first historical display quality evaluation sequence; performing central value calculation according to the first historical display quality evaluation sequence, and obtaining a first display quality fitting coefficient; adding the first display quality fitting coefficient to the display quality fitting sequence, and continuing to perform display quality fitting analysis based on the backlight control history set, and updating the display quality fitting sequence.

[0010] Preferably, the machine vision detection method for backlight Mura defects further comprises: image fitting according to the display requirement information to determine a display requirement image; performing brightness defect detection on the display image according to the plurality of machine vision models based on the display requirement image to obtain a display brightness defect detection result; performing color defect detection on the display image according to the plurality of machine vision models based on the display requirement image to obtain a display color defect detection result; performing pattern defect detection on the display image according to the plurality of machine vision models based on the display requirement image to obtain a display pattern defect detection result; and generating the Mura defect detection atlas according to the display brightness defect detection result, the display color defect detection result, and the display pattern defect detection result.

[0011] Preferably, the machine vision detection method for backlight Mura defects further comprises: respectively performing brightness feature extraction on the display requirement image and the display image to obtain requirement brightness feature point cloud and display brightness feature point cloud; performing point position alignment processing on the requirement brightness feature point cloud and the display brightness feature point cloud to obtain brightness feature comparison point cloud; respectively supervising training of display brightness defect detection record sets according to the plurality of machine vision models to obtain a plurality of brightness defect detectors; performing integrated fusion training according to the plurality of brightness defect detectors to obtain a brightness defect detection channel; and inputting the brightness feature comparison point cloud into the brightness defect detection channel to generate the display brightness defect detection result.

[0012] Preferably, the machine vision detection method for backlight Mura defects further comprises: based on the display image, synchronously collecting real-time monitoring parameters of the backlight to obtain a backlight monitoring sequence; associating and tracing the backlight monitoring sequence according to the display brightness defect detection result to obtain a brightness defect association and tracing result; associating and tracing the backlight monitoring sequence according to the display color defect detection result to obtain a color defect association and tracing result; associating and tracing the backlight monitoring sequence according to the display pattern defect detection result to obtain a pattern defect association and tracing result; and collating the brightness defect association and tracing result, the color defect association and tracing result, and the pattern defect association and tracing result to generate a defect backlight tracing atlas.

[0013] Preferably, the machine vision detection method for backlight Mura defects further comprises: performing multi-dimensional anomaly detection according to the backlight monitoring sequence to determine a brightness anomaly detection result, a color temperature anomaly detection result, a current-voltage anomaly detection result, and a temperature anomaly detection result; performing cause tracing on the display brightness defect detection result according to the brightness anomaly detection result to obtain a first brightness defect tracing feature; performing cause tracing on the display brightness defect detection result according to the color temperature anomaly detection result to obtain a second brightness defect tracing feature; performing cause tracing on the display brightness defect detection result according to the current-voltage anomaly detection result to obtain a third brightness defect tracing feature; performing cause tracing on the display brightness defect detection result according to the temperature anomaly detection result to obtain a fourth brightness defect tracing feature; and collating the first brightness defect tracing feature, the second brightness defect tracing feature, the third brightness defect tracing feature, and the fourth brightness defect tracing feature to generate the brightness defect correlation traceability result.

[0014] Preferably, the machine vision detection method for backlight Mura defects further comprises: generating a backlight anomaly early warning signal according to the brightness anomaly detection result, the color temperature anomaly detection result, the current-voltage anomaly detection result, and the temperature anomaly detection result.

[0015] Preferably, the machine vision detection method for backlight Mura defects further comprises: generating a display early warning signal according to the Mura defect detection map.

[0016] In a second aspect, the present application further provides a machine vision detection platform for backlight Mura defects, which is used to execute the machine vision detection method for backlight Mura defects as described in the first aspect, and comprises: a display task receiving module, which is used to receive a display task of a liquid crystal display, the liquid crystal display comprising a backlight and a liquid crystal panel, and the display task comprising display requirement information; a backlight control strategy determining module, which is used to perform control parameter mining on the backlight according to display environment real-time data of the liquid crystal display and the display requirement information to determine a backlight control strategy; a display image obtaining module, which is used to control the liquid crystal display to execute the display task according to the backlight control strategy to obtain a display image; a Mura defect detection map obtaining module, which is used to perform multi-level Mura defect detection on the display image according to a plurality of machine vision models to obtain a Mura defect detection map; a defect backlight traceability map obtaining module, which is used to perform correlation traceability on the backlight according to the Mura defect detection map to obtain a defect backlight traceability map; and an optimization adjustment module, which is used to perform optimization adjustment on the backlight control strategy according to the defect backlight traceability map.

[0017] The technical solutions provided in the application have at least the following technical effects or advantages: by achieving the technical target of multi-level intelligent detection and adaptive control based on the association analysis of the running state of the backlight source and display defects, the technical effects of real-time identification and tracing of the root cause of Mura defects, dynamic optimization of backlight control strategies, improvement of display uniformity and color consistency, and improvement of production yield and long-term operation reliability of equipment are achieved.

[0018] The above description is only a summary of the technical solutions of the application. In order to more clearly understand the technical means of the application, the specific embodiments of the application can be implemented according to the content of the description, and in order to make the above and other purposes, characteristics and advantages of the application more obvious and easy to understand, the following specific embodiments of the application are described. It should be understood that the content described in this part is not intended to identify the key or important features of the embodiments of the application, nor is it intended to limit the scope of the application. Other features of the application will become apparent from the following description. BRIEF DESCRIPTION OF DRAWINGS

[0019] In order to more clearly illustrate the technical solutions in the application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiments or the prior art description. Obviously, the drawings in the following description are only exemplary, and for those skilled in the art, other drawings can be obtained without creative labor on the basis of the provided drawings.

[0020] Figure 1 The flowchart of the machine vision detection method for the backlight source Mura defect of the application.

[0021] Figure 2 The structural schematic diagram of the machine vision detection platform for the backlight source Mura defect of the application.

[0022] Explanation of reference signs: display task receiving module 1, backlight source control strategy determination module 2, display image obtaining module 3, Mura defect detection atlas obtaining module 4, defect backlight source tracing atlas obtaining module 5, and optimization adjustment module 6. DETAILED DESCRIPTION

[0023] The application provides a machine vision detection method and platform for backlight Mura defects, solves the technical problem that in the prior art, due to the lack of real-time monitoring and analysis of the multi-dimensional running state of the backlight of a liquid crystal display, Mura defects cannot be detected and traced in time and accurately, and the display quality is difficult to be effectively guaranteed, achieves the technical goal of multi-level intelligent detection and adaptive control based on the correlation analysis of the running state of the backlight and display defects, and achieves the technical effects of being capable of identifying and tracing the root cause of Mura defects in real time, dynamically optimizing the backlight control strategy, improving display uniformity and color consistency, and improving production yield and long-term operation reliability of equipment.

[0024] In the following, the technical solutions in the application will be described clearly and completely with reference to the drawings. Obviously, the described embodiments are only a part of the embodiments of the application, rather than all the embodiments of the application. It should be understood that the application is not limited by the example embodiments described herein. Based on the embodiments of the application, all other embodiments obtained by a person of ordinary skill in the art without creative work fall within the scope of protection of the application. In addition, it should be noted that, for the convenience of description, only parts related to the application are shown in the drawings, rather than all.

[0025] Embodiment one, please refer to the accompanying Figure 1 The application provides a machine vision detection method for backlight Mura defects, which is applied to a machine vision detection platform for backlight Mura defects, and specifically includes the following steps:

[0026] S1: receiving a display task of a liquid crystal display, the liquid crystal display including a backlight and a liquid crystal panel, and the display task including display requirement information.

[0027] Specifically, the display task of the liquid crystal display is received to obtain the current display operation instruction of the liquid crystal display, and the display operation instruction contains the image or video display content to be executed by the liquid crystal display and the corresponding display parameters such as brightness, color, resolution, etc. Further, the structure of the liquid crystal display includes a backlight for providing illumination and a liquid crystal panel for forming an image, the backlight provides uniform illumination for the liquid crystal panel, and the liquid crystal panel realizes image display by adjusting light transmission through liquid crystal pixels. Further, the display requirement information included in the display task is used to define the specific display standards that should be met by the liquid crystal display when executing the display task, and the display requirement information can include image brightness target value, color space requirement, pattern detail requirement, refresh frequency and other parameters to ensure the accuracy and stability of the display effect of the liquid crystal display.

[0028] S2: performing control parameter mining on the backlight according to the display environment real-time data of the liquid crystal display and the display requirement information, and determining a backlight control strategy.

[0029] Further, the application further comprises: interconnecting the backlight sources of the same type to obtain a backlight source interconnected group; constructing a display task feature matrix according to the display environment real-time data and the display requirement information; performing control parameter historical retrieval on the backlight source interconnected group according to the display task feature matrix to obtain a backlight source control history set; performing display quality fitting analysis according to the backlight source control history set to obtain a display quality fitting sequence; and performing optimization identification on the backlight source control history set according to the display quality fitting sequence to generate the backlight source control strategy.

[0030] Further, the application further comprises: extracting a first control history scheme according to the backlight source control history set; performing display quality evaluation retrieval according to the first control history scheme to obtain a first historical display quality evaluation sequence; performing central value calculation according to the first historical display quality evaluation sequence to obtain a first display quality fitting coefficient; adding the first display quality fitting coefficient to the display quality fitting sequence, and continuing to perform display quality fitting analysis based on the backlight source control history set to update the display quality fitting sequence.

[0031] Specifically, the backlight sources are interconnected by the same type of backlight sources, that is, the backlight source units with the same technical specifications and electrical characteristics in the liquid crystal display are connected by electrical or communication means to form an interconnected group, so as to facilitate unified control and data acquisition, and obtain a backlight source interconnected group.

[0032] Then, a display task feature matrix is constructed according to the display environment real-time data and the display requirement information. For example, the environmental light intensity is 300 lux, the working temperature is 25 degrees Celsius, and the brightness, color and resolution information required by the display task are arranged in a specific dimension to form a matrix, each row of the matrix represents a display condition combination, and each column represents a feature parameter, thereby providing a structured data basis for subsequent backlight source control.

[0033] Subsequently, control parameter historical retrieval is performed on the backlight source interconnected group according to the display task feature matrix to obtain a backlight source control history set, that is, the parameter information in the display task feature matrix is used to find the backlight source control scheme and effect data matched with the current task feature in the historical control record. The backlight source control history set includes current, voltage and dimming mode data under different brightness, color and environmental conditions, and corresponding display quality evaluation results.

[0034] Further, a first control history scheme is randomly extracted from the backlight control history set. Then, display quality evaluation retrieval is performed according to the first control history scheme, corresponding display effect data is searched in the history display quality evaluation record, including brightness uniformity, color accuracy and image clarity, etc., a display quality evaluation sequence arranged in time or task order is generated, and a first history display quality evaluation sequence is obtained.

[0035] Subsequently, the central value calculation is performed according to the first history display quality evaluation sequence, that is, the statistical analysis is performed on the first history display quality evaluation sequence, and a fitting coefficient representing the overall display effect of the control scheme is obtained by calculating the average value, the median or the weighted mean, etc., to obtain a first display quality fitting coefficient.

[0036] Then, the first display quality fitting coefficient is added to the display quality fitting sequence, and the display quality fitting analysis is continued based on the backlight control history set, the display quality fitting sequence is updated, that is, the first display quality fitting coefficient is included in the overall display quality fitting sequence, and the above extraction, evaluation and fitting process is repeated in combination with other control schemes in the backlight control history set, to form an updated display quality fitting sequence, so that the display quality change law under different control schemes can be fully reflected, for example, after adding a second control scheme, the display quality coefficient in the fitting sequence may increase from 91 to 92, showing the overall display effect improvement trend.

[0037] Finally, the backlight control history set is optimized and identified according to the display quality fitting sequence, that is, the contribution of different control parameter combinations to the display effect is analyzed, the optimal parameter combination is selected to form a new backlight control strategy, and the backlight control strategy is generated, including the best driving current, voltage, dimming mode and synchronous control sequence, to ensure that the liquid crystal display reaches the best display effect under the current display task and environmental conditions. For example, the optimization and identification are performed by extracting the display quality fitting coefficient with the maximum value.

[0038] S3: controlling the liquid crystal display to perform the display task according to the backlight control strategy, and obtaining a display image.

[0039] Specifically, the backlight control strategy refers to the control scheme formulated for the luminous intensity, brightness distribution, color temperature adjustment, and on-off timing of the backlight module inside the liquid crystal display. Combining factors such as input signal characteristics, ambient light conditions, and power consumption limitations, the output state of the backlight module is dynamically adjusted through the coordinated action of hardware drive circuits and software algorithms to achieve comprehensive goals such as optimal energy consumption, enhanced visual effects, and extended display life. The liquid crystal display refers to a display device that uses liquid crystal material to adjust the transmittance under the action of an electric field to form an image. It consists of components such as a liquid crystal panel, a backlight module, a drive circuit, and a control chip. The liquid crystal panel is responsible for modulating light, the backlight module provides the necessary light source, and the drive and control system processes and outputs image signals. The display task refers to the image presentation operation that the liquid crystal display needs to complete in a specific application scenario, including video signal decoding, pixel driving, brightness and contrast adjustment, and a series of processing steps. Its goal is to convert input signals into visual image output according to preset instructions. The display image refers to the final image content visible to the human eye after backlight adjustment and liquid crystal modulation. It has multiple parameter characteristics such as resolution, color depth, brightness, and dynamic range. Its quality is directly affected by the backlight control strategy and the execution effect of the display task.

[0040] S4: Perform multi-level Mura defect detection on the display image according to the plurality of machine vision models to obtain a Mura defect detection map.

[0041] Further, the present application also includes: performing image fitting according to the display requirement information to determine a display requirement image; performing brightness defect detection on the display image according to the plurality of machine vision models based on the display requirement image to obtain a display brightness defect detection result; performing color defect detection on the display image according to the plurality of machine vision models based on the display requirement image to obtain a display color defect detection result; performing pattern defect detection on the display image according to the plurality of machine vision models based on the display requirement image to obtain a display pattern defect detection result; and generating the Mura defect detection map according to the display brightness defect detection result, the display color defect detection result, and the display pattern defect detection result.

[0042] Further, the application further comprises: performing brightness feature extraction on the display requirement image and the display image respectively to obtain requirement brightness feature point cloud and display brightness feature point cloud; performing point position alignment processing on the requirement brightness feature point cloud and the display brightness feature point cloud to obtain brightness feature comparison point cloud; performing supervised training on the display brightness defect detection record set according to the plurality of machine vision models to obtain a plurality of brightness defect detectors; performing integrated fusion training according to the plurality of brightness defect detectors to obtain a brightness defect detection channel; inputting the brightness feature comparison point cloud into the brightness defect detection channel to generate the display brightness defect detection result.

[0043] Further, the application further comprises: generating a display warning signal according to the Mura defect detection map.

[0044] Specifically, the display requirement information refers to the image output characteristic parameter set required by the liquid crystal display device or other types of display terminals in a specific application scenario. Image fitting according to the display requirement information means using interpolation, resampling or regression modeling methods to convert the input display requirement information into a target image with specific structure and attributes, so as to construct a display requirement image that can reflect the ideal display effect. The display requirement image refers to the standardized image sample that should theoretically be presented on the display terminal after fitting processing.

[0045] Further, brightness feature extraction is performed on the display requirement image and the display image respectively, which refers to the process of extracting multi-dimensional data features that can represent the brightness distribution characteristics from the input image data, including pixel intensity value, brightness gradient distribution, local contrast curve and regional brightness variance, etc., which are used to quantitatively describe the brightness structure of the image, and then obtain requirement brightness feature point cloud and display brightness feature point cloud, i.e. a high-dimensional data set composed of a large number of brightness feature points, each point contains position coordinates and brightness attribute information, which can reflect the distribution form of image brightness in space.

[0046] The point position alignment processing on the requirement brightness feature point cloud and the display brightness feature point cloud is to align the requirement brightness feature point cloud and the display brightness feature point cloud in coordinates and feature dimensions through spatial registration, similarity transformation or least mean square error matching algorithm, so that they have a one-to-one point position relationship in space, and obtain the brightness feature comparison point cloud, which can eliminate the spatial deviation caused by shooting angle, display area deviation or equipment installation error, and ensure the accuracy of subsequent comparison.

[0047] The display brightness defect detection record set refers to a sample data set accumulated from historical detection tasks, containing input data, defect labels, and detection labels, which contains typical defect information such as brightness unevenness, light leakage, local dark spots, etc. According to multiple machine vision models, the display brightness defect detection record set is supervised trained, that is, using the labeled display brightness defect detection record set as training data, the multiple machine vision models are supervised parameter learning, so that the model can automatically identify and classify different types of brightness defects, and then obtain multiple brightness defect detectors. Among them, through the supervised training of different models, multiple brightness defect detectors can be obtained, each detector has high sensitivity and recognition ability to specific type of brightness anomaly.

[0048] According to the integrated fusion training of multiple brightness defect detectors, that is, the joint optimization of multiple brightness defect detectors is carried out to fully utilize the detection advantages of each other, and the overall detection performance is improved through weighted integration, voting fusion or feature sharing, and then the brightness defect detection channel is obtained. The brightness defect detection channel generated after integration and fusion is a composite detection structure with multi-model collaborative discrimination ability, which can maintain high robustness and generalization ability in different brightness anomaly scenes.

[0049] Finally, the brightness feature comparison point cloud is input into the brightness defect detection channel, that is, the fused brightness difference data is input as input features into the detection model after integrated training for automatic analysis, so as to output the display brightness defect detection result. The display brightness defect detection result is a structured representation of brightness anomaly, including defect type, position coordinate, area size and severity.

[0050] Further, based on the display requirement image, the display image is detected for color defects according to multiple machine vision models, that is, the display requirement image is taken as an ideal reference, and the color features of the display image are compared and analyzed by multiple machine vision models to obtain the display color defect detection result. Among them, the multiple machine vision models refer to a set of intelligent recognition algorithms with different structures such as deep convolutional neural network, feature classifier, clustering analyzer, etc., which are used to extract color information from different dimensions.

[0051] Based on the display requirement image, the display image is detected for pattern defects according to multiple machine vision models, that is, the geometric structure, contour shape, texture distribution and pixel arrangement of the display image are compared and recognized by multiple machine vision models, and abnormal situations such as pattern misplacement, edge fracture, pixel loss, texture distortion or repeated structure deviation are recognized through structured feature extraction and pattern matching, and then the display pattern defect detection result is obtained. The display pattern defect detection result refers to the pattern anomaly record inconsistent with the expected structure detected, including the position coordinate, range size, pattern offset, and structure integrity score of the defect.

[0052] According to the display brightness defect detection result, the display color defect detection result and the display pattern defect detection result, feature fusion, spatial mapping and multidimensional statistics are performed to form a comprehensive analysis atlas intuitively presenting defect distribution characteristics and severity, and then a Mura defect detection atlas is generated. The Mura defect is a kind of visual non-uniformity defect with low contrast and uneven distribution in the display panel, which is manifested as dark spots, bright bands, color drift or pattern discontinuity, etc. By uniformly mapping the detection results of brightness, color and pattern, a spatial distribution model of panel overall defects, defect density gradient and evolution trend can be obtained for global quality control and long-term reliability evaluation.

[0053] Further, the Mura defect detection atlas is used to intuitively present the unevenness and abnormal distribution existing on the surface or inside of the display screen. Not only the position, range and type of the defect are marked, but also the severity of the defect can be reflected through color or gray level, thereby helping the system to quickly locate the potential problem area. According to the Mura defect detection atlas, a display warning signal is generated. The display warning signal is a kind of active feedback information for the running state of the display device, which is used to prompt that the device may have quality problems or performance degradation trend at present. That is, different warning levels can be divided according to different defect levels, for example, low-level warning is triggered when the Mura area ratio exceeds 1%, medium-level warning is triggered when it exceeds 3%, and high-level warning is triggered when it exceeds 5%. The grading mechanism can make the maintenance system intervene in advance when the defect is slight to prevent the problem from further deteriorating.

[0054] S5: According to the Mura defect detection atlas, the backlight source is associated and traced to obtain a defect backlight source tracing atlas.

[0055] Further, the present application also includes: based on the display image, synchronously collecting real-time monitoring parameters of the backlight source to obtain a backlight source monitoring sequence; according to the display brightness defect detection result, the backlight source monitoring sequence is associated and traced to obtain a brightness defect association and tracing result; according to the display color defect detection result, the backlight source monitoring sequence is associated and traced to obtain a color defect association and tracing result; according to the display pattern defect detection result, the backlight source monitoring sequence is associated and traced to obtain a pattern defect association and tracing result; the brightness defect association and tracing result, the color defect association and tracing result and the pattern defect association and tracing result are sorted to generate the defect backlight source tracing atlas.

[0056] Further, the application further comprises: performing multi-dimensional anomaly detection according to the backlight monitoring sequence, determining a brightness anomaly detection result, a color temperature anomaly detection result, a current-voltage anomaly detection result, and a temperature anomaly detection result; performing cause tracing on the display brightness defect detection result according to the brightness anomaly detection result, to obtain a first brightness defect tracing feature; performing cause tracing on the display brightness defect detection result according to the color temperature anomaly detection result, to obtain a second brightness defect tracing feature; performing cause tracing on the display brightness defect detection result according to the current-voltage anomaly detection result, to obtain a third brightness defect tracing feature; performing cause tracing on the display brightness defect detection result according to the temperature anomaly detection result, to obtain a fourth brightness defect tracing feature; and collating the first brightness defect tracing feature, the second brightness defect tracing feature, the third brightness defect tracing feature, and the fourth brightness defect tracing feature, to generate the brightness defect correlation traceability result.

[0057] Further, the application further comprises: generating a backlight abnormality early warning signal according to the brightness anomaly detection result, the color temperature anomaly detection result, the current-voltage anomaly detection result, and the temperature anomaly detection result.

[0058] Specifically, while analyzing the display image, real-time monitoring parameters of the backlight are synchronously collected, and key physical quantities of the working state of the backlight are dynamically acquired, including but not limited to brightness output intensity, color temperature distribution, current-voltage fluctuation, driving power, and working temperature, etc. The backlight monitoring sequence refers to a multi-dimensional monitoring data set collected within a certain time window, which is stored in a structured manner according to time sequence and parameter dimension, and is used for subsequent traceability analysis. Through synchronous collection, it can be ensured that the defect detection result and the working state of the backlight are one-to-one corresponding in time and space, so as to realize cause-effect correlation modeling.

[0059] Further, multi-dimensional anomaly detection is performed according to the backlight monitoring sequence, that is, multi-dimensional statistical analysis, feature modeling, and anomaly recognition are performed on various running data of the backlight system during the display process, to identify different types of abnormal behavior patterns. Among them, the brightness anomaly detection result, the color temperature anomaly detection result, the current-voltage anomaly detection result, and the temperature anomaly detection result. The brightness anomaly detection result refers to an abnormal state identified in the brightness output dimension, such as brightness fluctuation exceeding the design tolerance range or local brightness attenuation being obvious; the color temperature anomaly detection result refers to the drift or imbalance of the light source spectrum center or color temperature distribution; the current-voltage anomaly detection result refers to the phenomenon of current impact, voltage instability, or power fluctuation in the driving circuit; and the temperature anomaly detection result refers to the temperature rise of the internal or key components of the backlight module exceeding the safety threshold or the occurrence of thermal imbalance distribution.

[0060] Secondly, according to the brightness anomaly detection result, the display brightness defect detection result is traced back to the cause, that is, the display brightness defect (such as dark spot, brightness unevenness or local attenuation) is analyzed causally based on the brightness anomaly data as input. The first traceability feature of the brightness defect refers to the causal feature information extracted based on the brightness anomaly, such as the coupling degree of brightness attenuation and driving power drop, the relevance between brightness fluctuation and PWM dimming accuracy, etc.

[0061] Then, according to the color temperature anomaly detection result, the display brightness defect detection result is traced back to the cause, which means analyzing the indirect influence of color temperature anomaly on the formation of brightness defect based on color temperature shift or spectral drift data. The second traceability feature of the brightness defect refers to the brightness performance deviation feature caused by the color temperature anomaly, such as the visual brightness reduction caused by blue light peak shift, or the regional brightness difference caused by uneven spectral energy distribution, etc.

[0062] Next, according to the current-voltage anomaly detection result, the display brightness defect detection result is traced back to the cause, which means modeling the dynamic relationship between electrical parameter fluctuation and brightness anomaly, and analyzing the action mechanism of driving circuit performance on brightness defect. The third traceability feature of the brightness defect refers to the brightness response characteristic parameters caused by electrical fluctuation, such as the synchronicity of current transient impact and brightness peak flicker, or the time correlation of voltage drop and brightness sustained attenuation, etc.

[0063] Further, according to the temperature anomaly detection result, the display brightness defect detection result is traced back to the cause, which means identifying the influence of temperature rise, heat diffusion or thermal stress on brightness stability by analyzing the coupling relationship between thermal parameters and brightness defect. The fourth traceability feature of the brightness defect refers to the brightness change characteristics caused by temperature anomaly, such as the light efficiency attenuation caused by high temperature area, the local light output deviation of backlight module caused by thermal stress, etc.

[0064] Finally, the four types of traceability features of the first traceability feature of the brightness defect, the second traceability feature of the brightness defect, the third traceability feature of the brightness defect and the fourth traceability feature of the brightness defect are sorted and fused to generate the brightness defect correlation traceability result, that is, the causal features of different physical dimensions are feature mapped, weight modeled and comprehensively analyzed to form a complete traceability data structure to describe the multi-source formation mechanism and coupling path of the brightness defect, which not only reveals the direct cause of the brightness anomaly, but also reflects the combined influence of color temperature, electrical and thermal effects on the brightness performance.

[0065] Further, the correlation tracing of the backlight monitoring sequence according to the color defect detection result refers to the multi-dimensional correlation analysis of the backlight spectrum distribution, LED segmented driving power, color temperature compensation state and other monitoring parameters with reference to the color deviation, saturation abnormality or color temperature drift and other detection information, so as to identify the specific operation factors causing the color unevenness. The color defect correlation tracing result refers to the color deviation source characteristic data obtained through causal inference, for example, it is found that the output power attenuation of a certain wavelength segment is highly consistent with the blue shift region, so the LED array aging or color temperature adjustment failure can be located as the main reason.

[0066] Then, the correlation tracing of the backlight monitoring sequence according to the display pattern defect detection result refers to the comparison and modeling of the detected pattern breakage, bright band shift or stripe discontinuity and other abnormal phenomena with the spatial light distribution uniformity, local driving channel fluctuation, scattering layer thermal expansion parameter and other monitoring data of the backlight, so as to explore the operation root cause of the structural defect. The pattern defect correlation tracing result is the spatial characteristic mapping data generated after the tracing modeling, which is used to reveal the formation path of the defect in the backlight system, for example, it can be determined that the unstable driving current of a certain region directly leads to the shift of a specific stripe pattern.

[0067] Finally, the three types of correlation tracing results of the brightness defect correlation tracing result, the color defect correlation tracing result and the pattern defect correlation tracing result are sorted and fused to generate a defect backlight tracing map, which refers to mapping the correlation tracing results of different dimensions to the same backlight structure model, so as to present the full-link causal relationship of the defect formation in a visualized and modeled manner. The defect backlight tracing map not only describes the type, position and severity of the defect, but also reveals the interaction and coupling influence between different parameters, providing a decision basis for subsequent backlight control strategy adjustment.

[0068] Further, the brightness abnormality detection result refers to the detection conclusion of the brightness not meeting the preset standard or expected range obtained by monitoring and analyzing the brightness intensity of the backlight; the color temperature abnormality detection result is the abnormal state judgment obtained by analyzing the light source color temperature parameter, if the color temperature deviates from the standard range, it means that the backlight system may have light emitting material aging or driving control abnormality; the current voltage abnormality detection result refers to the result of whether the current and voltage signals of the power supply system deviate from the normal working interval obtained by collecting and analyzing the current and voltage signals, which may mean power module failure or driving circuit damage; the temperature abnormality detection result refers to the abnormal situation found by monitoring the working temperature of the backlight module through the temperature sensor, and the temperature that is too high may cause light source thermal decay and color deviation, and the temperature that is too low may affect the starting performance of the device.

[0069] According to the brightness anomaly detection result, the color temperature anomaly detection result, the current-voltage anomaly detection result and the temperature anomaly detection result, comprehensive analysis and fusion processing are performed to generate a backlight abnormality early warning signal, which is used to early inform the system or the operation and maintenance personnel that the backlight may have a failure, performance degradation or failure risk. For example, the backlight abnormality early warning signal can trigger an automatic adjustment mechanism, such as reducing the driving current or adjusting the heat dissipation strategy, and can also be used for manual intervention decision, such as scheduling maintenance or replacing components.

[0070] S6: Optimizing and adjusting the backlight control strategy according to the defect backlight traceability map.

[0071] Specifically, the defect backlight traceability map refers to a structured mapping result formed by correlatively analyzing the backlight monitoring sequence collected during the display process and the multi-dimensional defect detection results such as brightness, color and pattern. The defect backlight traceability map not only includes information such as the occurrence time, spatial position and severity of the defect, but also contains the key backlight parameter change trajectory leading to the defect, which can intuitively present the causal relationship and evolution path between different operating states of the backlight and display defects. For example, current fluctuation, temperature drift, color temperature shift, etc., thereby providing a quantifiable reference basis for subsequent strategy adjustment.

[0072] The backlight control strategy refers to a series of parameter settings and control logic for adjusting the operating state of the display device backlight system, including current output curve, driving frequency, PWM modulation mode, temperature compensation mechanism and color temperature correction model, etc., which directly determines the light output stability, spectral characteristics and response speed of the backlight, and has a decisive influence on the brightness uniformity, color reproduction and image consistency of the display panel. For example, when the map shows that the temperature anomaly and the brightness defect are highly correlated, the control strategy can suppress the brightness fluctuation caused by temperature drift by adjusting the heat dissipation control threshold or the dynamic compensation curve; when the current fluctuation is significantly associated with color shift, the current stability can be improved by optimizing the current driving algorithm or adding a feedback control link.

[0073] In summary, the machine vision detection method for backlight Mura defects provided in the present application has the following technical effects: by achieving the technical target of multi-level intelligent detection and adaptive control based on the correlation analysis of the backlight operating state and display defects, the technical effects of real-time identification and tracing of the root cause of Mura defects, dynamic optimization of backlight control strategy, improvement of display uniformity and color consistency, and improvement of production yield and long-term operation reliability of the equipment are achieved.

[0074] Embodiment two, based on the same inventive concept as the machine vision detection method for backlight Mura defects in the foregoing embodiments, the present application also provides a machine vision detection platform for backlight Mura defects, please refer to the accompanying drawings Figure 2The method comprises the following steps: receiving a display task of a liquid crystal display, wherein the liquid crystal display comprises a backlight source and a liquid crystal panel, and the display task comprises display requirement information; performing control parameter mining on the backlight source according to display environment real-time data of the liquid crystal display and the display requirement information, and determining a backlight source control strategy; controlling the liquid crystal display to execute the display task according to the backlight source control strategy, and obtaining a display image; performing multi-level Mura defect detection on the display image according to a plurality of machine vision models, and obtaining a Mura defect detection atlas; performing correlation tracing on the backlight source according to the Mura defect detection atlas, and obtaining a defective backlight source tracing atlas; and optimizing and adjusting the backlight source control strategy according to the defective backlight source tracing atlas.

[0075] Further, the machine vision detection platform for backlight source Mura defects is also used for: interconnecting backlight sources of the same type to obtain a backlight source interconnected group; constructing a display task feature matrix according to the display environment real-time data and the display requirement information; performing control parameter historical retrieval on the backlight source interconnected group according to the display task feature matrix, and obtaining a backlight source control history set; performing display quality fitting analysis according to the backlight source control history set, and obtaining a display quality fitting sequence; and performing optimization identification on the backlight source control history set according to the display quality fitting sequence, and generating the backlight source control strategy.

[0076] Further, the machine vision detection platform for backlight source Mura defects is also used for: extracting a first control history scheme according to the backlight source control history set; performing display quality evaluation retrieval according to the first control history scheme, and obtaining a first historical display quality evaluation sequence; performing concentration value calculation according to the first historical display quality evaluation sequence, and obtaining a first display quality fitting coefficient; adding the first display quality fitting coefficient to the display quality fitting sequence, and continuously performing display quality fitting analysis based on the backlight source control history set, and updating the display quality fitting sequence.

[0077] Further, the machine vision detection platform for backlight Mura defects is further used for: image fitting according to the display requirement information to determine a display requirement image; performing brightness defect detection on the display image according to the plurality of machine vision models based on the display requirement image to obtain a display brightness defect detection result; performing color defect detection on the display image according to the plurality of machine vision models based on the display requirement image to obtain a display color defect detection result; performing pattern defect detection on the display image according to the plurality of machine vision models based on the display requirement image to obtain a display pattern defect detection result; and generating the Mura defect detection atlas according to the display brightness defect detection result, the display color defect detection result and the display pattern defect detection result.

[0078] Further, the machine vision detection platform for backlight Mura defects is further used for: respectively performing brightness feature extraction on the display requirement image and the display image to obtain requirement brightness feature point cloud and display brightness feature point cloud; performing point position alignment processing on the requirement brightness feature point cloud and the display brightness feature point cloud to obtain brightness feature comparison point cloud; respectively supervising training of display brightness defect detection record set according to the plurality of machine vision models to obtain a plurality of brightness defect detectors; performing integrated fusion training according to the plurality of brightness defect detectors to obtain a brightness defect detection channel; and inputting the brightness feature comparison point cloud into the brightness defect detection channel to generate the display brightness defect detection result.

[0079] Further, the machine vision detection platform for backlight Mura defects is further used for: based on the display image, synchronously collecting real-time monitoring parameters of the backlight to obtain a backlight monitoring sequence; according to the display brightness defect detection result, associating and tracing the backlight monitoring sequence to obtain a brightness defect association and tracing result; according to the display color defect detection result, associating and tracing the backlight monitoring sequence to obtain a color defect association and tracing result; according to the display pattern defect detection result, associating and tracing the backlight monitoring sequence to obtain a pattern defect association and tracing result; and collating the brightness defect association and tracing result, the color defect association and tracing result and the pattern defect association and tracing result to generate a defect backlight tracing atlas.

[0080] Further, the machine vision detection platform for backlight Mura defects is further used for: performing multi-dimensional anomaly detection according to the backlight monitoring sequence, determining a brightness anomaly detection result, a color temperature anomaly detection result, a current voltage anomaly detection result and a temperature anomaly detection result; performing cause tracing on the display brightness defect detection result according to the brightness anomaly detection result, to obtain a first brightness defect tracing feature; performing cause tracing on the display brightness defect detection result according to the color temperature anomaly detection result, to obtain a second brightness defect tracing feature; performing cause tracing on the display brightness defect detection result according to the current voltage anomaly detection result, to obtain a third brightness defect tracing feature; performing cause tracing on the display brightness defect detection result according to the temperature anomaly detection result, to obtain a fourth brightness defect tracing feature; and collating the first brightness defect tracing feature, the second brightness defect tracing feature, the third brightness defect tracing feature and the fourth brightness defect tracing feature, to generate a brightness defect correlation trace result.

[0081] Further, the machine vision detection platform for backlight Mura defects is further used for: generating a backlight anomaly early warning signal according to the brightness anomaly detection result, the color temperature anomaly detection result, the current voltage anomaly detection result and the temperature anomaly detection result.

[0082] Further, the machine vision detection platform for backlight Mura defects is further used for: generating a display early warning signal according to the Mura defect detection atlas.

[0083] The various embodiments in the specification are described in a progressive manner, and each embodiment focuses on the difference from other embodiments. The machine vision detection platform for backlight Mura defects in the foregoing embodiment one is also applicable to the machine vision detection platform for backlight Mura defects in the present embodiment. Through the foregoing detailed description of the machine vision detection method for backlight Mura defects, those skilled in the art can clearly know the machine vision detection platform for backlight Mura defects in the present embodiment. Therefore, for the sake of brevity of the specification, the machine vision detection platform for backlight Mura defects in the present embodiment will not be described in detail.

[0084] The above description of disclosed embodiments enables a person skilled in the art to implement or use the present application. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to these embodiments shown herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.

[0085] Obviously, many modifications and variations of the present application are possible in light of the above teachings. It is, therefore, to be understood that within the scope of the application and its equivalents, the application can be practiced otherwise than as specifically described.

Claims

1. A machine vision inspection method for backlight Mura defects, characterized in that, The method comprises: receiving a display task of a liquid crystal display, the liquid crystal display comprising a backlight and a liquid crystal panel, the display task comprising display requirement information; mining control parameters of the backlight according to display environment real-time data of the liquid crystal display and the display requirement information, and determining a backlight control strategy; controlling the liquid crystal display to execute the display task according to the backlight control strategy, and obtaining a display image; performing multi-level Mura defect detection on the display image according to a plurality of machine vision models, and obtaining a Mura defect detection atlas; performing correlation tracing of the backlight according to the Mura defect detection atlas, and obtaining a defect backlight tracing atlas; optimizing and adjusting the backlight control strategy according to the defect backlight tracing atlas; wherein the mining of the control parameters of the backlight according to the display environment real-time data of the liquid crystal display and the display requirement information, and the determination of the backlight control strategy, comprise: interconnecting the backlights of the same type, and obtaining a backlight interconnected group; constructing a display task feature matrix according to the display environment real-time data and the display requirement information; performing control parameter historical retrieval on the backlight interconnected group according to the display task feature matrix, and obtaining a backlight control history set; performing display quality fitting analysis according to the backlight control history set, and obtaining a display quality fitting sequence; performing optimization identification on the backlight control history set according to the display quality fitting sequence, and generating the backlight control strategy; wherein the multi-level Mura defect detection on the display image according to the plurality of machine vision models, and the obtaining of the Mura defect detection atlas, comprise: performing image fitting according to the display requirement information, and determining a display requirement image; performing brightness defect detection on the display image according to the plurality of machine vision models based on the display requirement image, and obtaining a display brightness defect detection result; performing color defect detection on the display image according to the plurality of machine vision models based on the display requirement image, and obtaining a display color defect detection result; performing pattern defect detection on the display image according to the plurality of machine vision models based on the display requirement image, and obtaining a display pattern defect detection result; generating the Mura defect detection atlas according to the display brightness defect detection result, the display color defect detection result and the display pattern defect detection result; wherein the brightness defect detection on the display image according to the plurality of machine vision models based on the display requirement image, and the obtaining of the display brightness defect detection result, comprise: performing brightness feature extraction on the display requirement image and the display image respectively, and obtaining requirement brightness feature point cloud and display brightness feature point cloud; performing point position alignment processing on the requirement brightness feature point cloud and the display brightness feature point cloud, and obtaining brightness feature comparison point cloud; performing supervised training on display brightness defect detection record sets according to the plurality of machine vision models respectively, and obtaining a plurality of brightness defect detectors; Integrative fusion training is performed according to the plurality of luminance defect detectors to obtain a luminance defect detection channel; The luminance feature comparison point cloud is input into the luminance defect detection channel to generate the display luminance defect detection result.

2. The machine vision inspection method for backlight Mura defects as claimed in claim 1, wherein, According to the backlight control history set, display quality fitting analysis is performed to obtain a display quality fitting sequence, including: According to the backlight control history set, a first control history scheme is extracted; According to the first control history scheme, display quality evaluation retrieval is performed to obtain a first historical display quality evaluation sequence; According to the first historical display quality evaluation sequence, a central value is calculated to obtain a first display quality fitting coefficient; The first display quality fitting coefficient is added to the display quality fitting sequence, and display quality fitting analysis is continued based on the backlight control history set to update the display quality fitting sequence. 3.The machine vision inspection method for backlight Mura defect of claim 1, wherein, According to the Mura defect detection map, the backlight is associated and traced to obtain a defect backlight tracing map, including: Based on the display image, real-time monitoring parameters of the backlight are synchronously collected to obtain a backlight monitoring sequence; According to the display luminance defect detection result, the backlight monitoring sequence is associated and traced to obtain a luminance defect association and tracing result; According to the display color defect detection result, the backlight monitoring sequence is associated and traced to obtain a color defect association and tracing result; According to the display pattern defect detection result, the backlight monitoring sequence is associated and traced to obtain a pattern defect association and tracing result; The luminance defect association and tracing result, the color defect association and tracing result, and the pattern defect association and tracing result are sorted to generate the defect backlight tracing map.

4. The machine vision inspection method for backlight Mura defects as claimed in claim 3, wherein, According to the display luminance defect detection result, the backlight monitoring sequence is associated and traced to obtain a luminance defect association and tracing result, including: According to the backlight monitoring sequence, multi-dimensional anomaly detection is performed to determine a luminance anomaly detection result, a color temperature anomaly detection result, a current and voltage anomaly detection result, and a temperature anomaly detection result; According to the luminance anomaly detection result, the display luminance defect detection result is traced to obtain a luminance defect first tracing feature; According to the color temperature anomaly detection result, the display luminance defect detection result is traced to obtain a luminance defect second tracing feature; According to the current and voltage anomaly detection result, the display luminance defect detection result is traced to obtain a luminance defect third tracing feature; According to the temperature anomaly detection result, the display luminance defect detection result is traced to obtain a luminance defect fourth tracing feature; The luminance defect first tracing feature, the luminance defect second tracing feature, the luminance defect third tracing feature, and the luminance defect fourth tracing feature are sorted to generate the luminance defect association and tracing result.

5. The machine vision inspection method for backlight Mura defects as claimed in claim 4, wherein, According to the luminance anomaly detection result, the color temperature anomaly detection result, the current and voltage anomaly detection result, and the temperature anomaly detection result, a backlight anomaly early warning signal is generated. 6.The machine vision inspection method for backlight Mura defect of claim 1, wherein, According to the Mura defect detection map, a display early warning signal is generated.

7. A machine vision inspection platform for backlight Mura defects, characterized in that, Steps for implementing the machine vision detection method for backlight Mura defects according to any one of claims 1 to 6, comprising: a display task receiving module configured to receive a display task of a liquid crystal display, the liquid crystal display comprising a backlight and a liquid crystal panel, the display task comprising display requirement information; a backlight control strategy determining module configured to mine control parameters of the backlight according to real-time data of a display environment of the liquid crystal display and the display requirement information, and determine a backlight control strategy; a display image obtaining module configured to control the liquid crystal display to execute the display task according to the backlight control strategy, and obtain a display image; a Mura defect detection map obtaining module configured to perform multi-level Mura defect detection on the display image according to a plurality of machine vision models, and obtain a Mura defect detection map; a defective backlight traceability map obtaining module configured to associate and trace the backlight according to the Mura defect detection map, and obtain a defective backlight traceability map; an optimization adjusting module configured to optimize and adjust the backlight control strategy according to the defective backlight traceability map.

Images