LCD Screen Bad Pixel Detection Method, Detection Device and Storage Medium

Abstract

The present invention relates to the technical field of liquid crystal display screen detection, in particular to a method, device and storage medium for detecting dead pixels of a liquid crystal display screen. The test equipment is controlled based on a preset test scheme to test a target liquid crystal display screen, and a screen characteristic image of the target liquid crystal display screen is obtained. An actual gradient value matrix of the target liquid crystal display screen is constructed according to the screen characteristic image; the position of dead pixels of the target liquid crystal display screen is analyzed according to the actual gradient value matrix and the standard gradient value matrix of the target liquid crystal display screen, and a dead pixel position area in the screen characteristic image is obtained; the screen characteristic image is segmented to obtain a sub-characteristic image corresponding to the dead pixel position area in the target liquid crystal display screen; based on the dead pixel feature knowledge tree diagram general network, feature analysis is performed on each sub-characteristic image to obtain the dead pixel names and dead pixel incentives of all dead pixel position areas in the target liquid crystal display screen. Through this method, comprehensive and accurate dead pixel detection of the target liquid crystal display screen can be realized, and the detection efficiency and accuracy are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of liquid crystal display screen detection, and particularly to a method and device for detecting dead pixels of a liquid crystal display screen and a storage medium. Background Art

[0002] With the popularization of smart phones and the continuous improvement of users' requirements for screen display quality, the problem of dead pixels in mobile phone liquid crystal display screens has attracted increasing attention. In recent years, in order to improve the efficiency and accuracy of detecting dead pixels in mobile phone liquid crystal display screens, dead pixel detection methods based on image processing technology have gradually emerged. By using a camera to capture an image of the liquid crystal display screen and identifying the positions of dead pixels through image processing algorithms, automatic dead pixel detection is achieved. However, the existing detection methods based on image processing technology still have certain limitations: First, due to the complex structure and optical characteristics of the liquid crystal display screen, the image acquisition and processing process are easily affected by factors such as uneven illumination, reflected light, and noise, resulting in misjudgment of the detection results. Second, some subtle dead pixels, such as color deviation and light leakage, are difficult to accurately identify through existing image processing algorithms and require more complex algorithms and more precise equipment for detection. Finally, the existing methods have low linkage and low integration with the production line and cannot perform intelligent feedback adjustment on the production line according to the detection results, resulting in a high defective product rate. Therefore, overcoming these technical obstacles and developing a more accurate, intelligent, and automated dead pixel detection method is an important direction for the development of mobile phone liquid crystal display screen quality detection technology. Summary of the Invention

[0003] The present invention overcomes the limitations of the existing detection methods and provides a method and device for detecting dead pixels of a liquid crystal display screen and a storage medium.

[0004] To achieve the above object, the present invention discloses a method for detecting dead pixels of a liquid crystal display screen, including the following steps:

[0005] Obtain a preset test scheme for the target liquid crystal display screen, determine the standard gradient value corresponding to each pixel point in the target liquid crystal display screen when testing the target liquid crystal display screen according to the preset test scheme, and generate a standard gradient value matrix of the target liquid crystal display screen according to the standard gradient value corresponding to each pixel point;

[0006] Control a test device to test the target liquid crystal display screen based on the preset test scheme, obtain a screen characteristic image of the target liquid crystal display screen, and construct an actual gradient value matrix of the target liquid crystal display screen according to the screen characteristic image;

[0007] Analyze the position of dead pixels on the target liquid crystal screen based on the actual gradient value matrix and the standard gradient value matrix of the target liquid crystal screen, and obtain one or more dead pixel position areas in the screen characteristic image; perform segmentation processing on the screen characteristic image based on the watershed algorithm to obtain sub-characteristic images corresponding to the dead pixel position areas in the target liquid crystal screen;

[0008] Based on the dead pixel feature knowledge tree diagram general network, perform feature analysis on each sub-characteristic image to obtain the dead pixel names and dead pixel inducements of all dead pixel position areas in the target liquid crystal screen, and upload the dead pixel names and dead pixel inducements of all dead pixel position areas in the target liquid crystal screen to the detection database.

[0009] Preferably, construct the actual gradient value matrix of the target liquid crystal screen according to the screen characteristic image, specifically:

[0010] Perform grayscale processing on the screen characteristic image, and import the grayscale processed screen characteristic image into the OpenCV library;

[0011] Perform convolution operations on each pixel point in the screen characteristic image based on the cv2.Laplacian function in the OpenCV library; use the convolution result as the gradient value of each pixel point, where the magnitude of the gradient value represents the degree of change of the pixel point, and the direction represents the direction of change;

[0012] Construct a blank matrix, store the gradient values of each pixel point in the blank matrix, and obtain the preliminary gradient value matrix of the target liquid crystal screen;

[0013] Perform Fourier transform processing on the preliminary gradient value matrix to obtain the frequency spectrum diagram of the preliminary gradient value matrix; among them, the horizontal axis of the frequency spectrum diagram represents frequency, and the vertical axis represents the energy corresponding to the frequency;

[0014] Obtain the frequency intensity values of the elements in the preliminary gradient value matrix according to the frequency spectrum diagram of the preliminary gradient value matrix; compare the frequency intensity values of the elements in the preliminary gradient value matrix with the preset frequency intensity value threshold;

[0015] Mark the position area of the elements corresponding to the frequency intensity values greater than the preset frequency intensity value threshold in the preliminary gradient value matrix, and define it as the high-frequency position area;

[0016] Perform filtering processing on the high-frequency position area of the preliminary gradient value matrix based on the median filtering algorithm to generate the actual gradient value matrix of the target liquid crystal screen.

[0017] Preferably, analyze the position of dead pixels on the target liquid crystal screen based on the actual gradient value matrix and the standard gradient value matrix of the target liquid crystal screen, and obtain one or more dead pixel position areas in the screen characteristic image, specifically:

[0018] Calculate the absolute difference between the two elements at each same matrix position of the actual gradient value matrix and the standard gradient value matrix in the target liquid crystal screen;

[0019] Construct an absolute difference matrix based on the absolute difference between the two elements at each same matrix position of the actual gradient value matrix and the standard gradient value matrix;

[0020] Obtain the process precision requirement information of the target liquid crystal screen, and preset the limit difference threshold according to the process precision requirement information; compare each absolute difference in the absolute difference matrix with the limit difference threshold one by one;

[0021] If each absolute difference in the absolute difference matrix is not greater than the limit difference threshold, it indicates that there are no bad points in the target liquid crystal screen, indicating that the target liquid crystal screen is a qualified product, and upload the detection result that the target liquid crystal screen is a qualified product to the detection database;

[0022] If there is one or more absolute differences in the absolute difference matrix that are greater than the limit difference threshold, it indicates that there are bad points in the target liquid crystal screen, indicating that the target liquid crystal screen is an unqualified product, and upload the detection result that the target liquid crystal screen is an unqualified product to the detection database;

[0023] Moreover, if there is one or more absolute differences in the absolute difference matrix that are greater than the limit difference threshold, mark the element position nodes in the actual gradient value matrix where the absolute difference is greater than the limit difference threshold, define them as the bad point positions, and mark the marked bad point positions on the screen characteristic image;

[0024] Obtain all the independent closed regions in the screen characteristic image that are marked as bad point positions, and obtain one or more bad point position regions in the screen characteristic image.

[0025] Preferably, based on the bad point feature knowledge tree graph general network, perform feature analysis on each sub-characteristic image to obtain the bad point names and bad point causes of all bad point position regions in the target liquid crystal screen, specifically:

[0026] Obtain the bad point feature images corresponding to various types of bad points in the liquid crystal screen through the big data network, and obtain the bad point causes of various types of bad points;

[0027] Construct a knowledge tree graph general network, and if each type of bad point in the liquid crystal screen is a cutting node, cut out several branch nodes in the knowledge tree graph general network according to each cutting node, and assign corresponding bad point name labels to each branch node according to the corresponding type of bad point;

[0028] Map the bad point feature images and bad point causes of various types of bad points in the liquid crystal screen into the corresponding branch nodes respectively to obtain the bad point feature knowledge tree graph general network;

[0029] Obtain the sub-feature image corresponding to the defective pixel position area in the target liquid crystal screen, and import the sub-feature image into the defective pixel feature knowledge tree diagram general network;

[0030] Calculate the image similarity between the sub-feature image and the defective pixel feature images to which each branch node in the defective pixel feature knowledge tree diagram general network belongs based on the perceptual hashing algorithm;

[0031] Mark the branch node with the largest image similarity among the calculated image similarities, and extract the corresponding defective pixel name and defective pixel cause in the branch node with the largest image similarity to obtain the defective pixel name and defective pixel cause of the corresponding defective pixel position area in the target liquid crystal screen;

[0032] Repeat the above steps to obtain the defective pixel names and defective pixel causes of all defective pixel position areas in the target liquid crystal screen.

[0033] It further includes the following steps:

[0034] Obtain the detection data of each liquid crystal screen within a preset time period in the detection database; wherein, the detection data includes the detection result situation of each liquid crystal screen, and the defective pixel names and defective pixel causes of the liquid crystal screens with unqualified detection results;

[0035] Statistically obtain the defective product rate of the liquid crystal screens in the current production batch according to the detection data of each liquid crystal screen within the preset time period, and judge whether the defective product rate is greater than the preset defective product rate threshold;

[0036] If the defective product rate is greater than the preset defective product rate threshold, calculate the required supplementary production quantity of the liquid crystal screens in the current production batch according to the defective product rate, generate a supplementary warning message according to the required supplementary production quantity, and send the supplementary warning message to a preset terminal;

[0037] Meanwhile, obtain the production performance parameter information of the production line for producing the liquid crystal screens in the current production batch, and obtain the order information of the liquid crystal screens in the current production batch, and determine the delivery time node of the liquid crystal screens in the current production batch according to the order information;

[0038] Obtain the production lines that are idle in the production workshop before the delivery time node, and obtain the production performance parameter information of the idle production lines;

[0039] Analyze the attention scores between the production performance parameter information of the production line for producing the liquid crystal screens in the current production batch and the production performance parameter information of the idle production lines based on the local sensitive attention mechanism;

[0040] Only extract the production lines in the idle state with attention scores greater than the preset score threshold, and define them as available supplementary production lines; obtain the operation log information of each available supplementary production line, and obtain the failure probability of each available supplementary production line according to the operation log information.

[0041] Obtain the available supplementary production line with the minimum failure probability value as the recommended supplementary production line, and send the recommended supplementary production line to the preset terminal.

[0042] It also includes the following steps:

[0043] If the defective rate is greater than the preset defective rate threshold, obtain the bad point causes of the liquid crystal displays with unqualified detection results within the preset time period.

[0044] Statistically analyze the bad point causes of the liquid crystal displays with unqualified detection results within the preset time period to obtain the occurrence frequency values of various bad point causes.

[0045] Mark the bad point causes with occurrence frequency values greater than the preset frequency threshold as high-occurrence bad point causes, and mark the bad point causes with occurrence frequency values not greater than the preset frequency threshold as low-occurrence bad point causes.

[0046] Obtain the process characteristic information of each process station in the production line, combine the process characteristic information of each process station, and introduce a regression analysis algorithm to analyze the influence degree of each process station in the production line on the high-occurrence bad point causes.

[0047] Based on the Internet of Things method, obtain various real-time process parameters of the process stations with influence degrees greater than the preset influence degree threshold, and obtain various preset process parameters of the process stations with influence degrees greater than the preset influence degree threshold.

[0048] Calculate the process parameter differences between various real-time process parameters and preset process parameters of the process stations with influence degrees greater than the preset influence degree threshold.

[0049] Mark the real-time process parameters corresponding to the process parameter differences greater than the preset difference, and perform correction processing on the marked real-time process parameters based on the corresponding process parameter differences.

[0050] In addition, to achieve the above object, the present invention also discloses a liquid crystal display bad point detection device, which includes a memory and a processor. The memory stores a liquid crystal display bad point detection method program. When the liquid crystal display bad point detection method program is executed by the processor, the steps of any one of the liquid crystal display bad point detection methods are implemented.

[0051] In addition, to achieve the above object, the present invention also discloses a computer-readable storage medium, the readable storage medium includes a liquid crystal screen dead pixel detection method program, and when the liquid crystal screen dead pixel detection method program is executed by a processor, the steps of any one of the liquid crystal screen dead pixel detection methods are implemented.

[0052] The present invention solves the technical defects existing in the background art, and the present invention has the following beneficial effects: controlling a test device to test a target liquid crystal screen based on a preset test scheme, obtaining a screen characteristic image of the target liquid crystal screen, and constructing an actual gradient value matrix of the target liquid crystal screen according to the screen characteristic image; analyzing the dead pixel positions of the target liquid crystal screen based on the actual gradient value matrix and the standard gradient value matrix of the target liquid crystal screen to obtain one or more dead pixel position regions in the screen characteristic image; performing segmentation processing on the screen characteristic image based on the watershed algorithm to obtain a sub-characteristic image corresponding to the dead pixel position region in the target liquid crystal screen; performing feature analysis on each sub-characteristic image based on the dead pixel feature knowledge tree diagram generalization network to obtain the dead pixel names and dead pixel inducements of all dead pixel position regions in the target liquid crystal screen. Through this method, comprehensive and accurate dead pixel detection of the target liquid crystal screen can be realized, and the detection efficiency and accuracy are improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the following drawings are only some embodiments of the present invention, and for those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings.

[0054] Figure 1 It is a first method flow chart of a liquid crystal screen dead pixel detection method;

[0055] Figure 2 It is a second method flow chart of a liquid crystal screen dead pixel detection method;

[0056] Figure 3 It is a third method flow chart of a liquid crystal screen dead pixel detection method. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0057] In order to be able to more clearly understand the above objects, features, and advantages of the present invention, the following will further describe the present invention in detail with reference to the drawings and specific embodiments. It should be noted that, without conflict, the embodiments of the present application and the features in the embodiments can be combined with each other.

[0058] In the following description, many specific details are set forth in order to provide a thorough understanding of the present invention. However, the present invention may be practiced in other ways than those specifically described herein, and thus, the scope of protection of the present invention is not limited by the specific embodiments disclosed below.

[0059] As Figure 1 shown, the present invention discloses a method for detecting dead pixels of a liquid crystal display screen, comprising the following steps:

[0060] S102: Obtain a preset test scheme for the target liquid crystal display screen, determine the standard gradient values corresponding to each pixel point in the target liquid crystal display screen when testing the target liquid crystal display screen according to the preset test scheme, and generate a standard gradient value matrix of the target liquid crystal display screen;

[0061] S104: Control a test device to test the target liquid crystal display screen based on the preset test scheme, obtain a screen characteristic image of the target liquid crystal display screen, and construct an actual gradient value matrix of the target liquid crystal display screen according to the screen characteristic image;

[0062] S106: Analyze the dead pixel positions of the target liquid crystal display screen according to the actual gradient value matrix and the standard gradient value matrix of the target liquid crystal display screen to obtain one or more dead pixel position regions in the screen characteristic image; perform segmentation processing on the screen characteristic image based on the watershed algorithm to obtain sub-characteristic images corresponding to the dead pixel position regions in the target liquid crystal display screen;

[0063] S108: Perform feature analysis on each sub-characteristic image based on the dead pixel feature knowledge tree diagram general network to obtain the dead pixel names and dead pixel causes of all dead pixel position regions in the target liquid crystal display screen, and upload the dead pixel names and dead pixel causes of all dead pixel position regions in the target liquid crystal display screen to a detection database.

[0064] It should be noted that first, obtain the preset test scheme of the target liquid crystal display screen, and determine the standard gradient values of each pixel point according to this scheme to generate a standard gradient value matrix. Then, control the test device to test the target liquid crystal display screen, obtain a screen characteristic image, and construct an actual gradient value matrix. Next, analyze the dead pixel positions of the target liquid crystal display screen according to the actual gradient value matrix and the standard gradient value matrix to obtain dead pixel position regions. Then, use the watershed algorithm to perform segmentation processing on the screen characteristic image to obtain sub-characteristic images. Finally, perform feature analysis on each sub-characteristic image based on the dead pixel feature knowledge tree diagram general network to obtain dead pixel names and dead pixel causes, and upload them to the detection database. Through this method, comprehensive and accurate dead pixel detection of the target liquid crystal display screen can be achieved, improving the detection efficiency and accuracy. Performing feature analysis on each sub-characteristic image based on the dead pixel feature knowledge tree diagram general network can identify different types of dead pixels and analyze their causes, providing strong support for the quality evaluation and improvement of the liquid crystal display screen.

[0065] Preferably, an actual gradient value matrix of the target liquid crystal screen is constructed according to the screen characteristic image, specifically as follows:

[0066] The screen characteristic image is grayscale processed, and the grayscale processed screen characteristic image is imported into the OpenCV library;

[0067] Based on the cv2.Laplacian function in the OpenCV library, convolution operations are performed on each pixel point in the screen characteristic image; the convolution result is used as the gradient value of each pixel point, where the magnitude of the gradient value represents the degree of change of the pixel point, and the direction represents the direction of change;

[0068] A blank matrix is constructed, and the gradient values of each pixel point are stored in the blank matrix to obtain a preliminary gradient value matrix of the target liquid crystal screen;

[0069] The preliminary gradient value matrix is subjected to Fourier transform processing to obtain a spectrogram of the preliminary gradient value matrix; among them, the horizontal axis of the spectrogram represents frequency, and the vertical axis represents the energy corresponding to the frequency;

[0070] The frequency intensity values of the elements in the preliminary gradient value matrix are obtained according to the spectrogram of the preliminary gradient value matrix; the frequency intensity values of the elements in the preliminary gradient value matrix are compared with a preset frequency intensity value threshold;

[0071] The position regions corresponding to the elements in the preliminary gradient value matrix with frequency intensity values greater than the preset frequency intensity value threshold are marked and defined as high-frequency position regions;

[0072] Based on the median filtering algorithm, the high-frequency position region of the preliminary gradient value matrix is filtered to generate an actual gradient value matrix of the target liquid crystal screen.

[0073] It should be noted that first, the screen characteristic image is grayscaled to simplify the image data for subsequent processing. Then, the cv2.Laplacian function in the OpenCV library is used to perform a convolution operation on the image to calculate the gradient value of each pixel point. The gradient value reflects the degree and direction of pixel change around the pixel point, which helps to identify the edges and details in the image. A blank matrix is constructed to store the gradient values of all pixel points, forming a preliminary gradient value matrix. Through Fourier transform, the preliminary gradient value matrix is converted into a spectrogram, with frequency and energy as the coordinate axes, intuitively showing the distribution of different frequency components in the image. According to the analysis of the spectrogram, the high-frequency position regions can be identified, which contain the edges and important details of the image. Next, the elements in the preliminary gradient value matrix with frequency intensity values greater than the preset threshold are marked as high-frequency position regions. This is based on the understanding of the image characteristics, and the high-frequency regions are the noise points in the matrix. To further purify these high-frequency position regions and reduce the noise impact, a median filtering algorithm is used for processing. Median filtering is a non-linear filtering method, especially suitable for removing isolated noise points in the image, reducing interference while maintaining edges and details. Finally, after the above processing steps, the actual gradient value matrix of the target liquid crystal screen is obtained. This matrix contains the detailed characteristic information of the liquid crystal screen, including not only the positioning of edges and details but also the suppression of noise, providing basic data for subsequent analysis and quality control. Through this method, the characteristics of the liquid crystal screen can be analyzed efficiently and accurately, which is of great significance for improving product quality and performance.

[0074] Preferably, as Figure 2 shown, the bad point position analysis of the target liquid crystal screen is carried out according to the actual gradient value matrix and the standard gradient value matrix of the target liquid crystal screen, and one or more bad point position regions in the screen characteristic image are obtained. Specifically:

[0075] S202: Calculate the absolute difference between the two elements at each same matrix position in the actual gradient value matrix and the standard gradient value matrix of the target liquid crystal screen;

[0076] S204: Construct an absolute difference matrix according to the absolute difference between the two elements at each same matrix position in the actual gradient value matrix and the standard gradient value matrix;

[0077] S206: Obtain the process precision requirement information of the target liquid crystal screen, and preset a limit difference threshold according to the process precision requirement information; compare each absolute difference in the absolute difference matrix with the limit difference threshold one by one;

[0078] S208: If all the absolute differences in the absolute difference matrix are not greater than the limit difference threshold, it indicates that there are no dead pixels in the target LCD screen, which means the target LCD screen is a qualified product, and the detection result of the target LCD screen being a qualified product is uploaded to the detection database;

[0079] S210: If there is one or more absolute differences in the absolute difference matrix that are greater than the limit difference threshold, it indicates that there are dead pixels in the target LCD screen, which means the target LCD screen is a non - qualified product, and the detection result of the target LCD screen being a non - qualified product is uploaded to the detection database;

[0080] S212: Moreover, if there is one or more absolute differences in the absolute difference matrix that are greater than the limit difference threshold, mark the element position nodes in the actual gradient value matrix whose absolute differences are greater than the limit difference threshold, define them as dead pixel positions, and mark the marked dead pixel positions on the screen characteristic image;

[0081] S214: Obtain all the independent closed regions in the screen characteristic image that are marked as dead pixel positions, and get one or more dead pixel position regions in the screen characteristic image.

[0082] It should be noted that by comparing the actual gradient value matrix of the target liquid crystal display (LCD) screen with the standard gradient value matrix, it is determined whether there are quality problems with the LCD screen. Specifically, first, the absolute differences between the actual gradient value matrix and the standard gradient value matrix at corresponding positions are calculated, and then these differences are organized into a new matrix, that is, the absolute difference matrix. This step aims to quantify the difference between the two and provide a data basis for subsequent quality evaluation. Secondly, according to the process precision requirements of the target LCD screen, a limit difference threshold is preset in advance, and this threshold is used to judge whether the LCD screen meets the quality standard. Next, each difference in the absolute difference matrix is compared with the limit difference threshold. If all differences are less than or equal to the threshold, it indicates that the LCD screen meets the quality requirements and is judged as a qualified product, and its test results are recorded in the database. On the contrary, if there is one or more differences exceeding the threshold, it indicates that there are quality problems with the LCD screen, and it is judged as an unqualified product, and the test results are also recorded in the database. For those cases where the differences exceed the threshold, the specific element positions in the actual gradient value matrix are marked as "bad point positions". At the same time, these positions are also marked on the screen characteristic image to visually display the problem areas on the LCD screen. Finally, all independent and enclosed areas are extracted from the screen characteristic image marked with bad point positions to form one or more bad point position areas. These areas represent the specific positions of the quality problems existing on the LCD screen, providing a clear reference for subsequent further analysis or improvement work. This method realizes the efficient and accurate detection of the quality of the LCD screen through mathematical calculations and image processing means. It can not only quickly identify unqualified products but also accurately locate the problems, improving the reliability of product quality inspection and control.

[0083] Preferably, as Figure 3 shown, based on the bad point feature knowledge tree diagram concept network, feature analysis is performed on each sub-feature image to obtain the bad point names and bad point causes of all bad point position areas in the target LCD screen, specifically:

[0084] S302: Obtain the bad point feature images corresponding to various types of bad points in the LCD screen through the big data network, and obtain the bad point causes of various types of bad points;

[0085] Among them, the types of bad points on the LCD screen mainly include the following several types, and each type has its specific appearance and potential causes, that is, "bad point causes":

[0086] Monochromatic bad points: These bad points usually appear as one or more points of a fixed color (usually black, white or other specific colors) on the screen. They may be caused by the open circuit or short circuit of a certain pixel's circuit;

[0087] Light leakage and dead pixels: Light leakage and dead pixels usually appear as light leakage at a certain place on the screen under a dark background, forming visible bright spots or stripes. This may be caused by poor sealing of the backlight module or the screen panel, or a malfunction in the control circuit of the LED backlight source;

[0088] Rainbow effect: The rainbow effect is caused by uneven color mixing between pixels and usually appears as color bands or color patches when the screen displays transitional colors. This is mainly due to uneven arrangement of liquid crystal molecules;

[0089] Color deviation dead pixels: Color deviation dead pixels are manifested as incorrect color output of a certain pixel under a specific color. For example, a pixel that should display red actually shows other colors. This may be caused by a malfunction in the red, green, and blue color filters of the pixel or related circuits;

[0090] Vignetting: Vignetting usually appears at the edges of the screen and is manifested as lower brightness than the central area. This may be caused by uneven backlighting, optical design problems, or defects in the screen panel itself;

[0091] S304: Construct a knowledge tree diagram general network, and if each type of dead pixel in the liquid crystal screen is a cutting node, cut out several branch nodes in the knowledge tree diagram general network according to each cutting node, and assign corresponding dead pixel name labels to each branch node according to the corresponding type of dead pixel;

[0092] S306: Map the dead pixel characteristic images and dead pixel inducements of various types of dead pixels in the liquid crystal screen into the corresponding branch nodes respectively to obtain a dead pixel characteristic knowledge tree diagram general network;

[0093] S308: Obtain the sub-feature image corresponding to the dead pixel position area in the target liquid crystal screen, and import the sub-feature image into the dead pixel characteristic knowledge tree diagram general network;

[0094] S310: Calculate the image similarity between the sub-feature image and the dead pixel characteristic images to which each branch node in the dead pixel characteristic knowledge tree diagram general network belongs based on the perceptual hashing algorithm;

[0095] S312: Mark the branch node with the largest image similarity among the calculated image similarities, and extract the corresponding dead pixel name and dead pixel inducement in the branch node with the largest image similarity to obtain the dead pixel name and dead pixel inducement of the corresponding dead pixel position area in the target liquid crystal screen;

[0096] S314: Repeat the above steps to obtain the dead pixel names and dead pixel inducements of all dead pixel position areas in the target liquid crystal screen.

[0097] It should be noted that by collecting and analyzing a large amount of bad pixel data of liquid crystal displays, including the characteristic images of various types of bad pixels and their inducing factors, a bad pixel characteristic knowledge tree graph general network containing rich knowledge has been constructed. This network structure allows for the division of branch nodes with specific attributes by cutting nodes (representing different types of bad pixels), and corresponding bad pixel name labels are assigned to each branch node. In practical applications, when diagnosing bad pixels of a specific liquid crystal display, the system will obtain the sub-characteristic image of a bad pixel position area in the liquid crystal display and import it into the constructed bad pixel characteristic knowledge tree graph general network. By using the perceptual hashing algorithm, the system can calculate the similarity between the sub-characteristic image and the bad pixel characteristic images represented by each branch node in the knowledge tree graph general network. The branch node with the highest similarity usually best matches the bad pixel type represented by the sub-characteristic image. Therefore, the system can extract the corresponding bad pixel name and inducing factor from this branch node, thereby identifying the bad pixel type and possible causes in the liquid crystal display. Through big data and machine learning methods, the system can quickly and accurately identify and classify various bad pixels on the liquid crystal display, which can effectively improve the detection efficiency, improve the reliability of detection results, and is of great significance for improving product quality. In addition, this system can also continuously improve its diagnostic ability through continuous learning and updating to adapt to different types of bad pixels and changing market demands.

[0098] It also includes the following steps:

[0099] Obtain the detection data of each liquid crystal display in the detection database within a preset time period; wherein, the detection data includes the detection result situation of each liquid crystal display, as well as the bad pixel name and bad pixel inducing factor of the liquid crystal display with a non-conforming product detection result;

[0100] According to the detection data of each liquid crystal display within a preset time period, statistically obtain the defective product rate of the liquid crystal displays in the current production batch, and judge whether the defective product rate is greater than a preset defective product rate threshold;

[0101] If the defective product rate is greater than the preset defective product rate threshold, calculate the required supplementary production quantity of the liquid crystal displays in the current production batch according to the defective product rate, generate a supplementary warning message according to the required supplementary production quantity, and send the supplementary warning message to a preset terminal;

[0102] Meanwhile, obtain the production performance parameter information of the production line for producing the liquid crystal displays in the current production batch, and obtain the order information of the liquid crystal displays in the current production batch, and determine the delivery time node of the liquid crystal displays in the current production batch according to the order information;

[0103] Among them, the production performance parameter information usually includes but is not limited to production rate, equipment utilization rate, downtime, energy consumption, raw material consumption, etc.;

[0104] Obtain the production lines that are idle before the delivery time node in the production workshop, and obtain the production performance parameter information of the idle production lines;

[0105] Analyze the attention scores between the production performance parameter information of the production line for producing the current production batch of liquid crystal displays and the production performance parameter information of the idle production lines based on the local sensitive attention mechanism;

[0106] Only extract the idle production lines with attention scores greater than the preset score threshold, and define them as available supplementary production lines; obtain the operation log information of each available supplementary production line, and obtain the failure probability of each available supplementary production line according to the operation log information;

[0107] Obtain the available supplementary production line with the minimum failure probability value as the recommended supplementary production line, and send the recommended supplementary production line to a preset terminal.

[0108] It should be noted that, first of all, the system obtains the detection data of each liquid crystal display within a preset time period in the detection database, including the detection results and the detailed information of the defective products. By analyzing these data, the system can calculate the defective product rate of the current production batch and compare it with the preset threshold. If the defective product rate exceeds the standard, the system will calculate the required quantity of supplementary production and generate a supplementary warning message to notify the relevant departments in a timely manner to ensure the adjustment of the production plan and the effective allocation of resources. Secondly, the system obtains the order information and delivery time node of the current production batch, and combines the production performance parameters of the production line to predict the possible production capacity gap before delivery. Through the local sensitive attention mechanism, the system analyzes the performance matching degree between the current production line and the idle production lines, and screens out the idle production lines that match the performance of the current production line and have a lower failure probability as the recommended supplementary production lines. This step helps to quickly respond to production needs, improve production efficiency and product quality. The entire system realizes the refined management of the liquid crystal display production process through real-time data processing, predictive analysis and decision support, effectively improves production efficiency, reduces the defective product rate, optimizes resource allocation at the same time, enhances production flexibility and response speed. Through intelligent prediction and optimization strategies, the system can effectively respond to market changes, ensure the smooth execution of the production plan, and improve the competitiveness of the enterprise.

[0109] It also includes the following steps:

[0110] If the defective product rate is greater than the preset defective product rate threshold, obtain the bad point inducement reasons of the liquid crystal displays with unqualified detection results within the preset time period;

[0111] Conduct statistical analysis on the bad point inducement reasons of the liquid crystal displays with unqualified detection results within the preset time period to obtain the occurrence frequency values of various bad point inducement reasons;

[0112] Mark the bad point causes with frequency values greater than the preset frequency threshold as high - incidence bad point causes, and mark the bad point causes with frequency values not greater than the preset frequency threshold as low - incidence bad point causes;

[0113] Obtain the process characteristic information of each process station in the production line, combine the process characteristic information of each process station, and introduce a regression analysis algorithm to analyze the influence degree of each process station in the production line on the high - incidence bad point causes;

[0114] Among them, the process characteristic information of each process station in the production line refers to the unique attributes and characteristics of each process station during the production process. These information describe the role of each station in the entire production process, the operation mode, the equipment and tools used, the raw materials or parts required, the process parameters, the quality control requirements, and the detailed connection relationship with other stations;

[0115] Extract data related to high - incidence bad point causes from the database, including but not limited to process parameters, equipment operation status, operation records, quality inspection results, etc.; Select a suitable regression analysis model, such as linear regression, logistic regression, multiple linear regression, etc., according to the actual situation. The construction of the model needs to consider independent variables (the process characteristic information of each process station) and dependent variables (the occurrence frequency of high - incidence bad point causes). Use historical data to train the model, and by adjusting the model parameters, make the model better fit the historical data. And verify the accuracy and reliability of the model through methods such as cross - validation and residual analysis. According to the results of the regression analysis, evaluate the influence degree of the process characteristic information of each process station on high - incidence bad point causes. This can be obtained through coefficient estimation, significance test, etc. The magnitude and sign of the coefficient can reflect the influence direction and intensity of the variable on the result; Interpret the analysis results as the contribution degree of each process station in the production line to high - incidence bad point causes, providing a decision - making basis for the optimization of the production process, such as adjusting process parameters, improving equipment, optimizing operation processes, etc., to reduce the defective product rate, improve production efficiency and product quality;

[0116] Based on the Internet of Things method, obtain various real - time process parameters of the process stations with influence degrees greater than the preset influence degree threshold, and obtain various preset process parameters of the process stations with influence degrees greater than the preset influence degree threshold;

[0117] Calculate the process parameter differences between various real - time process parameters and preset process parameters of the process stations with influence degrees greater than the preset influence degree threshold;

[0118] Mark the real - time process parameters corresponding to the process parameter differences greater than the preset difference, and perform correction processing on the marked real - time process parameters based on the corresponding process parameter differences.

[0119] It should be noted that when the defective product rate exceeds the preset threshold, the system will obtain the causes of the defective points of the unqualified products, conduct statistical analysis, and find out the high-incidence defective point causes with higher occurrence frequencies and the low-incidence defective point causes with lower occurrence frequencies. Then, the system will combine the process characteristic information of each process station in the production line and use the regression analysis algorithm to analyze the influence degree of each process station on the high-incidence defective point causes. The system obtains the real-time process parameters and the preset process parameters of the process stations with greater influence degree and calculates the difference between the two. Finally, the system will mark the real-time process parameters with larger process parameter differences and correct them according to the corresponding differences. Through this method, the main reasons for defective products can be quickly and accurately found, and by real-time monitoring and correcting the process parameters, the production process can be adjusted in time, the defective product rate can be reduced, and the product quality and production efficiency can be improved.

[0120] In the actual application process of the present invention, the following steps may further be included:

[0121] Obtain the historical working data of each process station in each production line when it is in an abnormal state during the historical working process; and obtain the historical working data of each process station in each production line before it is in an abnormal state during the historical working process;

[0122] Based on the fuzzy algorithm, conduct fuzzy analysis on the historical working data of each process station when it is in an abnormal state during the historical working process to obtain the fuzzy characteristics of each process station when it is in an abnormal state; and based on the fuzzy algorithm, conduct fuzzy analysis on the historical working data of each process station in each production line before it is in an abnormal state during the historical working process to obtain the fuzzy characteristics of each process station before it is in an abnormal state;

[0123] Introduce the Monte Carlo model, import the fuzzy characteristics of each process station when it is in an abnormal state and the fuzzy characteristics before it is in an abnormal state into the Monte Carlo model for probabilistic state random deduction, and obtain the transition probabilities of each process station from before being in an abnormal state to each abnormal state;

[0124] Based on the deep learning network, construct an operating state prediction model, construct a probabilistic state transition matrix according to the transition probabilities of each process station from before being in an abnormal state to each abnormal state, import the probabilistic state transition matrix into the operating state prediction model for backpropagation learning and training until the model meets the preset requirements, and output the trained operating state prediction model;

[0125] If the defective product rate is greater than the preset defective product rate threshold, obtain various real-time process parameters of the process stations with an influence degree greater than the preset influence degree threshold, and import the various real-time process parameters of the process stations with an influence degree greater than the preset influence degree threshold into the trained operation state prediction model for prediction, and output the state transition probability of the process stations with an influence degree greater than the preset influence degree threshold;

[0126] When the state transition probability of the process stations with an influence degree greater than the preset influence degree threshold is greater than the preset probability value, control the process stations with an influence degree greater than the preset influence degree threshold to stop production, and generate a fault warning message.

[0127] It should be noted that by analyzing historical data, a probability model of the operation state of the process stations is established, and real-time data is used for prediction to warn of potential fault risks in advance, thereby effectively reducing the downtime rate and defective product rate during the production process. First, the system collects the historical working data of each process station in normal and abnormal states, and uses a fuzzy algorithm to analyze these data to extract the characteristics of each station in different states. Then, the system imports these characteristic data into the Monte Carlo model to simulate the probability transfer process of the station from the normal state to different abnormal states, and obtains the state transition probabilities of each state. Next, a deep learning network is used to construct an operation state prediction model, and the probability state transition matrix calculated by the Monte Carlo model is used as the training data of the model for backpropagation learning, and finally a model that can predict the state change of the process stations is obtained. When the defective product rate exceeds the preset threshold, the system will obtain the real-time process parameters of the process stations with a greater influence degree and input them into the trained operation state prediction model to predict the probability of the station having a fault. If the prediction result shows that the fault probability exceeds the preset threshold, the system will control the station to stop production and issue a fault warning message to remind the staff to check and repair. This method combines a fuzzy algorithm, a Monte Carlo model and a deep learning network to achieve accurate prediction of the operation state of the process stations, can warn of potential faults in advance, effectively reduce the defective product rate during the production process, and improve production efficiency and product quality.

[0128] In the actual application process of the present invention, the following steps may also be included:

[0129] Obtain the negative comment data of the liquid crystal display screens feedback by users, and construct a heat map of the negative comments of the liquid crystal display screens according to the negative comment data of the liquid crystal display screens;

[0130] Obtain the detection data of each liquid crystal display screen in a preset time period from the detection database, and construct a heat map of the detection results of the liquid crystal display screens according to the detection data;

[0131] Perform registration processing on the heat map of negative reviews of the liquid crystal screen and the heat map of liquid crystal screen detection results based on the ICP algorithm. After the registration is completed, calculate the cosine similarity between each sub-region in the heat map of negative reviews of the liquid crystal screen and the heat map of liquid crystal screen detection results based on the cosine similarity algorithm;

[0132] Only extract the sub-regions corresponding to the cosine similarity greater than the cosine similarity, which are defined as the overlapping regions of negative reviews, and obtain the process stations that are functionally related to the overlapping regions of negative reviews during the production of the liquid crystal screen, which are defined as the key monitoring process stations;

[0133] Obtain the real-time process parameters of the key monitoring process stations, import the real-time process parameters of the key monitoring process stations into the trained operation status prediction model for prediction, and output the state transition probability of the key monitoring process stations;

[0134] When the state transition probability of the key monitoring process stations is greater than the preset probability value, control the corresponding key monitoring process stations to stop production and generate a fault warning message.

[0135] It should be noted that by collecting and analyzing the negative evaluations (i.e., "bad reviews") of the liquid crystal display (LCD) screens from users and the detection data of the LCD screens, potential problem areas in the production process can be identified and located. First, the bad review data of the LCD screens are collected from user feedback, and then these data are used to construct a heat map of the bad reviews of the LCD screens. This step helps to visually show which areas or parts of the LCD screens are more frequently negatively evaluated, thus indicating potential quality problems. Next, the detection data of each LCD screen within a preset time period are obtained, and a heat map of the detection results of the LCD screens is constructed. This step aims to compare the actual detection results with user feedback to further verify and locate the problem areas. Subsequently, the heat map of the bad reviews of the LCD screens and the heat map of the detection results are registered through the ICP algorithm to ensure their spatial correspondence. After that, the cosine similarity algorithm is applied to calculate the similarity between each sub-region in the two heat maps, so as to evaluate the consistency between the bad reviews and the detection results. The system will screen out the sub-regions with high cosine similarity, define them as "bad review coincidence regions", and identify the process workstations that are functionally relevant to these regions as "key monitoring process workstations". These workstations may be the key links prone to problems in the production process. Next, the system will obtain the real-time process parameters of these key monitoring process workstations and input these parameters into a pre-trained operation state prediction model. The model predicts the state transition probability of the workstation, that is, the probability of potential problems in the future. Finally, if the state transition probability of a certain key monitoring process workstation exceeds the preset threshold, the system will automatically control the workstation to stop production and generate a fault warning message to take timely measures to avoid potential quality problems. Generally speaking, this process realizes the real-time monitoring and problem location of the production process by integrating user feedback, detection data, and machine learning technology, which helps to improve product quality, reduce production costs, and enhance the overall production efficiency. An intelligent system for realizing the quality control of LCD screen production by combining user feedback with production process data. By performing correlation analysis on user feedback and production process data, potential quality problems can be identified in advance and timely measures can be taken for prevention, thus effectively reducing the defective product rate, improving product quality, and enhancing user satisfaction.

[0136] In addition, to achieve the above object, the present invention also discloses a liquid crystal display screen dead pixel detection device, which includes a memory and a processor. A liquid crystal display screen dead pixel detection method program is stored in the memory. When the liquid crystal display screen dead pixel detection method program is executed by the processor, the steps of any one of the liquid crystal display screen dead pixel detection methods are realized.

[0137] In addition, to achieve the above object, the present invention also discloses a computer-readable storage medium, which includes a liquid crystal display screen dead pixel detection method program. When the liquid crystal display screen dead pixel detection method program is executed by a processor, the steps of any one of the liquid crystal display screen dead pixel detection methods are realized.

[0138] The above are only specific embodiments of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention can easily think of changes or substitutions, which should all be covered within the protection scope of the present invention.

Claims

1. A method for detecting bad pixels of a liquid crystal display, characterized in that: include: Determine the standard gradient value corresponding to each pixel point in the target LCD screen when testing the target LCD screen according to the preset test scheme, and generate a standard gradient value matrix of the target LCD screen; Controlling the test equipment to test the target LCD screen based on a preset test plan, obtaining a screen characteristic image of the target LCD screen, and constructing an actual gradient value matrix of the target LCD screen; According to the actual gradient value matrix and the standard gradient value matrix of the target LCD screen, the bad pixel position analysis of the target LCD screen is performed to obtain one or more bad pixel position areas in the screen characteristic image, specifically: Calculate the absolute difference between two elements in each identical matrix position of the actual gradient value matrix and the standard gradient value matrix in the target LCD screen; Construct an absolute difference matrix based on the absolute difference between two elements in each same matrix position of the actual gradient value matrix and the standard gradient value matrix; Obtaining the process accuracy requirement information of the target LCD screen, and presetting the limit difference threshold value according to the process accuracy requirement information; comparing each absolute difference in the absolute difference matrix with the limit difference threshold value; If there is one or more absolute difference values ​​in the absolute difference matrix that are greater than the limit difference threshold, the element position node whose absolute difference value is greater than the limit difference threshold is marked in the actual gradient value matrix, which is defined as the bad pixel position, and the marked bad pixel position is marked in the screen characteristic image; Acquire all independent closed areas marked as bad pixel locations in the screen characteristic image to obtain the bad pixel location area in the screen characteristic image; The screen characteristic image is segmented based on the watershed algorithm to obtain a sub-characteristic image corresponding to the bad pixel location area in the target LCD screen; Based on the bad pixel feature knowledge tree graph network, feature analysis is performed on each sub-characteristic image to obtain the bad pixel names and causes of all bad pixel location areas in the target LCD screen and upload them to the detection database.

2. A method for detecting bad pixels of a liquid crystal display according to claim 1, characterized in that: Construct the actual gradient value matrix of the target LCD screen, specifically: The screen characteristic image is grayed out, and the grayed out screen characteristic image is imported into the OpenCV library; Based on the cv2.Laplacian function in the OpenCV library, a convolution operation is performed on each pixel in the screen characteristic image; the convolution result is used as the gradient value of each pixel, the magnitude of the gradient value indicates the degree of change of the pixel, and the direction indicates the direction of change; Construct a blank matrix, store the gradient value of each pixel in the blank matrix, and obtain a preliminary gradient value matrix of the target LCD screen; Performing Fourier transform processing on the preliminary gradient value matrix to obtain a frequency spectrum of the preliminary gradient value matrix; wherein the horizontal axis of the frequency spectrum represents the frequency, and the vertical axis represents the energy corresponding to the frequency; Obtain the frequency intensity value of each element in the preliminary gradient value matrix according to the spectrum diagram of the preliminary gradient value matrix; compare the frequency intensity value of each element in the preliminary gradient value matrix with a preset frequency intensity value threshold; Mark the position area of ​​the elements corresponding to the frequency intensity value greater than the preset frequency intensity value threshold in the preliminary gradient value matrix, and define it as the high-frequency position area; The high-frequency position area of ​​the preliminary gradient value matrix is ​​filtered based on the median filtering algorithm to generate the actual gradient value matrix of the target LCD screen.

3. A method for detecting bad pixels of a liquid crystal display according to claim 1, characterized in that: If all absolute differences in the absolute difference matrix are not greater than the limit difference threshold, it means that there are no bad pixels in the target LCD screen, indicating that the target LCD screen is a qualified product, and the test result that the target LCD screen is a qualified product is uploaded to the test database; If there is one or more absolute differences in the absolute difference matrix that are greater than the limit difference threshold, it means that there are bad pixels in the target LCD screen, indicating that the target LCD screen is a defective product, and the detection result that the target LCD screen is a defective product is uploaded to the detection database.

4. A method for detecting bad pixels of a liquid crystal display according to claim 1, characterized in that: Based on the bad pixel feature knowledge tree graph network, the feature analysis of each sub-feature image is performed to obtain the bad pixel names and bad pixel causes of all bad pixel location areas in the target LCD screen, specifically: Obtain the bad pixel feature images corresponding to various types of bad pixels in the LCD screen through the big data network, and obtain the bad pixel causes of various types of bad pixels; Construct a knowledge tree graph network, and treat each type of bad pixel in the LCD screen as a cutting node, cut out a number of branch nodes in the knowledge tree graph network according to each cutting node, and assign a corresponding bad pixel name label to each branch node according to the corresponding type of bad pixel; The bad pixel feature images and bad pixel inducements of various types of bad pixels in the LCD screen are mapped to the corresponding branch nodes to obtain the bad pixel feature knowledge tree graph network; Obtain a sub-characteristic image corresponding to the bad pixel location area in the target LCD screen, and import the sub-characteristic image into the bad pixel feature knowledge tree graph network; The image similarity between the sub-feature image and the bad pixel feature image to which each branch node in the bad pixel feature knowledge tree graph network belongs is calculated based on the perceptual hash algorithm; The calculated similarities of each image are used to mark the branch node with the greatest image similarity, and the corresponding bad pixel name and bad pixel inducement are extracted from the branch node with the greatest image similarity to obtain the bad pixel name and bad pixel inducement of the corresponding bad pixel position area in the target LCD screen; Repeat the above steps to obtain the bad pixel names and bad pixel causes of all bad pixel locations in the target LCD screen.

5. A method for detecting bad pixels of a liquid crystal display according to claim 1, characterized in that: The following steps are also included: Acquire the detection data of each LCD screen within a preset time period in the detection database; wherein the detection data includes the detection result of each LCD screen, and the name and cause of the bad pixel of the LCD screen with the detection result being unqualified; Obtain the defective rate of the current production batch of LCD screens according to the inspection data of each LCD screen within a preset time period, and determine whether the defective rate is greater than a preset defective rate threshold; If the defective rate is greater than the preset defective rate threshold, the required replenishment output of the current production batch of LCD screens is calculated according to the defective rate, and supplementary warning information is generated according to the required replenishment output, and the supplementary warning information is sent to the preset terminal; At the same time, obtain the production performance parameter information of the production line that produces the current production batch of LCD screens, and obtain the order information of the current production batch of LCD screens, and determine the delivery time node of the current production batch of LCD screens according to the order information; Obtain the production lines in the production workshop that are in an idle state before the delivery time node, and obtain the production performance parameter information of the production lines in the idle state; Analyze the attention score between the production performance parameter information of the production line producing the current production batch of LCD screens and the production performance parameter information of the production line in an idle state based on the local sensitive attention mechanism; Only the idle production lines whose attention scores are greater than the preset score threshold are extracted and defined as available supplementary production lines; the operation log information of each available supplementary production line is obtained, and the failure probability of each available supplementary production line is obtained according to the operation log information; An available supplementary production line with the smallest failure probability value is obtained as a recommended supplementary production line, and the recommended supplementary production line is sent to a preset terminal.

6. A method for detecting bad pixels of a liquid crystal display according to claim 5, characterized in that: The following steps are also included: If the defective product rate is greater than a preset defective product rate threshold, the bad pixel cause of the LCD screen whose detection result is defective within a preset time period is obtained; Performing statistical analysis on the causes of bad pixels of LCD screens that are unqualified products within a preset time period to obtain the occurrence frequency values ​​of various causes of bad pixels; Mark the bad pixel inducement whose occurrence frequency value is greater than the preset frequency threshold as a high-incidence bad pixel inducement, and mark the bad pixel inducement whose occurrence frequency value is not greater than the preset frequency threshold as a low-incidence bad pixel inducement; Obtain the process characteristic information of each process station in the production line, combine the process characteristic information of each process station and introduce regression analysis algorithm to analyze the influence of each process station in the production line on the cause of high incidence of bad pixels; Acquire various real-time process parameters of process stations whose influence is greater than a preset influence threshold based on the Internet of Things, and acquire various preset process parameters of process stations whose influence is greater than a preset influence threshold; Calculate the process parameter difference between various real-time process parameters of the process stations whose influence degree is greater than a preset influence degree threshold and the preset process parameters; The real-time process parameters corresponding to the process parameter differences that are greater than the preset differences are marked, and the marked real-time process parameters are corrected based on the corresponding process parameter differences.

7. A liquid crystal screen bad pixel detection device, characterized in that: The LCD screen bad pixel detection device includes a memory and a processor. The memory stores a LCD screen bad pixel detection method program. When the LCD screen bad pixel detection method program is executed by the processor, the steps of the LCD screen bad pixel detection method as claimed in any one of claims 1 to 6 are implemented.

8. A computer-readable storage medium, characterized in that: The readable storage medium includes a program for detecting a method for detecting a bad pixel of a liquid crystal screen. When the program for detecting a method for detecting a bad pixel of a liquid crystal screen is executed by a processor, the steps of the method for detecting a bad pixel of a liquid crystal screen as claimed in any one of claims 1 to 6 are implemented.

Images