LCD module alignment calibration system based on image analysis

Abstract

The invention relates to the technical field of LCD module calibration, and discloses an LCD module alignment calibration system based on image analysis. An image acquisition module of the system acquires multi-view image data to generate an image basic data set; the feature extraction module extracts feature information to obtain an image feature data set; the deviation analysis module analyzes to obtain a position deviation parameter set; the calibration threshold value setting module sets a threshold value to obtain a calibration displacement threshold value; the displacement prediction module deduces the trend to obtain a displacement distribution prediction value; and the calibration feedback control module adjusts the device according to the data to obtain an automatic calibration regulation and control scheme. According to the system, through cooperation of multiple modules, multi-view image analysis, dynamic threshold setting and real-time feedback control, the problems of insufficient precision, low efficiency, poor adaptability and the like in a traditional calibration mode are solved, the accuracy and stability of alignment calibration of the LCD module are improved, and the requirements of high-requirement production scenes are met.

Description

Technical Field

[0001] This invention relates to the field of LCD module calibration technology, specifically to an LCD module alignment calibration system based on image analysis. Background Technology

[0002] In today's rapidly developing display technology, LCD modules, as a core component of various display devices, directly impact the display quality and lifespan of the product due to the alignment accuracy during their production and assembly. As display screens evolve towards higher resolution, narrower bezels, and greater flexibility, more stringent requirements are being placed on the alignment calibration of LCD modules. Traditional LCD module alignment calibration often relies on manual operation or semi-automated equipment. These methods are frequently affected by factors such as differences in operator experience, visual fatigue, and changes in ambient lighting, making it difficult to consistently guarantee calibration accuracy. During manual calibration, operators must use tools such as microscopes to observe the module edges and positioning marks, adjusting the module position based on subjective judgment. This is not only inefficient but also difficult to accurately identify micron-level positional deviations. In large-scale mass production scenarios, the poor consistency of manual operations can easily lead to fluctuations in product yield and increase production costs. Furthermore, the long adjustment cycle of manual calibration cannot meet the high-speed requirements of modern production lines, severely hindering the improvement of production efficiency. Existing automated alignment and calibration technologies sometimes employ mechanical positioning combined with single-sensor feedback. While this reduces human intervention to some extent, it still has many limitations. For example, traditional mechanical positioning relies on a preset mechanical reference, which is prone to positioning deviations when the module experiences minor deformation or assembly stress. Furthermore, the data collected by a single sensor has limited dimensions, making it difficult to fully reflect the module's true position, leading to misjudgments or delayed adjustments during the calibration process. Image-based alignment calibration methods are increasingly being applied in LCD module production. However, existing technologies mostly employ single-view image acquisition, which makes it difficult to fully capture the module's three-dimensional position information and easily leads to incomplete feature extraction due to blind spots in the viewing angle. In the feature matching stage, traditional algorithms often rely on single feature points or simple contour comparisons. When there are stains, scratches, or reflections on the module surface, the accuracy of feature point recognition drops significantly, resulting in large errors in position deviation calculation. Existing systems often use fixed parameters for calibration threshold settings, failing to dynamically adjust based on the physical characteristics and assembly environment of different module models. This leads to over- or under-calibration during the calibration process. In the displacement prediction stage, the lack of effective analysis of displacement change trends makes it difficult to anticipate potential cumulative errors during calibration, causing calibration accuracy to gradually drift with increasing production batches. Furthermore, most calibration systems suffer from lagging feedback mechanisms, unable to adjust control strategies in real-time based on module position changes. This results in calibration efficiency and stability that cannot meet the production requirements of high-end LCD modules. These problems hinder the improvement of LCD module production yields, restricting the development of display devices towards higher performance and better quality. Summary of the Invention

[0003] The purpose of this invention is to provide an LCD module alignment calibration system based on image analysis to solve the problems mentioned in the background art.

[0004] To achieve the above objectives, the present invention provides an LCD module alignment calibration system based on image analysis, the system comprising: The image acquisition module acquires multi-view image data based on the optical signals of the alignment area of ​​the LCD module, integrates image resolution and contrast parameters, and generates a basic image dataset. Based on the image dataset, the feature extraction module extracts feature information of module edge contours and positioning marks, filters key feature points and feature regions, and obtains image feature dataset; Based on the image feature dataset, the deviation analysis module extracts the difference between the current module position and the standard position, analyzes the relationship between contour matching degree and marker point offset, matches the position deviation and feature correspondence, and obtains the position deviation parameter set. The calibration threshold setting module extracts the current calibration displacement change value based on the position deviation parameter set, combines it with real-time image feedback data, allocates position and displacement, sets a threshold, and applies the threshold to the adjustment of module alignment to obtain the calibration displacement threshold. The displacement prediction module captures image sampling point displacement data based on the calibration displacement threshold, combines image analysis to infer the trend of displacement change of sampling points, analyzes the displacement change trend corresponding to the calibration displacement, classifies and organizes the displacement change trend according to the inference results, and combines displacement change information to perform image analysis adjustment and filtering on the classified data to obtain displacement distribution prediction values. Based on the displacement distribution prediction value, the calibration feedback control module analyzes the position and displacement error value through real-time module position and displacement data, and adjusts the module alignment device in combination with the error value to obtain an automatic calibration and control scheme for LCD module alignment.

[0005] Preferably, the image base dataset includes a resolution parameter set, a contrast parameter set, and an image integration parameter set; the image feature dataset includes contour feature parameters, marker point coordinate parameters, and feature filtering parameters; the position deviation parameter set includes difference extraction parameters, matching degree analysis parameters, and deviation correspondence parameters; the calibration displacement threshold includes displacement change parameters, feedback matching parameters, and threshold application parameters; the displacement distribution prediction value includes trend analysis parameters and position displacement relationship parameters; and the LCD module alignment automatic calibration and control scheme includes error analysis parameters and calibration adjustment parameters.

[0006] Preferably, the image acquisition module includes: The data acquisition submodule acquires image data from multiple perspectives based on the optical signals of the alignment area of ​​the LCD module, locates blurred data, removes noise data, and arranges the acquired image data in the order of perspectives to generate a regional image dataset. The difference integration submodule analyzes the image resolution and contrast between viewpoints based on the regional image dataset, calculates the ratio of viewpoint parameter changes, sorts the difference values ​​between viewpoints by weight, marks viewpoints with excessive fluctuation differences, and obtains image acquisition difference data. The basic integration submodule, based on the image acquisition difference data, calls the perspective difference value to perform multi-dimensional summarization, screens the differences in image acquisition values, classifies them according to the size of the values, and arranges the image acquisition values ​​in an orderly manner to generate a basic image dataset.

[0007] Preferably, the feature extraction module includes: Based on the image dataset, the feature recognition submodule identifies the contour state of each module edge, records the curvature of the contour and the position of the marker point, normalizes the recorded data, sorts the normalized data by curvature and position, and generates a feature parameter dataset. The feature optimization submodule analyzes the curvature and position values ​​in the feature parameter dataset, filters feature combinations with high matching degree with the image base, records the matching results through pattern matching, adjusts the feature combinations, and generates feature group optimization results. The feature selection submodule retrieves the optimized result of the feature combination, determines the optimal curvature and position combination with matching degree, adjusts the feature extraction parameters, inputs the recognition configuration, verifies the stability of the parameter set, and generates an image feature dataset.

[0008] Preferably, the deviation analysis module includes: The deviation factor analysis submodule, based on the image feature dataset, collects key data through image monitoring, including grayscale values, contour continuity, and marker point clarity. It performs time series analysis on the data, removes outliers, partitions the remaining data, and obtains deviation factor analysis data. The parameter matching submodule analyzes the influence of image variables on the deviation between the module position and the standard position by analyzing the deviation factor analysis data, calculates the degree of influence of the change of each image factor on the deviation adjustment, determines the optimal deviation setting for matching based on the influence score, iteratively adjusts the parameters to capture the optimal combination, and obtains the deviation docking result. The deviation parameter integration submodule selects the position and standard position deviation combination that matches the current image conditions from the deviation docking results, conducts parameter adjustment experiments, optimizes the parameter settings through multiple adjustments and verifications, determines and solidifies the parameters as the operation standard, and generates a position deviation parameter set.

[0009] Preferably, the calibration threshold setting module includes: The displacement extraction submodule locates the displacement change monitoring point based on the position deviation parameter set, extracts the displacement change value in the monitoring area, continuously records the displacement increase and decrease rate, extracts multiple key change nodes corresponding to the change rate, sorts the node values ​​in order, and obtains the current displacement change feature value. The displacement matching submodule analyzes the node change values ​​and real-time image feedback data based on the current displacement change feature value, performs calibration according to a predetermined matching criterion, calls the matching criterion to perform distribution redistribution within the displacement interval, and obtains the position displacement matching structure. The calibration allocation submodule, based on the position displacement matching structure, uses an image threshold dynamic adjustment method to calculate the distribution of displacement among changing nodes, sets upper and lower limits for node thresholds, applies the thresholds to the module alignment values, and allocates them to obtain the calibration displacement threshold.

[0010] Preferably, the image threshold dynamic adjustment method includes obtaining the allocation calibration value of the matching displacement change node, and performing a comprehensive calculation using the real-time displacement value, the displacement matching value adjusted by the upstream node, the real-time position height calculated by the current node, the position matching value adjusted by the downstream node, the displacement weight coefficient, the position weight coefficient, the dynamic adjustment weight coefficient, and the threshold lower limit set by the node.

[0011] Preferably, the displacement prediction module includes: The displacement data capture submodule, based on the calibrated displacement threshold, applies image analysis algorithms to capture displacement data of sampling points, remove outliers and correct errors, stores the data in a hierarchical manner according to intervals, performs feature processing, and generates a feature displacement dataset. Based on the characteristic displacement dataset, the displacement load analysis submodule divides the intervals according to the calibration displacement, extracts the changing trend and fluctuation characteristics, and generates a calibration displacement and displacement change feature set. The displacement distribution inference submodule adjusts the feature parameters and calibrates the trend data based on the calibrated displacement and displacement change feature set, extracts the distribution interval, and performs numerical prediction to obtain the displacement distribution prediction value.

[0012] Preferably, the image analysis algorithm includes calculating displacement feature values ​​and generating a feature displacement dataset, wherein the displacement feature values ​​are calculated comprehensively based on the weight of each data point, the original displacement value of each sampling point, the position value of each sampling point, and the calibration coefficient of each sampling point, and the total number of sampling points is the total number of sampling points.

[0013] Preferably, the calibration feedback control module includes: The error analysis submodule extracts real-time module position and displacement data based on the displacement distribution prediction value, analyzes the real-time displacement value and the prediction value, and generates a position displacement error value by matching the displacement difference with the current position information. The parameter adjustment submodule sets the calibration adjustment parameters of the module alignment device based on the position displacement error value. It sets the adjustment range for areas with large errors and makes fine adjustments for areas with low errors. By comparing the calibration effect, it filters and integrates the matching parameter set to generate a calibration adjustment parameter set. Based on the set of calibration adjustment parameters, the calibration control submodule applies adjustment parameters to each alignment device position, performs calibration operations item by item, synchronously monitors the module position and displacement, gradually adjusts the calibration operation sequence of each area, and generates an automatic calibration and control scheme for LCD module alignment.

[0014] Compared with the prior art, the beneficial effects of the present invention are: The image analysis-based LCD module alignment calibration system provided by this invention effectively solves many problems existing in current LCD module alignment calibration technologies through the collaborative work of multiple modules. The system includes an image acquisition module that can acquire multi-view image data based on the optical signals of the LCD module alignment area and integrate image resolution and contrast parameters to generate a basic image dataset. Compared with traditional single-view image acquisition methods, this system can more comprehensively capture the module's positional features, reducing the loss of feature information due to limited viewing angles, thereby providing richer and more accurate raw data for subsequent feature extraction. The feature extraction module extracts feature information from the module's edge contours and positioning markers based on the image dataset, and filters key feature points and regions to obtain an image feature dataset. This process, through accurate identification and filtering of key features, avoids interference from irrelevant information, improves the effectiveness of feature information, and lays a reliable foundation for position deviation analysis. The deviation analysis module extracts the difference between the current module position and the standard position, analyzes the relationship between contour matching degree and marker point offset, and matches the position deviation with the feature correspondence to obtain a position deviation parameter set. This allows for more accurate quantification of the module's position deviation, reducing misjudgments caused by inaccurate feature matching compared to traditional deviation calculation methods. The calibration threshold setting module combines the position deviation parameter set and real-time image feedback data to allocate position and displacement, sets a threshold, and applies it to module alignment adjustment. The resulting calibration displacement threshold can be dynamically adjusted according to actual conditions, overcoming the shortcomings of traditional fixed thresholds that cannot adapt to different working conditions. This makes the threshold setting more closely match actual calibration needs, improving the flexibility and adaptability of calibration. The displacement prediction module infers the trend of displacement changes at sampling points based on the calibration displacement threshold, and categorizes, organizes, and filters the displacement change trends. The resulting displacement distribution prediction value can predict displacement change patterns in advance, helping to take countermeasures in advance during actual calibration and reducing calibration errors caused by the uncertainty of displacement changes. The calibration feedback control module analyzes the error value by using real-time module position and displacement data, and adjusts the module alignment device based on the error value. The resulting automatic calibration control scheme can achieve real-time closed-loop control, ensuring that the module position adjustment responds promptly to error changes, and avoiding the problems of low calibration efficiency and insufficient accuracy caused by traditional feedback lag. Attached Figure Description

[0015] Figure 1 This is a schematic diagram illustrating the working principle of the LCD module alignment and calibration system based on image analysis described in this invention. Figure 2 Flowcharts generated for the image base dataset and feature dataset; Figure 3 This is a schematic diagram illustrating the working principle of the image acquisition module. Figure 4 This is a flowchart of the feature extraction module. Detailed Implementation

[0016] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0017] Please see Figures 1-4 This invention provides an LCD module alignment calibration system based on image analysis, and the specific implementation steps are as follows: The image acquisition module acquires multi-view image data based on the optical signals of the alignment area of ​​the LCD module, integrates image resolution and contrast parameters, and generates a basic image dataset. Specifically, it captures optical signals of the alignment area of ​​the module under different lighting conditions using an optical sensor, and acquires images of the same area from multiple perspectives, including horizontal, vertical, and 45° tilt. During this process, the resolution and contrast of the images are monitored in real time, and the image data from different perspectives are integrated according to a preset format to form a basic image dataset containing complete regional features.

[0018] Based on the aforementioned image dataset, the feature extraction module extracts feature information from the module's edge contours and positioning markers, filters key feature points and feature regions, and obtains an image feature dataset. An edge detection algorithm is used to identify the contour curves of the module edges, recording parameters such as inflection points and tangent slopes. Simultaneously, positioning markers in the image (such as markers of specific shapes) are located, and their center coordinates and dimensions are extracted. A feature filtering algorithm is then used to remove redundant features, retaining feature points and regions that play a crucial role in positioning.

[0019] The deviation analysis module, based on the image feature dataset, extracts the difference between the current module position and the standard position, analyzes the relationship between contour matching degree and marker point offset, matches the position deviation with the feature correspondence, and obtains a position deviation parameter set. The extracted feature information is compared with pre-stored standard module feature information to calculate differences such as edge contour overlap and marker point coordinate offset, establishing a correlation model between deviation and feature change, thereby quantifying the position deviation.

[0020] The calibration threshold setting module extracts the current calibration displacement change value based on the position deviation parameter set, combines it with real-time image feedback data, allocates position and displacement, sets a threshold, and applies the threshold to the module alignment adjustment to obtain the calibration displacement threshold. The required displacement adjustment amount is determined based on the magnitude of the position deviation, the adjustment direction is determined by combining real-time acquired image data, and the maximum and minimum threshold ranges for displacement adjustment are set to ensure that the adjustment is within a reasonable range.

[0021] The displacement prediction module, based on the calibrated displacement threshold, captures displacement data of image sampling points. It then combines image analysis to infer the trend of displacement changes at the sampling points, analyzes the displacement change trend corresponding to the calibrated displacement, and categorizes the displacement change trends based on the inference results. Combining this with displacement change information, it performs image analysis adjustments and filtering on the categorized data to obtain predicted displacement distribution values. By continuously acquiring displacement data from multiple sampling points in the image, a trend analysis algorithm is used to predict the direction and magnitude of subsequent displacement changes. The prediction results are categorized according to the change trend, and then the categorized data is optimized by combining actual displacement information.

[0022] Based on the predicted displacement distribution, the calibration feedback control module analyzes the position and displacement error values ​​using real-time module position and displacement data. It then adjusts the module alignment device according to the error values, resulting in an automatic calibration and control scheme for LCD module alignment. By comparing the real-time acquired module position and displacement data with the predicted values, calculating the error magnitude, and driving the motors and other actuators of the alignment device to make corresponding adjustments based on the error values, a complete automatic calibration operation process is formed.

[0023] Example 1: The image dataset comprises a resolution parameter set, a contrast parameter set, and an image integration parameter set. The resolution parameter set records pixel dimension information of images from different viewpoints, covering the number of horizontal pixels, the number of vertical pixels, and the physical size of a single pixel. Images from different viewpoints exhibit different resolution characteristics due to differences in optical lens parameters; some viewpoints use higher pixel densities to capture details, while others use lower pixel densities to expand coverage. These parameters collectively constitute the core content of the resolution parameter set. The contrast parameter set involves information on brightness differences between bright and dark areas in the image, including the maximum brightness value, the minimum brightness value, and the gradient change between them. By sampling the brightness of different areas, a set of parameters reflecting the degree of contrast in the image is formed. Differences in lighting conditions under different viewpoints cause fluctuations in contrast parameters, and these fluctuations are also included in the contrast parameter set. The image integration parameter set is used to standardize the stitching process of multi-view images. It includes the overlapping area ratio of images from different viewpoints, the stitching order, and color consistency adjustment parameters. The overlapping area ratio is determined based on the viewpoint coverage, the stitching order is set according to the viewpoint orientation, and the color consistency adjustment parameters are used to eliminate color deviations caused by different sensors. Through the synergistic effect of these parameters, seamless integration of multi-view images is achieved.

[0024] The image feature dataset consists of contour feature parameters, marker coordinate parameters, and feature filtering parameters. Contour feature parameters describe the geometric characteristics of the module's edge contours, including the length, curvature, and turning angle of contour segments. These geometric quantities are extracted and formed into a continuous parameter sequence through point-by-point analysis of the edge contours. Contour features differ at different edge locations, and the parameter sequence accurately reflects these differences. Marker coordinate parameters record the specific position of the positioning markers in the image coordinate system, including x-axis coordinates, y-axis coordinates, and z-axis coordinates in some scenarios. The extraction method of coordinate parameters is adjusted according to the shape and size of the positioning markers to ensure that the coordinate values ​​accurately correspond to the actual position of the markers. Feature filtering parameters are used to filter key content from a large amount of feature information, including the saliency score of feature points and the area ratio of feature regions. The saliency score is determined based on the influence of the feature on the positioning, while the area ratio reflects the size of the feature region in the overall image. By setting appropriate filtering conditions, feature information that is practically meaningful for calibration is retained.

[0025] The positional deviation parameter set encompasses difference extraction parameters, matching degree analysis parameters, and deviation correspondence parameters. Difference extraction parameters define the calculation method and range of positional differences, including the calculation starting point, measurement step size, and effective extraction area. The calculation starting point is typically set as the module's reference point. The measurement step size is determined based on accuracy requirements. The effective extraction area covers key areas for module alignment, ensuring that all potentially deviating locations are included in the analysis. Matching degree analysis parameters assess the degree of matching between the current feature and the standard feature, including the overlap ratio of the contour and the allowable offset range of the marker point. The overlap ratio is calculated by comparing the overlapping portion of the current contour and the standard contour. The allowable offset range is set according to calibration accuracy requirements and is used to determine whether the marker point offset is within acceptable limits. Deviation correspondence parameters establish the correlation between feature deviations and positional deviations, recording the quantitative relationship between the changes in different features and the overall positional deviation of the module. For example, the offset of a marker point has a fixed proportional relationship with the module's positional deviation in a specific direction. These relationship data are compiled into deviation correspondence parameters.

[0026] The calibration displacement threshold includes displacement change parameters, feedback matching parameters, and threshold application parameters. Displacement change parameters describe the dynamic characteristics of the displacement adjustment process, including the displacement increment for each adjustment, the adjustment speed, and the acceleration of the displacement change. The displacement increment is set in stages according to the magnitude of the deviation, while the adjustment speed and acceleration take into account the operating performance of the mechanical device to avoid instability caused by excessively rapid adjustments. Feedback matching parameters measure the degree of agreement between real-time feedback data and expected data, including the sampling frequency of the feedback data and the matching error range. The sampling frequency is determined based on the adjustment speed to ensure timely capture of displacement changes, and the matching error range is used to determine the validity of the feedback data. Threshold application parameters specify how the threshold is used in different calibration stages, including the threshold's activation conditions, adjustment weights, and priority in multi-region calibration. Activation conditions are determined based on whether the deviation reaches a set value, adjustment weights reflect the degree of influence of different thresholds on the calibration results, and priority determines the processing order when multiple regions need to be calibrated simultaneously.

[0027] Displacement distribution predictions include trend analysis parameters and position-displacement relationship parameters. Trend analysis parameters reflect the overall trend of displacement change, including the rate of change of displacement over time, the direction of change, and potential inflection points. The rate of change is calculated from the displacement change over a continuous time period, the direction of change is determined by the sign of the displacement increment, and the inflection point is the point in time when the displacement trend changes. These parameters together constitute a complete description of the displacement trend. Position-displacement relationship parameters reflect the mutual influence of displacement between different locations, including the displacement transfer coefficient between adjacent locations and the degree of correlation between distant locations. The displacement transfer coefficient indicates the degree of influence of a displacement change at one location on adjacent locations, while the degree of correlation reflects the synchronicity of displacement changes between non-adjacent locations. These parameters allow for the analysis of the spatial distribution patterns of displacement.

[0028] The automatic alignment calibration and control scheme for LCD modules consists of error analysis parameters and calibration adjustment parameters. Error analysis parameters process and evaluate error data, including the statistical period, distribution characteristics, and criteria for identifying abnormal errors. The statistical period is set according to the calibration speed to ensure timely reflection of error changes. Distribution characteristics are described using statistical quantities such as the mean and variance of the errors. Criteria for identifying abnormal errors are used to identify error values ​​that exceed the normal range. Calibration adjustment parameters guide the actual calibration operation, including the motion parameters of the actuator, the limit on the number of adjustments, and the stabilization time after calibration. Motion parameters include the motor's rotation direction and speed. The limit on the number of adjustments is used to avoid invalid repeated adjustments. The stabilization time is the time interval after one adjustment during which the module position stabilizes. Through the synergistic effect of these parameters, automatic alignment calibration of the LCD module is achieved.

[0029] Example 2: The image acquisition module includes a data acquisition submodule, a difference integration submodule, and a basic integration submodule. These submodules work together to generate the basic image dataset.

[0030] When the data acquisition submodule operates, it first initiates the acquisition process based on the optical signals of the LCD module alignment area. These optical signals are captured by optical sensors distributed at different locations, arranged at preset angles to cover the horizontal, vertical, and multiple tilted viewpoints of the module alignment area. During acquisition, the submodule processes the acquired image data in real time, identifying blurry data using a sharpness assessment algorithm. Blurry data typically manifests as unclear edges and loss of detail. For such data, the submodule triggers a re-acquisition command until an image meeting the sharpness requirements is obtained. Simultaneously, a noise filtering algorithm removes noise from the image, which may be caused by sensor interference or ambient light fluctuations. The filtering process preserves key image features, preventing the loss of valid data. After data purification, the submodule arranges the image data in viewpoint order, for example, first the horizontal viewpoint, then the vertical viewpoint, and finally images from tilted viewpoints such as 45° and 135°. Each image is labeled with a viewpoint identifier and an acquisition timestamp, forming a regional image dataset containing raw data from multiple viewpoints.

[0031] The difference integration submodule operates on a regional image dataset, with its core task being to analyze the resolution and contrast differences between images from different viewpoints. The submodule first extracts the resolution parameters for each image, including the number of horizontal and vertical pixels, calculating the pixel density corresponding to the actual physical size. Simultaneously, it extracts the contrast parameter, quantifying the contrast by statistically analyzing the ratio of the highest to the lowest brightness in the image. Subsequently, the submodule compares the parameters of each viewpoint with preset reference viewpoint parameters, calculating the resolution change ratio and contrast change ratio. For example, when the reference viewpoint resolution is 1920×1080 and the resolution of a certain tilted viewpoint is 1280×720, the resolution change ratio is approximately 0.67. The submodule assigns weights to these ratios based on the importance of different viewpoints in alignment calibration; horizontal and vertical viewpoints, because they directly affect the main positioning direction, have relatively higher weights, while tilted viewpoints have lower weights. Next, the submodule performs fluctuation analysis on the ratios after weight sorting, and marks the angles where the ratio fluctuation exceeds the preset range. The image data of these angles may have acquisition anomalies and need to be checked again. Finally, image acquisition difference data containing angle parameter differences, weight values ​​and anomaly markers are generated.

[0032] The basic integration submodule performs multi-dimensional aggregation based on image acquisition difference data, including dimensions such as resolution, contrast, acquisition time, and viewpoint weight. The submodule uses numerical comparison algorithms to screen for numerical differences across each dimension. For example, resolution change ratios are categorized into intervals such as 0-0.2 and 0.2-0.4, and contrast change ratios are categorized into intervals such as 1-1.2 and 1.2-1.4. For each category interval, the submodule counts the number of images within that interval and their corresponding viewpoint information, analyzing the distribution characteristics of different intervals. After classification, the submodule sorts all image acquisition values ​​by numerical value, maintaining the association between image data and viewpoint identifiers and acquisition parameters during the sorting process to ensure data source traceability for subsequent calls. Through this processing, the submodule integrates the scattered viewpoint data into a unified format dataset. This dataset contains both the key parameters of the original images and reflects the differences between different viewpoints, ultimately generating a basic image dataset that can be used by subsequent modules.

[0033] The entire image acquisition module operates in a closed loop. The data acquisition submodule provides the raw materials, the difference integration submodule analyzes the difference features, and the basic integration submodule completes the standardization and integration of the data. The three work together to ensure that the generated image basic dataset can accurately reflect the multi-view features of the LCD module alignment area, providing a reliable data foundation for subsequent feature extraction and calibration analysis.

[0034] Example 3: The feature extraction module consists of a feature recognition submodule, a feature optimization submodule, and a feature selection submodule. These submodules work together to extract and optimize key feature information from the basic image dataset.

[0035] The feature recognition submodule operates based on the image dataset. First, it uses an edge detection algorithm to identify the edges of the LCD module in the image. By scanning the image pixels line by line and comparing the grayscale changes of adjacent pixels, when the grayscale change exceeds a set threshold, it is identified as an edge point. Continuous edge points are connected to form a complete module edge contour. During the recognition process, the curvature value of each point on the contour is recorded simultaneously. The curvature value is obtained by calculating the rate of change of the tangent direction of the contour at that point, reflecting the degree of curvature of the contour. For positioning markers in the image, a template matching algorithm is used for recognition. A preset marker template is compared with the region in the image. When the similarity exceeds a certain value, the position of the positioning marker is determined, and its coordinates in the image coordinate system are recorded. After data recording, the curvature data and marker coordinate data are normalized using the following formula:

[0036] in, Represents the normalized eigenvalues. Represents the original feature values. This represents the minimum value among the characteristic values ​​of this class. This represents the maximum value among the features of this type. This process transforms feature data of different magnitudes into the same data range, facilitating subsequent analysis and comparison. After processing, the data is classified according to the magnitude of the curvature values ​​and the distribution range of the marker coordinates. For example, curvature values ​​are divided into three categories: low curvature, medium curvature, and high curvature. Marker coordinates are divided into different groups according to their image regions. Finally, a feature parameter dataset containing various feature parameters is generated.

[0037] After receiving the feature parameter dataset, the feature optimization submodule begins analyzing the curvature and position values ​​within it. It calculates the matching degree between each feature combination and the underlying image dataset. The matching degree is determined by comparing the degree of agreement between the feature parameters and the actual regions in the image; a higher agreement indicates a better match. During the analysis, pattern matching technology is used to compare the feature parameter dataset with preset standard feature patterns, recording the results of each match, including the number of successfully matched features and the matching error. Based on the matching results, the feature combinations are adjusted, for example, by increasing the number of feature points with high matching degrees, removing feature regions with low matching degrees, and optimizing the structure of the feature combinations. For the adjusted feature combinations, pattern matching verification is performed again, and this process is repeated until a stable optimized feature group result is obtained. This result includes multiple candidate feature combinations and their corresponding matching parameters.

[0038] The feature selection submodule retrieves the optimized feature group results, compares the matching degree values ​​of each candidate feature combination, and selects the curvature and position combination with the highest matching degree. Based on this optimal combination, relevant parameters of the feature extraction algorithm are adjusted, such as modifying the grayscale threshold for edge detection and adjusting the template size for marker recognition, to make the feature extraction process more adaptable to the current image features. After adjustment, new recognition configuration parameters are input, and multiple feature extraction experiments are initiated. During the experiments, the stability of the parameter set is monitored, and the consistency of feature extraction results in different image frames is observed. Higher consistency indicates a more stable parameter set. After multiple verifications, when the parameter set can stably extract high-matching feature information, it is determined as the final feature extraction parameters. Based on these parameters, an image feature dataset is generated. This dataset contains filtered and optimized contour features and marker point information, accurately reflecting the key features of the LCD module.

[0039] The entire feature extraction module operates in a progressively refined process, from feature recognition of raw image data to optimization of feature combinations and final determination of feature parameters. Through the collaborative efforts of each submodule, the extracted image feature dataset ensures that it contains sufficient feature information while eliminating redundant and interfering data, providing high-quality feature data support for subsequent stages of the LCD module alignment and calibration system.

[0040] Example 4: The deviation analysis module includes a deviation factor analysis submodule, a parameter matching submodule, and a deviation parameter integration submodule. Each submodule processes image feature data in sequence and finally generates a set of position deviation parameters for calibration.

[0041] The deviation factor analysis submodule operates based on an image feature dataset. It continuously monitors key areas in the image to collect data related to positional deviations. This data includes the grayscale values ​​of each pixel in the image, obtained through the light sensitivity of an optical sensor; different module surfaces reflect light of varying intensities, leading to differences in grayscale values. The continuity of edge contours is determined by detecting breaks in contour segments; continuous contour segments more accurately reflect the actual edges of the module. The clarity of positioning markers is judged based on the contrast between the marked area and its surrounding area; higher contrast indicates clearer markings. During collection, this data is recorded chronologically to form a time-series dataset. This dataset is then analyzed using a sliding window method to identify outliers. A fixed-length window is set, and the mean and standard deviation of the data within the window are calculated. Data points whose difference from the mean exceeds three times the standard deviation are considered outliers and removed. The remaining valid data is partitioned according to the physical structure of the module, such as into the upper left corner region, the upper right corner region, the center region, etc. The gray-scale distribution, outline integrity and marker clarity characteristics of each region are statistically analyzed to form deviation factor analysis data, which includes the key factor values ​​of each region and the corresponding timestamp.

[0042] After receiving the deviation factor analysis data, the parameter matching submodule analyzes the impact of various image variables on the module's position deviation. It establishes a correlation model using grayscale value changes, contour continuity changes, and marker point sharpness changes as independent variables, and the deviation between the module's actual position and the standard position as the dependent variable. By calculating the contribution of different image factors to the deviation change, the influence degree of each factor is determined. For example, when the grayscale value of a certain area changes by 10%, the module's position deviation changes by 0.02mm, while when the contour continuity decreases by 20%, the position deviation changes by 0.05mm. This shows that contour continuity has a greater impact on the deviation. These influence degrees are scored; a higher score indicates a more significant impact of the factor on deviation adjustment. Based on the scoring results, the initial deviation setting is determined. Then, the weights of various parameters are adjusted iteratively, such as increasing the weight of high-scoring factors and decreasing the weight of low-scoring factors. The deviation changes are observed, and the parameter combination that minimizes the deviation is captured. During the adjustment process, the deviation value is recorded after each parameter change. Through multiple iterations, the deviation alignment result is finally obtained, which includes the optimal parameter combination and the corresponding deviation adjustment effect data.

[0043] The deviation parameter integration submodule filters deviation combinations from the deviation alignment results to find those that match the current image conditions, including illumination intensity, module model, and ambient temperature, ensuring the selected combinations are suitable for the current scenario. Parameter adjustment experiments are conducted on the selected deviation combinations. During the experiments, the value of a specific parameter is changed, and the change in module alignment accuracy is observed. For example, the weighting coefficient of grayscale values ​​is adjusted, and the deviation values ​​under different coefficients are recorded. Through multiple adjustments and verifications, the parameter settings are gradually optimized. For instance, if adjusting the grayscale value weighting coefficient from 0.3 to 0.25 results in a smaller deviation value, then 0.25 is adopted as the optimal value for that parameter. This process is repeated to optimize all parameters until the parameter settings maintain a stable deviation control effect in multiple consecutive experiments. These optimized parameters are then compiled into standardized operating parameters, clarifying the value range, applicable conditions, and adjustment rules for each parameter, solidifying them into operating standards, and generating a position deviation parameter set. This parameter set can be directly used to guide the deviation calculation and analysis process for module alignment.

[0044] The entire deviation analysis module achieves precise quantification of LCD module position deviation by extracting deviation factors, matching the degree of influence, and optimizing and integrating parameters. This provides a reliable parameter basis for subsequent calibration threshold setting and ensures that the calibration process can specifically handle different types of position deviations.

[0045] Example 5: The calibration threshold setting module includes a displacement extraction submodule, a displacement matching submodule, and a calibration allocation submodule. The displacement extraction submodule, based on a set of position deviation parameters, presets multiple displacement change monitoring points on the image of the LCD module's alignment area. These monitoring points are evenly distributed along the module's edge line and the center of the positioning mark; the number is determined by the module size to ensure coverage of the entire alignment area. By real-time acquisition of pixel position changes at the monitoring points, the physical displacement change value is calculated. Simultaneously, a high-frequency sampling method is used to record the rate of displacement increase or decrease. The sampling interval is set according to the module's movement speed to ensure that instantaneous rate changes are captured. Key change nodes are identified from the recorded rate data, such as the inflection point where the rate changes from positive to negative, or the point where the rate reaches its peak. The displacement values ​​corresponding to these nodes are arranged in chronological order to form the current displacement change characteristic values, which reflect the key stages of displacement change.

[0046] After receiving the current displacement change feature value, the displacement matching submodule compares and analyzes it with the real-time image feedback data. The real-time image feedback data includes the latest acquired monitoring point location information and overall image features. By comparing the correspondence between the node change values ​​and the feedback data, calibration is performed according to preset matching rules. If the difference between the displacement value of a node and the feedback data is within a set range, the match is considered successful; if it exceeds the range, the node value is corrected using interpolation to make the corrected node value more closely match the real-time feedback. After matching is completed, the displacement data within each interval is redistributed according to the range of displacement changes to ensure that the displacement distribution is consistent with the actual situation in the image feedback, forming a position displacement matching structure. This structure includes information such as the matched node values ​​and interval allocation ratios.

[0047] The calibration allocation submodule, based on a position displacement matching structure, employs an image threshold dynamic adjustment method to calculate the displacement distribution among changing nodes. This method comprehensively considers real-time acquired displacement values, the adjusted displacement matching results of upstream nodes, the real-time position height of the current node, the adjusted position matching results of downstream nodes, as well as parameters such as displacement weight coefficient, position weight coefficient, dynamic adjustment weight coefficient, and node threshold lower limit. Through multi-parameter collaborative calculation, the threshold range for each node is determined. An upper and lower threshold are set for each node; the upper limit is the maximum allowable displacement adjustment amount for that node, and the lower limit is the minimum adjustment amount. These thresholds are applied to the module alignment adjustment process, allocating the application ratio of the thresholds according to the importance of different regions to form a calibration displacement threshold. This threshold includes the upper and lower limit values ​​of each node and its application priority.

[0048] The displacement prediction module consists of a displacement data capture submodule, a displacement load analysis submodule, and a displacement distribution inference submodule. The displacement data capture submodule, based on a calibrated displacement threshold, uses image analysis algorithms to track sampling points in an image and collect displacement data for each sampling point. During the acquisition process, an outlier detection mechanism identifies and removes data exceeding a reasonable range, and error correction is performed on the remaining data to eliminate systematic errors introduced by the image sensor. The corrected displacement data is then stored hierarchically according to intervals defined by the calibrated displacement threshold. Each layer of data undergoes feature processing to extract features such as the frequency and amplitude of displacement changes, generating a feature-based displacement dataset. This dataset contains the displacement feature parameters and corresponding sampling times for each interval.

[0049] After receiving the characteristic displacement dataset, the displacement load analysis submodule divides the data into multiple continuous intervals based on the magnitude of the calibrated displacement. The span of each interval is determined according to the required precision of the displacement change. Trend analysis is performed on the displacement data within each interval. By calculating the displacement difference between adjacent sampling points, the direction of the displacement change trend is identified, such as increasing, decreasing, or remaining stable. The characteristics of displacement fluctuations, such as fluctuation period and amplitude, are also analyzed. These trend and fluctuation characteristics are then associated with the corresponding calibrated displacement intervals to form a calibrated displacement and displacement change feature set. This feature set contains information such as trend parameters and fluctuation parameters for each interval.

[0050] The displacement distribution inference submodule adjusts the values ​​of characteristic parameters based on the calibration displacement and displacement change feature set, such as correcting the time window size of trend analysis to make the trend data more closely reflect actual changes. Based on the adjusted trend data, it determines the range of displacement distribution intervals and uses statistical analysis methods to predict the displacement values ​​within each interval, forecasting displacement changes over a future period, thus forming a displacement distribution prediction value. This prediction value includes information such as the predicted displacement range and probability of change for each interval.

[0051] The calibration feedback control module includes an error analysis submodule, a parameter adjustment submodule, and a calibration control submodule. The error analysis submodule, based on displacement distribution predictions, acquires the module's current position and displacement data through a real-time positioning system. It compares the real-time displacement values ​​with the predicted values, calculates the difference, and obtains the position displacement error value for each monitoring point. The error value is then associated with the corresponding position information, marking the error magnitude at each location and forming an error distribution map. This map visually displays the error situation in each region.

[0052] The parameter adjustment submodule sets the calibration adjustment parameters for the module alignment device based on the magnitude of the position displacement error. For areas with large error values, a larger adjustment range is set to accelerate the calibration process; for areas with small error values, a fine-tuning method is used to avoid over-adjustment that could lead to new deviations. By comparing the effects of different adjustment parameters through multiple experiments, the parameter combination that minimizes the error is selected. These parameters are then integrated by region to generate a calibration adjustment parameter set, which includes information such as the adjustment range and direction for each region.

[0053] The calibration control submodule, based on the calibration adjustment parameter set, sends control commands to the actuators of the module alignment device, applying the corresponding adjustment parameters at the position of each alignment device. Calibration operations are performed item by item according to a preset sequence, adjusting areas with larger errors first, followed by areas with smaller errors. During the adjustment process, a real-time monitoring system synchronously tracks the module's position and displacement changes, dynamically adjusting the calibration order and adjustment range of each area based on the monitoring results to ensure the accuracy and efficiency of the calibration process. After multiple rounds of adjustments, an automatic calibration and control scheme for LCD module alignment is generated, which includes the final adjustment parameters, operating procedures, and monitoring requirements.

[0054] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0055] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. An image analysis-based LCD module alignment calibration system, characterized by, The system comprises: The image acquisition module acquires multi-view image data based on the optical signals of the LCD module alignment area, integrates image resolution and contrast parameters, and generates an image basic data set; The feature extraction module extracts feature information of module edge profiles and positioning marks based on the image basic data set, screens key feature points and feature areas, and obtains an image feature data set; The deviation analysis module extracts the difference between the current module position and the standard position based on the image feature data set, analyzes the profile matching degree and the mark point offset relationship, matches the position deviation and the feature correspondence, and obtains a position deviation parameter set; The calibration threshold setting module extracts the current calibration displacement change value based on the position deviation parameter set, combines real-time image feedback data, allocates positions and displacements, sets a threshold value, and applies the threshold value to the adjustment of module alignment, and obtains a calibration displacement threshold value; The displacement prediction module captures image sampling point displacement data based on the calibration displacement threshold value, infers the displacement change trend of the sampling points by combining image analysis, analyzes the displacement change trend corresponding to the calibration displacement, classifies the displacement change trend according to the inference result, adjusts and screens the classified data by combining displacement change information, and obtains a displacement distribution prediction value; The calibration feedback control module analyzes the error value of position and displacement by combining real-time module position and displacement data based on the displacement distribution prediction value, adjusts the module alignment device by combining the error value, and obtains an LCD module alignment automatic calibration regulation scheme.

2. The image analysis based LCD module alignment calibration system of claim 1, wherein: The image basic data set includes a resolution parameter set, a contrast parameter set, and an image integration parameter set. The image feature data set includes profile feature parameters, mark point coordinate parameters, and feature screening parameters. The position deviation parameter set includes difference extraction parameters, matching degree analysis parameters, and deviation correspondence parameters. The calibration displacement threshold value includes displacement change parameters, feedback matching parameters, and threshold application parameters. The displacement distribution prediction value includes trend analysis parameters and position displacement relationship parameters. The LCD module alignment automatic calibration regulation scheme includes error analysis parameters and calibration adjustment parameters.

3. The image analysis based LCD module alignment calibration system of claim 1, wherein: The image acquisition module comprises: The data acquisition submodule acquires multi-view image data based on the optical signals of the LCD module alignment area, locates fuzzy data, removes noise data, arranges the acquired image data in order of view angle, and generates a regional image data set; The difference integration submodule analyzes the image resolution and contrast between views based on the regional image data set, calculates the view parameter change ratio, sorts the view difference values by weight, marks the view angles with excessively large fluctuation differences, and acquires image acquisition difference data; The basic integration submodule calls view difference values for multi-dimensional summarization based on the image acquisition difference data, screens image acquisition numerical differences, classifies them by numerical size, and arranges the image acquisition values in order, generating an image basic data set.

4. The image analysis based LCD module alignment calibration system of claim 1, wherein: The feature extraction module comprises: Based on the image dataset, the feature recognition submodule identifies the contour state of each module edge, records the curvature of the contour and the position of the marker point, normalizes the recorded data, sorts the normalized data by curvature and position, and generates a feature parameter dataset. The feature optimization submodule analyzes the curvature and position values ​​in the feature parameter dataset, filters feature combinations with high matching degree with the image base, records the matching results through pattern matching, adjusts the feature combinations, and generates feature group optimization results. The feature selection submodule retrieves the optimized result of the feature combination, determines the optimal curvature and position combination with matching degree, adjusts the feature extraction parameters, inputs the recognition configuration, verifies the stability of the parameter set, and generates an image feature dataset.

5. The image analysis based LCD module alignment calibration system of claim 1, wherein: The deviation analysis module includes: The deviation factor analysis submodule, based on the image feature dataset, collects key data through image monitoring, including grayscale values, contour continuity, and marker point clarity. It performs time series analysis on the data, removes outliers, partitions the remaining data, and obtains deviation factor analysis data. The parameter matching submodule analyzes the influence of image variables on the deviation between the module position and the standard position by analyzing the deviation factor analysis data, calculates the degree of influence of the change of each image factor on the deviation adjustment, determines the optimal deviation setting for matching based on the influence score, iteratively adjusts the parameters to capture the optimal combination, and obtains the deviation docking result. The deviation parameter integration submodule selects the position and standard position deviation combination that matches the current image conditions from the deviation docking results, conducts parameter adjustment experiments, optimizes the parameter settings through multiple adjustments and verifications, determines and solidifies the parameters as the operation standard, and generates a position deviation parameter set.

6. The image analysis based LCD module alignment calibration system of claim 1, wherein: The calibration threshold setting module includes: The displacement extraction submodule locates the displacement change monitoring point based on the position deviation parameter set, extracts the displacement change value in the monitoring area, continuously records the displacement increase and decrease rate, extracts multiple key change nodes corresponding to the change rate, sorts the node values ​​in order, and obtains the current displacement change feature value. The displacement matching submodule analyzes the node change values ​​and real-time image feedback data based on the current displacement change feature value, performs calibration according to a predetermined matching criterion, calls the matching criterion to perform distribution redistribution within the displacement interval, and obtains the position displacement matching structure. The calibration allocation submodule, based on the position displacement matching structure, uses an image threshold dynamic adjustment method to calculate the distribution of displacement among changing nodes, sets upper and lower limits for node thresholds, applies the thresholds to the module alignment values, and allocates them to obtain the calibration displacement threshold.

7. The image analysis based LCD module alignment calibration system of claim 6, wherein: The image threshold dynamic adjustment method includes obtaining the allocation calibration value of the matching displacement change node, and comprehensively calculating it by real-time displacement value, displacement matching value adjusted by upstream node, real-time position height calculated by current node, position matching value adjusted by downstream node, displacement weight coefficient, position weight coefficient, dynamic adjustment weight coefficient and threshold lower limit set by node.

8. The image analysis based LCD module alignment calibration system of claim 1, wherein: The displacement prediction module includes: The displacement data capture submodule, based on the calibrated displacement threshold, applies image analysis algorithms to capture displacement data of sampling points, remove outliers and correct errors, stores the data in a hierarchical manner according to intervals, performs feature processing, and generates a feature displacement dataset. Based on the characteristic displacement dataset, the displacement load analysis submodule divides the intervals according to the calibration displacement, extracts the changing trend and fluctuation characteristics, and generates a calibration displacement and displacement change feature set. The displacement distribution inference submodule adjusts the feature parameters and calibrates the trend data based on the calibrated displacement and displacement change feature set, extracts the distribution interval, and performs numerical prediction to obtain the displacement distribution prediction value.

9. The image analysis based LCD module alignment calibration system of claim 8, wherein: The image analysis algorithm includes calculating displacement feature values ​​and generating a feature displacement dataset. The displacement feature values ​​are calculated by comprehensively considering the weight of each data point, the original displacement value of each sampling point, the position value of each sampling point, and the calibration coefficient of each sampling point. The total number of sampling points is the total number of sampling points.

10. The image analysis based LCD module alignment calibration system of claim 1, wherein: The calibration feedback control module includes: The error analysis submodule extracts real-time module position and displacement data based on the displacement distribution prediction value, analyzes the real-time displacement value and the prediction value, and generates a position displacement error value by matching the displacement difference with the current position information. The parameter adjustment submodule sets the calibration adjustment parameters of the module alignment device based on the position displacement error value. It sets the adjustment range for areas with large errors and makes fine adjustments for areas with low errors. By comparing the calibration effect, it filters and integrates the matching parameter set to generate a calibration adjustment parameter set. Based on the set of calibration adjustment parameters, the calibration control submodule applies adjustment parameters to each alignment device position, performs calibration operations item by item, synchronously monitors the module position and displacement, gradually adjusts the calibration operation sequence of each area, and generates an automatic calibration and control scheme for LCD module alignment.

Images