An online screen size detection method based on image analysis

Abstract

The present invention proposes an online detection method for screen size based on image analysis, comprising constructing a virtual three-dimensional model prototype of a display screen and a camera, simulating the movement trajectory and zoom effect of the camera, and generating image previews at multiple angles and distances; acquiring three-dimensional simulation data and real measurement data; comparing the real measurement data with the three-dimensional simulation data, and adjusting the shooting parameters of the camera according to the deviation results; acquiring and processing pre-shot images, quantitatively evaluating the consistency of black edges, and optimizing the actual shooting configuration of the camera; adjusting the ambient light parameters in combination with the physical size data and curvature parameters of the display screen; acquiring image data, and repeatedly testing and finely adjusting the shooting parameters of the camera; correcting the pre-shot images; detecting the edge alignment and the width of the black edges according to the image data of the corrected pre-shot images, and comparing with the real measurement data at the same time, and performing subsequent image optimization to match the actual observation results.

Description

Technical Field

[0001] The present invention relates to the field of information technology, and in particular to an online screen size detection method based on image analysis. Background Art

[0002] With the rapid development of modern display technology, accurate measurement of screen size has become increasingly important, especially in production lines and quality control. However, existing screen size detection methods often rely on physical contact measurement tools or simple two-dimensional image analysis. These methods are not only time-consuming, but also difficult to adapt to complex screen curvature and frame designs. Especially when conducting online quality inspection, existing technologies are difficult to overcome the influence of camera angle, position and ambient light on image capture, which makes the image analysis results often interfered by perspective distortion, reflection and shadow, and cannot truly and accurately evaluate the edge consistency and dimensional accuracy of the screen.

[0003] Specifically, due to the physical characteristics of the display, such as the black border of the frame and the curvature of the screen, traditional image analysis methods often have difficulty in accurately measuring the screen size, resulting in deviations between the actual results and the theoretical values. In particular, when using a camera to shoot the screen from different angles, perspective distortion and ambient light effects (such as reflections and shadows) will further reduce the accuracy of the measurement. In addition, the visual differences in the device design itself also bring additional challenges to size detection. For example, when shooting a display, the distance and angle between the camera and the display, as well as the design of the display will affect the image, especially in terms of the consistency of the display frame.

[0004] First, the distance between the camera and the display screen is directly related to the degree of perspective distortion. When the camera is close to the display screen, the edge of the display screen close to the camera will appear wider due to the perspective relationship. On the contrary, the edge far away from the camera will appear narrower due to the opposite effect of perspective. This leads to inconsistent widths of the black borders at the edges of the display screen in the image. Secondly, if the display screen itself adopts a curved design, the black borders at its edges have different widths in physical size. In this case, the image captured by the camera will further magnify the inconsistency in physical size, making the curvature of the display screen more significant in the image. Finally, the degree of parallelism between the camera and the display screen will also affect the consistency of the black borders. If the camera is too high or too low relative to the display screen and cannot maintain a parallel relationship with the display screen, then in the final image, the black borders at the top and bottom or left and right parts of the display screen will appear to have different widths.

[0005] In summary, considering the distance and angle relationship between the camera and the display screen, as well as the physical characteristics of the display screen, targeted research is needed on how to arrange the camera's appropriate shooting angle and position, how to obtain display screen images that maintain consistency in black border width, and how to reduce perspective distortion and visual differences caused by the device design itself. Summary of the invention

[0006] In view of the deficiencies in the prior art, the present invention proposes an online screen size detection method based on image analysis.

[0007] The present invention discloses an online screen size detection method based on image analysis, comprising:

[0008] S1: constructing a virtual three-dimensional model prototype of a display screen and a camera, simulating a moving trajectory and a zoom effect of the camera, and generating image previews at multiple angles and distances according to the moving trajectory and the zoom effect;

[0009] S2: Acquire three-dimensional simulation data according to the virtual three-dimensional model prototype, and use an optical measurement tool to acquire real measurement data of the precise width of the actual black border of the display screen, wherein the real measurement data is used to correct the perspective error in the simulation process;

[0010] S3: comparing the real measurement data with the three-dimensional simulation data, and adjusting the shooting parameters of the camera according to the deviation result between the real measurement data and the three-dimensional simulation data, wherein the shooting parameters include a preset position and a shooting angle;

[0011] S4: acquiring and processing the pre-shot image, performing quantitative evaluation on the consistency of the black border, and optimizing the actual shooting configuration of the camera according to the result of the quantitative evaluation, wherein the actual shooting configuration includes an actual shooting position and an actual shooting angle;

[0012] S5: adjusting ambient light parameters in combination with the physical size data and curvature parameters of the display screen to reduce the impact of reflection and shadow on image quality;

[0013] S6: acquiring image data, and repeatedly testing and finely adjusting the height parameter and tilt angle parameter of the camera according to the image data until obtaining shooting parameters with the best consistency of black border width;

[0014] S7: Correcting the pre-captured image, including correcting the perspective distortion of the four corners of the display screen using a digital deformation technique, and adjusting the graphic ratio of the image data and the symmetry of the black border according to the edge data collected during the repeated tests;

[0015] S8: Detecting the black border edge alignment and the black border width according to the corrected image data of the pre-shot image, and comparing them with the real measurement data, and performing subsequent image optimization to match the actual observation result.

[0016] In an optional embodiment, the step of constructing a virtual three-dimensional model prototype of a display screen and a camera, simulating a movement trajectory and a zoom effect of the camera, and generating image previews at multiple angles and distances according to the movement trajectory and the zoom effect includes:

[0017] S101: Designing with three-dimensional modeling software, setting model parameters of the display screen and the camera, and constructing a virtual three-dimensional model prototype of the display screen and the camera;

[0018] S102: after the virtual three-dimensional model prototype is constructed, simulating the movement trajectory of the camera so that the simulated movement trajectory is close to the actual operation scenario, the movement trajectory including movement speed and angle change;

[0019] S103: adjusting variables in the movement trajectory that affect image quality, wherein the variables include light and light reflection in a simulated environment;

[0020] S104: simulating the zoom effect of the camera by the three-dimensional modeling software to obtain simulated images of the camera at different focal lengths, wherein the simulated images serve as a preset standard for subsequent testing and error correction;

[0021] S105: simulating the camera shooting at multiple angles and multiple distances, determining the imaging performance of the camera at different angles and different distances, and obtaining an image preview effect according to the imaging performance;

[0022] S106: comparing and analyzing the deviation between the image preview effect obtained under each set of angles and distances and the preset standard, and adjusting the virtual three-dimensional model prototype;

[0023] S107: Verify the results of the entire simulation process and adjust the simulation parameters through feedback to optimize the image generation process.

[0024] In an optional embodiment, the acquiring of three-dimensional simulation data according to the virtual three-dimensional model prototype, using an optical measurement tool to acquire real measurement data of the precise width of the actual black border of the display screen, wherein the real measurement data is used to correct the perspective error in the simulation process, comprises:

[0025] S201: extracting three-dimensional simulation data of the visual effect of the virtual three-dimensional model prototype using image processing technology to obtain expected width information of the black border;

[0026] S202: using a high-precision optical measurement tool to measure the actual black border, obtaining real measurement data of the precise width of the actual black border, and recording the actual width information of the actual black border;

[0027] S203: Compare and analyze the extracted expected width information with the actual width information recorded by the high-precision optical measurement tool to determine whether a deviation occurs; if a deviation occurs, calibrate the high-precision optical measurement tool;

[0028] S204: After the device is calibrated, the actual black border is measured to obtain new actual width information, and the actual width information is compared with the expected width information again;

[0029] S205: Determine the deviation by comparing the data in S204, optimize the image processing technology, and use the three-dimensional simulation data extracted by the optimized image processing technology to guide the precise setting of the camera;

[0030] S206: Determine the placement position of the camera to capture a complete image of the display screen;

[0031] S207: Obtaining guidance data through the device calibration and the placement position of the camera, and fine-tuning the position of the camera according to the guidance data;

[0032] S208: After fine-tuning, a final test is performed to measure the width of the black border using the high-precision optical measurement tool to obtain final actual width information, ensuring that the final actual width information is consistent with the expected width information of the virtual three-dimensional model prototype.

[0033] In an optional embodiment, the comparing the real measurement data with the three-dimensional simulation data and adjusting the shooting parameters of the camera according to the deviation result between the real measurement data and the three-dimensional simulation data, wherein the shooting parameters include a preset position and a shooting angle, comprises:

[0034] S301: Analyze the difference between the three-dimensional simulation data and the real measurement data to obtain a deviation result;

[0035] S302: Formulate data adjustment strategies through deviation result analysis;

[0036] S303: After obtaining the data adjustment strategy, automatically adjust the preset position of the camera so that the spatial positioning of the camera during shooting meets the model adjustment standard;

[0037] S304: determining whether the preset position of the camera meets the requirements of the three-dimensional simulation data, and if so, adjusting the shooting angle of the camera to obtain the best shooting posture;

[0038] S305: After adjusting the shooting angle, continue to evaluate the effect of the new shooting parameters through comparative analysis;

[0039] S306: When the effect brought by the new shooting parameters in S305 still has deviations, the correction process of steps S302-S305 is looped until the preset model adjustment standard is reached;

[0040] S307: When the deviation is controlled within an acceptable range, the new shooting parameter setting is saved as a part of the shooting plan.

[0041] In an optional embodiment, the acquiring and processing of the pre-shot image, performing a quantitative evaluation on the consistency of the black border, and optimizing the actual shooting configuration of the camera according to the result of the quantitative evaluation, wherein the actual shooting configuration includes an actual shooting position and an actual shooting angle, comprises:

[0042] S401: loading a pre-shot photo, identifying the outline of the black edge through a specially designed algorithm, and obtaining edge information of the black edge;

[0043] S402: performing geometric transformation operator processing on the obtained edge information to determine the actual width of the black edge and obtain an edge width value set;

[0044] S403: Calculating the standard deviation and the average value of the edge width value set by a statistical method, and presetting a quantization parameter, thereby quantitatively evaluating the consistency of the black edge and obtaining a quantization consistency result before adjustment;

[0045] S404: using machine learning technology to analyze the correlation between the quantization parameters and the actual visual effect, obtaining a predicted shooting configuration through a training model, simulating the predicted shooting configuration through a simulator, and simulating a simulated visual effect for the predicted shooting configuration, wherein the preset shooting configuration includes a shooting position and an angle configuration;

[0046] S405: Generate an optimization report by comparing the simulated visual result and the actual shooting visual effect, wherein the optimization report includes recommended parameters for shooting configuration;

[0047] S406: Automatically adjust the shooting position and shooting angle of the camera according to the recommended parameters in the optimization report, and accurately perform the adjustment process through a control algorithm;

[0048] S407: After the adjustment is completed, retake the photo of the display screen to obtain a new photo, and use image analysis software to evaluate the consistency of the black border of the new photo to obtain a quantitative consistency result after adjustment;

[0049] S408: Compare the quantization consistency result before the adjustment with the quantization consistency result after the adjustment, determine whether the continuity of the visual effects of the pre-taken photo and the new photo is improved, and update the machine learning model according to the comparison result.

[0050] In an optional embodiment, the adjusting the ambient light parameters in combination with the physical size data and curvature parameters of the display screen to reduce the impact of reflection and shadow on image quality includes:

[0051] S501: Obtaining physical size data of the display screen, and calculating the optimal viewing distance and viewing angle of the display in combination with curvature parameters;

[0052] S502: Detecting image quality problems existing on the current display screen, where the image quality problems include reflection problems and shadow problems;

[0053] S503: Acquire ambient light parameters and perform preliminary evaluation, wherein the ambient light parameters include light source type, brightness and color temperature;

[0054] S504: adjusting the ambient light parameter according to the curvature parameter and the physical size data of the display screen, and the data fed back by the image analysis software, to obtain an adjusted image effect;

[0055] S505: Detect the adjusted image effect again to determine whether there are still reflection problems and shadow problems. If yes, dynamically adjust the ambient light parameters until the predetermined standard is met;

[0056] S506: Setting the ambient light parameters when the shadow problem and the reflection problem are minimized as ideal ambient light parameters, and applying the ideal ambient light parameters to the ambient light source;

[0057] S507: Continuously monitor the image quality problem of the display screen and the ambient light source, and dynamically adjust the ambient light parameters according to the image quality problem and the ambient light source to maintain the best display effect of the display

[0058] In an optional embodiment, the acquiring of image data, and repeated testing and fine-tuning of the height parameter and tilt angle parameter of the camera according to the image data until the shooting parameters with the best consistency of black border width are obtained include:

[0059] S601: Setting the height parameters and tilt angle parameters of the initial camera, and automatically adjusting the camera to a preset starting position;

[0060] S602: Obtain image data through an image acquisition sensor, determine whether the width of the black border displayed by the image data is uneven, and if so, gradually adjust the height of the camera until the black border width is close to a preset range, thereby obtaining image data after the first adjustment;

[0061] S603: judging whether the best consistency of black border width is achieved according to the image data after the first adjustment, and if not, automatically adjusting the tilt angle parameter according to the image data to obtain the image data after the second adjustment;

[0062] S604: evaluating again whether the optimal consistency of the black border width is achieved according to the image data after the second adjustment, and if so, setting the height parameter and the tilt angle parameter as a shooting parameter; if not, entering the next round of adjustment;

[0063] S605: Using a statistical method, recording and analyzing multiple sets of image data, and establishing a correlation model between different height parameters of the camera and different tilt angle parameter adjustments;

[0064] S606: In a subsequent shooting task, shooting parameters when different height parameters and different tilt angle parameters are combined are set in advance according to the correlation model;

[0065] S607: According to the shooting parameters, the tilt angle parameter of the camera is automatically adjusted to a value that matches the height parameter, and the camera automatically executes the shooting parameters each time shooting is performed.

[0066] In an optional embodiment, the correcting the pre-captured image includes correcting the perspective distortion of the four corners of the display screen using a digital deformation technology, and adjusting the graphic ratio of the image data and the symmetry of the black border according to the edge data collected during the repeated experiments, including:

[0067] S701: Acquire perspective distortion parameters in the captured image, use digital deformation technology to perform perspective distortion correction operations, correct the distorted graphics at the four corners of the display screen, and obtain a corrected image;

[0068] S702: Based on the corrected image, an enhancement algorithm is called to refine the edge, improve the edge clarity of the image, and obtain an edge-enhanced image effect;

[0069] S703: combining the edge data collected in the repeated test process, performing graphic proportion adjustment on the edge enhanced image effect to obtain a proportion adjusted graphic;

[0070] S704: determining whether the symmetry of the proportion-adjusted graphic meets the expected requirement, and if not, performing detailed correction on the asymmetric part of the black border;

[0071] S705: If there is still a slight error in the perspective of the image corrected in S701 after adjustment, the digital deformation supplementation technology is used for adjustment again, and the perspective distortion parameters are optimized according to the feedback of the adjustment result;

[0072] S706: Reapply the optimized parameters in S701-S705 to ensure that the image effect after each correction is balanced and consistent;

[0073] S707: Determine whether the size of the scale-adjusted graphic in S703 meets the standard size. If not, reacquire the size data of the graphic, adjust the scale calibration algorithm and execute the graphic scale adjustment step again.

[0074] In an optional embodiment, detecting the black border edge alignment and the black border width according to the image data of the corrected pre-shot image, comparing with the real measurement data, and performing subsequent image optimization to match the actual observation result includes:

[0075] S801: extracting image edge features using an edge detection algorithm according to the corrected image data of the pre-shot image, and obtaining position information and accuracy requirements of the image edge;

[0076] S802: Automatically detecting the black edge width and the black edge alignment in the pre-shot image according to the extracted image edge features, and obtaining black edge width data and black edge alignment deviation data;

[0077] S803: Obtaining the black border edge alignment and black border width under ideal conditions according to the real measurement data;

[0078] S804: comparing the black border width data and the black border edge alignment deviation data with the real measurement data to determine whether there is a deviation, and if so, using an algorithm to calculate an image correction amount;

[0079] S805: adjusting the image data according to the image correction amount and performing edge alignment correction;

[0080] S806: After the correction, the black edge width is detected again using a high-precision edge detection algorithm to obtain corrected image data; judging whether the corrected image data is consistent with the real measurement data according to the detection result, and if so, determining that the image has met the optimization requirements;

[0081] S807: continuously monitor the black border width data and the black border edge alignment deviation data of the image and the comparison deviation value with the true measurement data, and execute S804-S806 in a loop.

[0082] Compared with the prior art, the present invention has the following advantages:

[0083] The present invention can obtain a display screen image with a consistent black border width by arranging a proper shooting angle and position of a camera, thereby reducing perspective distortion and visual differences caused by the device design itself. BRIEF DESCRIPTION OF THE DRAWINGS

[0084] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0085] Figure 1 The present invention is a flowchart of an online screen size detection method based on image analysis.

[0086] Figure 2 A schematic diagram of an online screen size detection method based on image analysis according to the present invention.

[0087] Figure 3 It is another schematic diagram of an online screen size detection method based on image analysis of the present invention. DETAILED DESCRIPTION

[0088] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0089] In the description of the present invention, it should be noted that the terms "center", "upper", "lower", "left", "right", "vertical", "horizontal", "inner", "outer", etc., indicating the orientation or positional relationship, are based on the orientation or positional relationship shown in the drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as limiting the present invention. In addition, the terms "first", "second", "third", "fourth", etc. are only used for descriptive purposes and cannot be understood as indicating or implying relative importance.

[0090] In the description of the present invention, it should be noted that, unless otherwise clearly specified and limited, the terms "installed", "connected", and "connected" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be indirectly connected through an intermediate medium, or it can be the internal communication of two components. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0091] See also Figure 1-3 The embodiment of the present invention discloses an online screen size detection method based on image analysis, comprising:

[0092] S1: constructing a virtual three-dimensional model prototype of a display screen and a camera, simulating a moving trajectory and a zoom effect of the camera, and generating image previews at multiple angles and distances according to the moving trajectory and the zoom effect;

[0093] In an optional embodiment, the step of constructing a virtual three-dimensional model prototype of a display screen and a camera, simulating a movement trajectory and a zoom effect of the camera, and generating image previews at multiple angles and distances according to the movement trajectory and the zoom effect includes:

[0094] S101: using 3D modeling software to design, set model parameters of the display screen and the camera, and build virtual 3D model prototypes of the display screen and the camera to ensure matching between the two in virtual space;

[0095] S102: after the virtual three-dimensional model prototype is constructed, simulating the movement trajectory of the camera so that the simulated movement trajectory is close to the actual operation scenario, the movement trajectory including movement speed and angle change;

[0096] S103: adjusting variables in the movement trajectory that affect the image quality, wherein the variables include light and light reflection in the simulated environment, and timely adjusting the light, reflection and other variables that may affect the image quality in the simulated environment to ensure that the generated image data has a high reference value;

[0097] S104: simulating the zoom effect of the camera through the 3D modeling software to obtain simulated images of the camera at different focal lengths to reflect the performance of the zoom function of the camera; the simulated images are used as a preset standard for subsequent testing and error correction to improve the imaging quality of the final product;

[0098] S105: simulating the camera shooting at multiple angles and multiple distances, determining the imaging performance of the camera at different angles and different distances, and obtaining an image preview effect according to the imaging performance;

[0099] S106: comparing and analyzing the deviation between the image preview effect obtained under each set of angles and distances and the preset standard, and adjusting the virtual three-dimensional model prototype to approach the real result;

[0100] S107: Verify the results of the entire simulation process and adjust the simulation parameters through feedback to optimize the image generation process.

[0101] For example, when creating high-precision display and camera models, 3D software such as Blender or Maya is used to introduce actual size parameters, such as display size 100mmx200mmx8mm and camera lens radius 10mm, to ensure that the model is proportional to the real object. Advanced rendering technology such as physically based rendering (PBR) is also used to ensure that the glossiness of the material (such as metalness 0.8 and roughness 0.2) matches the actual material.

[0102] Specifically, when setting the camera trajectory, the built-in animation component of the 3D software is used to define the trajectory with Bezier curves, and the lens is controlled to rotate within 0 to 30 degrees along the x-axis and 0 to 45 degrees along the y-axis. The speed is set to 0.5m / s to simulate the real motion trajectory and smooth the transition through the keyframe interpolation algorithm.

[0103] Specifically, when adjusting the light, ray tracing technology is used to simulate real-world light, the ambient light intensity is set to 1000 lumens, and the refractive index of the reflective surface is set to 1.5, so that a lighting effect close to reality can be obtained through light simulation.

[0104] Specifically, when simulating the zoom effect, the imaging formula at different focal lengths is calculated through a mathematical model, such as a simple thin lens formula f=d+di, where f is the focal length, d is the object distance, and di is the imaging distance, so as to maintain clarity when adjusting the focal length (35mm to 70mm).

[0105] Specifically, it automatically shoots at multiple angles, for example, setting a preview point every 10 degrees from 0 to 180 degrees, and uses a ray tracing algorithm to calculate the imaging conditions at each angle to predict the image effect. In order to obtain imaging performance at different working distances, the imaging distance is set from 0.5m, 1m to 2m, and the text clarity and color accuracy of the display screen at each distance are evaluated.

[0106] Specifically, an image processing algorithm such as an edge detection algorithm (eg, Canny algorithm) is used to analyze the difference between the simulated image and the real image, and the model is adjusted according to the obtained edge intensity difference and distribution statistics.

[0107] Specifically, statistical error analysis (such as root mean square error RMSE calculation) is used to quantitatively analyze the consistency between the generated image and the standard image, and the error reduction curve is used to verify the effectiveness of the optimization process.

[0108] S2: Acquire three-dimensional simulation data according to the virtual three-dimensional model prototype, and use an optical measurement tool to acquire real measurement data of the precise width of the actual black border of the display screen, wherein the real measurement data is used to correct the perspective error in the simulation process;

[0109] In an optional embodiment, the acquiring of three-dimensional simulation data according to the virtual three-dimensional model prototype, using an optical measurement tool to acquire real measurement data of the precise width of the actual black border of the display screen, wherein the real measurement data is used to correct the perspective error in the simulation process, comprises:

[0110] S201: extracting three-dimensional simulation data of the visual effect of the virtual three-dimensional model prototype using image processing technology to obtain expected width information of the black border;

[0111] S202: using a high-precision optical measurement tool to measure the actual black border, obtaining real measurement data of the precise width of the actual black border, and recording the actual width information of the actual black border;

[0112] S203: Compare and analyze the extracted expected width information with the actual width information recorded by the high-precision optical measurement tool to determine whether a deviation occurs; if a deviation occurs, calibrate the high-precision optical measurement tool to adjust the accuracy of the high-precision optical measurement tool to ensure the accuracy of subsequent data;

[0113] S204: After the device calibration is completed, the actual black border is measured to obtain new actual width information, and the data is compared with the expected width information again to verify the validity of the device calibration;

[0114] S205: Determine the deviation by comparing the data in S204, optimize the image processing technology, and improve the accuracy of the three-dimensional simulation data extracted by the optimized image processing technology to guide the precise setting of the camera to ensure that the camera is positioned to accurately capture the edge of the screen;

[0115] S206: Determine the placement position of the camera to capture a complete image of the display screen;

[0116] S207: Obtaining guidance data through the device calibration and the placement position of the camera, and fine-tuning the position of the camera according to the guidance data to obtain the best video recording effect;

[0117] S208: After fine-tuning, a final test is performed, using the high-precision optical measurement tool to measure the width of the black border to obtain final actual width information, ensuring that the final actual width information is consistent with the expected width information of the virtual three-dimensional model prototype, so as to ensure that the actual effect is consistent with the visual effect provided by the virtual three-dimensional model prototype.

[0118] For example, in the process of 3D model analysis, we first use image processing technology to perform edge detection to determine the width of the black border in the display model. For example, using the Canny edge detection algorithm, input the model image parameters, and output the edge width We. Setting a scaling factor ρ from pixels to actual length units, we can get the predicted width of the black border in the model Wm=We*ρ. Then, based on this theoretical value, use high-precision optical measurement tools to perform physical measurements, such as using a laser distance measurement tool to determine the actual black border width Wa.

[0119] Specifically, the measured value is compared with the theoretical value, and the deviation ΔW = |Wa-Wm| is calculated. If the deviation is found to exceed the acceptable error range ε ​​(assuming ε = 0.5 mm), we proceed to the next step of calibration. Fine-tune the optical measurement tool, such as calibrating the wavelength λ of the laser or correcting the scale calibration parameter s of the instrument. Use the following formula to calibrate the optical measurement equipment: Wc = (Wm / s)*λ, where Wc is the measured value after calibration.

[0120] Specifically, after the device is calibrated, the black border width Wv of the display is measured again, and the same method is used to evaluate the accuracy of the calibrated device by calculating the new deviation ΔW' between Wv and Wm to ensure that ΔW' ≤ ε. These corrected accurate data Wv are used to optimize the image processing algorithm. For example, the threshold parameters in the Canny algorithm are adjusted to accurately simulate the actual edge width to ensure that the camera is set correctly.

[0121] Specifically, if the edge width We' obtained after adjusting the threshold matches the corrected true measurement Wv (i.e., We'*ρ≈Wv), the image processing technique is considered to be optimized successfully. Finally, after the camera is accurately set up, fine-tuning is performed to obtain the desired shooting effect, such as changing the aperture size to adjust the image exposure, or adjusting the focal length to optimize the image clarity. After completing these adjustments, a final test is performed.

[0122] Specifically, the same optical measurement tool is used to measure the width of the black border of the screen again to verify whether the on-site effect is consistent with the optimized model analysis and the visual effect data captured by the camera. For example, the deviation ΔW” of the final black border width Wf and Wv is evaluated to ensure that ΔW” is not greater than ε.

[0123] S3: comparing the real measurement data with the three-dimensional simulation data, and adjusting the shooting parameters of the camera according to the deviation result between the real measurement data and the three-dimensional simulation data, wherein the shooting parameters include a preset position and a shooting angle;

[0124] In an optional embodiment, the comparing the real measurement data with the three-dimensional simulation data and adjusting the shooting parameters of the camera according to the deviation result between the real measurement data and the three-dimensional simulation data, wherein the shooting parameters include a preset position and a shooting angle, comprises:

[0125] S301: Analyze the difference between the three-dimensional simulation data and the real measurement data to obtain a deviation result;

[0126] S302: formulating a data adjustment strategy through deviation result analysis to reduce the perspective error between the virtual three-dimensional model prototype and the actual physical black edge;

[0127] S303: After obtaining the data adjustment strategy, automatically adjust the preset position of the camera so that the spatial positioning of the camera during shooting meets the model adjustment standard;

[0128] S304: determining whether the preset position of the camera meets the requirements of the three-dimensional simulation data, and if so, adjusting the shooting angle of the camera to obtain the best shooting posture;

[0129] S305: After adjusting the shooting angle, continue to evaluate the effect of the new shooting parameters through comparative analysis to ensure that the perspective error is minimized;

[0130] S306: When the effect brought by the new shooting parameters in S305 still has deviations, the correction process of steps S302-S305 is looped until the preset model adjustment standard is reached;

[0131] S307: When the deviation is controlled within an acceptable range, the new shooting parameter setting is saved as a part of the shooting plan to provide a basis for subsequent shooting tasks.

[0132] Exemplarily, a 3D model consistent with the real scene is generated using 3D simulation software (such as Blender or Maya), and the model point cloud data is output for use by the machine learning model. These point cloud data exist in XYZ coordinate format to ensure dimensional consistency with the optical measurement data, and are kept consistent with the format of the machine learning algorithm input through a conversion script, for example, a JSON or CSV file.

[0133] Specifically, the MLP model is used to match and compare the 3D simulation data and the optical measurement data to find the differences. By calculating the Euclidean distance between the two sets of data point clouds, we can accurately quantify their differences. For example, if the Euclidean distance between point A shown in the simulation data and the corresponding point A' in the optical measurement data is 0.5 mm, then it is determined that there is an error at this point.

[0134] Specifically, using the error analysis results obtained in the above steps, an optimization algorithm, such as a genetic algorithm, is used to adjust the camera preset position and shooting angle to find the configuration with the minimum error. Assuming that the initial angle is set to (θx, θy, θz), the genetic algorithm quickly explores the appropriate angle setting through an evolutionary strategy, and the fitness function of each generation is evaluated according to the error size until the angle configuration with the minimum error is found. Once the optimal configuration is obtained, the automated control finely adjusts the camera pose based on this data. This control can be based on machine vision feedback, adopting a proportional-integral-differential (PID) control strategy to adjust the camera position in real time until the predetermined confidence level position is reached. After the camera pose is adjusted, a new round of optical measurements will be performed, and the measurement data will be input into the machine learning model to evaluate the degree of improvement in perspective error.

[0135] Specifically, if the error range of the measured data is within an acceptable threshold (e.g., the error is less than 0.1 mm), it can be considered that the error correction has achieved the expected effect. Finally, the perspective parameters are updated to the 3D simulation software, and new simulation data is regenerated by the software and compared with the optimized optical measurement data again. This comparison will use similar methods as before, such as Euclidean distance calculation.

[0136] Specifically, if the error in a new round is reduced again, for example, the average error is reduced to within 0.05 mm, it means that the measurement and simulation process has reached a more optimized state.

[0137] S4: acquiring and processing the pre-shot image, performing quantitative evaluation on the consistency of the black border, and optimizing the actual shooting configuration of the camera according to the result of the quantitative evaluation, wherein the actual shooting configuration includes an actual shooting position and an actual shooting angle;

[0138] In an optional embodiment, the acquiring and processing of the pre-shot image, performing a quantitative evaluation on the consistency of the black border, and optimizing the actual shooting configuration of the camera according to the result of the quantitative evaluation, wherein the actual shooting configuration includes an actual shooting position and an actual shooting angle, comprises:

[0139] S401: loading a pre-shot photo, identifying the outline of the black edge through a specially designed algorithm, and obtaining edge information of the black edge;

[0140] S402: performing geometric transformation operator processing on the obtained edge information to determine the actual width of the black edge and obtain an edge width value set;

[0141] S403: Calculating the standard deviation and the average value of the edge width value set by a statistical method, and presetting a quantization parameter, thereby quantitatively evaluating the consistency of the black edge and obtaining a quantization consistency result before adjustment;

[0142] S404: using machine learning technology to analyze the correlation between the quantization parameters and the actual visual effect, obtaining a predicted shooting configuration through a training model, simulating the predicted shooting configuration through a simulator, and simulating a simulated visual effect for the predicted shooting configuration, wherein the preset shooting configuration includes a shooting position and an angle configuration;

[0143] S405: Generate an optimization report by comparing the simulated visual result with the actual visual effect, wherein the optimization report includes recommended parameters for the shooting configuration, and the report is used to explain in detail the specific effects of the shooting position and angle on the consistency of the black edge;

[0144] S406: Automatically adjust the shooting position and shooting angle of the camera according to the recommended parameters in the optimization report, and accurately perform the adjustment process through a control algorithm;

[0145] S407: After the adjustment is completed, retake the photo of the display screen to obtain a new photo, and use image analysis software to evaluate the consistency of the black border of the new photo to obtain a quantitative consistency result after adjustment;

[0146] S408: Compare the quantization consistency result before the adjustment with the quantization consistency result after the adjustment to determine whether the continuity of the visual effects of the pre-taken photo and the new photo is improved, and update the machine learning model according to the comparison result to optimize future prediction performance.

[0147] Exemplarily, based on edge detection algorithms, such as the Canny operator, the pre-shot display screen photo is first processed to extract pixel information of the edge of the display screen in the picture. The algorithm identifies the boundary between the black border and the display area by calculating the gradient strength and direction of the image. Then, the black border edge information is input into a geometric transformation model, such as using the Hough transform, to identify the straight line equation of the edge, thereby calculating the distance D between two parallel straight lines, which is the width of the black border.

[0148] Furthermore, the obtained multiple width values ​​D are analyzed by statistical methods. Generally, their mean μ and standard deviation σ are calculated, such as μ = (ΣDi) / n and σ = sqrt((Σ(Di-μ)2) / n), where Di represents a single measurement value and n represents the total number of measurements. Through these statistics, the consistency and distribution range of the black edge are quantitatively described.

[0149] Specifically, a linear regression analysis method is used to analyze the relationship between the statistical results and the observed visual effects, and a model Y = aX + b is established, where Y may represent the visual satisfaction score, X is the mean and / or standard deviation of the black border width obtained by the above-mentioned geometric transformation and statistical methods, and a and b are model parameters obtained by regression analysis of the training data set. In the simulation, small changes in the camera position and angle, such as the adjustment of Δx, Δy and Δθ in three-dimensional space, will be simulated to observe how they affect the visual effect of the black border of the display in the photo. Then, the simulated visual effect is comprehensively compared with the actual shooting results to test how to fine-tune the shooting parameters to reduce the visual difference. This comparison process produces an optimization report that summarizes the effect of improving the consistency of the black border by, for example, fine-tuning the individual units of the camera position by Δx = 1mm or adjusting the angle by Δθ = 0.5°. The optimization report will guide the adjustment of the camera configuration, automatically adjust its physical position and angle, and use information technology such as image recognition and robot control to achieve precise control of the shooting process and improve the final image.

[0150] In actual operation, the robot arm or automated guide rail can move accurately to λx, λy, and λθ according to the prediction results, corresponding to the predicted optimal x-axis position, y-axis position, and rotation angle, respectively. After the new shooting position and angle are set, the improvement in visual effect is verified by reshooting and evaluating the consistency of the black edge. For example, by comparing the edge width mean μ1 and μ2 before and after optimization, as well as the standard deviation σ1 and σ2, the improvement in consistency is evaluated, and the machine learning model parameters are improved accordingly to more accurately predict the optimal camera configuration in the future. If the adjusted standard deviation σ2 shows a significant decrease compared to σ1, it indicates that the consistency of the black edge width has been enhanced, thus verifying the effectiveness of the optimization process.

[0151] S5: adjusting ambient light parameters in combination with the physical size data and curvature parameters of the display screen to reduce the impact of reflection and shadow on image quality;

[0152] In an optional embodiment, the adjusting the ambient light parameters in combination with the physical size data and curvature parameters of the display screen to reduce the impact of reflection and shadow on image quality includes:

[0153] S501: Obtaining physical size data of the display screen, and calculating the optimal viewing distance and viewing angle of the display in combination with curvature parameters;

[0154] S502: Detecting image quality problems existing on the current display screen, where the image quality problems include reflection problems and shadow problems;

[0155] S503: Acquire ambient light parameters and perform preliminary evaluation, wherein the ambient light parameters include light source type, brightness and color temperature;

[0156] S504: adjusting the ambient light parameters according to the curvature parameter and the physical size data of the display screen, and the data fed back by the image analysis software, and obtaining an adjusted image effect to reduce reflections and shadows caused by direct irradiation or reflection of the light source;

[0157] S505: Detect the adjusted image effect again to determine whether there are still reflection problems and shadow problems. If yes, dynamically adjust the ambient light parameters until the predetermined standard is met;

[0158] S506: setting the ambient light parameters when the shadow problem and the reflection problem are minimized as ideal ambient light parameters, and applying the ideal ambient light parameters to the ambient light source to maintain the ambient light source;

[0159] S507: Continuously monitor the image quality problem of the display screen and the ambient light source, and dynamically adjust the ambient light parameters according to the image quality problem and the ambient light source to maintain an optimal display effect of the display.

[0160] For example, first, consider adaptive brightness adjustment, which detects the surrounding brightness through an ambient light sensor and adjusts the brightness of the display screen in real time. Specifically, a simple linear relationship can be used to map the ambient brightness (Ev) and the screen brightness (Sv): Sv = k*Ev+b, where k and b are parameters adjusted according to the screen characteristics and user experience.

[0161] Specifically, if the ambient brightness Ev is measured to be 250 lux, the screen brightness will be adjusted according to the set k value (assuming it is 0.8) and b value (assuming it is 10), calculated as Sv = 0.8*250+10 = 210 nits. In this way, the display can maintain a comfortable browsing brightness regardless of changes in the external environment.

[0162] Next, the color is corrected to ensure a consistent color experience under different ambient light. By using color management (CMS), a color profile is set to change the display parameters according to the ambient color temperature. The color correction coefficients of the red, green and blue (RGB) channels at a specific color temperature are calculated using the CIE1931 color space to ensure accurate colors. If the ambient color temperature is detected to be 6500K, the color correction coefficients adjust the RGB values ​​accordingly to compensate for the influence of ambient light.

[0163] Specifically, dynamic contrast optimization focuses on enhancing the image quality of the display screen, using algorithms such as local adaptive histogram equalization (LAHE) to improve contrast while preventing loss of image details. This algorithm changes the contrast of local areas of the image to obtain a more balanced contrast overall. Assuming that the pixel value range of the original image is 0-255, in a certain local area, by adjusting the histogram distribution, the pixel values ​​of the area are better distributed in the entire range, thereby improving the contrast in the area.

[0164] Furthermore, environmental noise suppression is an important part of ensuring visual clarity in complex environments, mainly referring to reflected light or other interference factors in the environment near the display device. By coating the surface of the display screen with a reflective and anti-glare coating, light reflection can be effectively reduced. The design of the coating uses Fourier transform (FT) to analyze the light waves and determine the coating thickness to minimize reflections within a specific wavelength range. When the wavelength of light is 550 nanometers (the center wavelength of green light), the coating thickness is designed to be one-quarter of the wavelength, that is, 137.5 nanometers, to form destructive interference, which greatly reduces the reflection of green light.

[0165] S6: acquiring image data, and repeatedly testing and finely adjusting the height parameter and tilt angle parameter of the camera according to the image data until obtaining shooting parameters with the best consistency of black border width;

[0166] In an optional embodiment, the acquiring of image data, and repeated testing and fine-tuning of the height parameter and tilt angle parameter of the camera according to the image data until the shooting parameters with the best consistency of black border width are obtained include:

[0167] S601: Setting the height parameters and tilt angle parameters of the initial camera, and automatically adjusting the camera to a preset starting position;

[0168] S602: Obtain image data through an image acquisition sensor, determine whether the width of the black border displayed by the image data is uneven, and if so, gradually adjust the height of the camera until the black border width is close to a preset range, thereby obtaining image data after the first adjustment;

[0169] S603: judging whether the best consistency of black border width is achieved according to the image data after the first adjustment, and if not, automatically adjusting the tilt angle parameter according to the image data to obtain the image data after the second adjustment;

[0170] S604: evaluating again whether the optimal consistency of the black border width is achieved according to the image data after the second adjustment, and if so, setting the height parameter and the tilt angle parameter as a shooting parameter; if not, entering the next round of adjustment;

[0171] S605: Using a statistical method, recording and analyzing multiple sets of image data, and establishing a correlation model between different height parameters of the camera and different tilt angle parameter adjustments;

[0172] S606: In subsequent shooting tasks, shooting parameters when different height parameters and different tilt angle parameters are combined are set in advance according to the correlation model to reduce the number of trial and error times;

[0173] S607: According to the shooting parameters, the tilt angle parameter of the camera is automatically adjusted to a value that matches the height parameter, and the camera automatically executes the shooting parameters each time shooting is performed.

[0174] Exemplarily, an initial camera height is set, such as 1.5 meters. At this time, the camera angle is also preset to an initial value, such as 15 degrees. This process can use a motor encoder to track the exact position of the camera and stop when it reaches the predetermined position. Next, a starting image is collected for quality analysis. Assuming that the black border width of the analyzed starting image is uneven, the algorithm uses an iterative adjustment method, such as a simple linear search.

[0175] Specifically, the height is adjusted in increments of 0.01 meters each time, and each time the height is moved up or down a certain amount, the image is taken and analyzed again. This incremental adjustment and image analysis is a cyclic process. A target tolerance range for the black border width is set, such as 0.5mm, and the cycle is repeated until the black border width difference is less than or equal to this tolerance. If the problem is not successfully solved after the height adjustment, the tilt angle is adjusted.

[0176] Specifically, an increment of the tilt angle is first defined, for example, 0.2 degrees is adjusted each time. Similarly, an iterative method is adopted to control the servo motor to adjust the camera angle, and a new image is captured and the black border width is analyzed after each adjustment until an angle is found that makes the black border consistency reach the target tolerance range.

[0177] Specifically, after these fine adjustments are completed, a new image data set is acquired and an image processing algorithm is used to detect whether the black border width is consistent with the specified standard. This process may use an edge detection algorithm, such as Canny edge detection, to extract the edges of the image and calculate the difference between the edge widths through a mathematical formula. Assume that the camera height is now adjusted to 1.52 meters and the tilt angle is adjusted to 15.6 degrees, and the image black border width consistency reaches the target standard. This set of parameters is stored as the standard shooting process parameters.

[0178] In order to further optimize future shooting tasks, multiple sets of image data are collected and analyzed. Regression analysis methods can be used to build a model that can express the statistical relationship between the influence of height and tilt angle on the consistency of image black edges. For example, a multivariate regression model is constructed using the least squares method to find the best combination of different parameters.

[0179] Specifically, in the future, when performing shooting tasks, the ideal camera height and tilt angle can be pre-set based on the statistical model built previously. If the model prediction results show that 1.53 meters height and 15.4 degrees tilt angle are close to ideal shooting parameters, these parameters will be used for new tasks to reduce the number of adjustments and improve efficiency.

[0180] S7: Correcting the pre-captured image, including correcting the perspective distortion of the four corners of the display screen using a digital deformation technique, and adjusting the graphic ratio of the image data and the symmetry of the black border according to the edge data collected during the repeated tests;

[0181] In an optional embodiment, the correcting the pre-captured image includes correcting the perspective distortion of the four corners of the display screen using a digital deformation technology, and adjusting the graphic ratio of the image data and the symmetry of the black border according to the edge data collected during the repeated experiments, including:

[0182] S701: Acquire perspective distortion parameters in the captured image, use digital deformation technology to perform perspective distortion correction operations, correct the distorted graphics at the four corners of the display screen, and obtain a corrected image;

[0183] S702: Based on the corrected image, an enhancement algorithm is called to refine the edge, improve the edge clarity of the image, and obtain an edge-enhanced image effect;

[0184] S703: Combining the edge data collected during the repeated experiments, adjusting the image scale of the edge enhanced image to obtain a scale-adjusted image, thereby ensuring the image scale accuracy;

[0185] S704: determining whether the symmetry of the proportion-adjusted graphic meets the expected requirement; if not, performing detailed correction on the asymmetric part of the black border to ensure the symmetry of the image;

[0186] S705: If there are still slight errors in the perspective of the image corrected in S701 after adjustment, the digital deformation supplementation technology is used again for adjustment, and the perspective distortion parameters are optimized according to the feedback of the adjustment result to achieve perfect perspective correction and improve the accuracy of processing perspective distortion;

[0187] S706: Reapply the optimized parameters in S701-S705 to ensure that the image effect after each correction is balanced and consistent;

[0188] S707: Determine whether the size of the scale-adjusted graphic in S703 meets the standard size. If not, reacquire the size data of the graphic, adjust the scale calibration algorithm and execute the graphic scale adjustment step again.

[0189] For example, when processing the perspective distortion of the image captured by the display screen, firstly, the straight line features in the image are analyzed by image processing software, such as Photoshop, to extract the parameters of the perspective distortion. For example, the perspective transformation is represented by a 3x3 matrix T, and for any point P(x, y) in the image, its perspective transformed position P'(x', y') can be calculated by P'=TP.

[0190] Specifically, after determining the perspective distortion parameters, digital deformation techniques such as bilinear interpolation are applied to correct the image. Bilinear interpolation considers the four closest pixels in the original image, and the calculated weights are used to generate the color values ​​of the new pixels, effectively correcting the distortion.

[0191] Specifically, after the image perspective correction is completed, the Canny edge detection algorithm is used to extract the edge of the corrected image, and then the edge is sharpened by enhancing the contrast and clarity. The Laplacian operator L is used as the algorithm core of edge enhancement, which can be expressed as a two-dimensional convolution kernel. The convolution operation of L and image I can be expressed as (I{enhancement}=I+k*(L*I)), where k is an adjustment factor used to control the degree of edge enhancement.

[0192] Further, on this basis, the symmetry of the black edges in the image is checked, and the image moment is used to calculate the image center of gravity. If the center of gravity is not located at the center of the image, it can be considered as insufficient symmetry. By relocating the center of the image and adjusting the black edges, the overall symmetry is ensured.

[0193] Specifically, for minor errors in perspective, further fine-tuning may be required after a rough adjustment. At this time, bilinear interpolation is used again, but with a finer distribution of control points and smaller weight adjustment parameters, to slightly modify the image to obtain a satisfactory final correction effect.

[0194] Specifically, by adjusting various parameters in the algorithm, such as adjusting the k value in edge definition enhancement, and precisely controlling the degree of rotation in symmetry adjustment, etc., the parameters are optimized to maximize the processing effect and ensure that each adjustment is close to the target result. For example, during fine-tuning, only 0.1-0.5 degree rotation correction or 1-5 pixel displacement correction may be required each time.

[0195] Specifically, for scale calibration, the conversion relationship between pixel size and actual size is used, such as 1 cm is represented as 100 pixels in the image, to ensure that the image size is consistent with the real world. This ratio can be modified, such as using transformation tools to enlarge or reduce the image according to the scale according to the actual situation. By iterating this process, after each correction, use a scale detection tool such as a ruler tool to determine whether the actual scale requirement has been achieved.

[0196] S8: Detecting the black border edge alignment and the black border width according to the corrected image data of the pre-shot image, and comparing them with the real measurement data, and performing subsequent image optimization to match the actual observation result.

[0197] In an optional embodiment, detecting the black border edge alignment and the black border width according to the image data of the corrected pre-shot image, comparing with the real measurement data, and performing subsequent image optimization to match the actual observation result includes:

[0198] S801: extracting image edge features using an edge detection algorithm according to the corrected image data of the pre-shot image, and obtaining position information and accuracy requirements of the image edge;

[0199] S802: Automatically detecting the black edge width and the black edge alignment in the pre-shot image according to the extracted image edge features, and obtaining black edge width data and black edge alignment deviation data;

[0200] S803: Obtaining the black border edge alignment and black border width under ideal conditions according to the real measurement data;

[0201] S804: comparing the black border width data and the black border edge alignment deviation data with the real measurement data to determine whether there is a deviation, and if so, using an algorithm to calculate an image correction amount;

[0202] S805: adjusting the image data according to the image correction amount and performing edge alignment correction;

[0203] S806: After the correction, the black edge width is detected again using a high-precision edge detection algorithm to obtain corrected image data; judging whether the corrected image data is consistent with the real measurement data according to the detection result, and if so, determining that the image has met the optimization requirements;

[0204] S807: continuously monitor the black border width data and the black border edge alignment deviation data of the image and the comparison deviation value with the true measurement data, and execute S804-S806 in a loop.

[0205] For example, a high-precision edge detection algorithm, such as the Canny algorithm, is applied to the image. This algorithm uses different filters to detect horizontal, vertical, and diagonal edges in the image, extracts edge features by setting appropriate thresholds (such as upper and lower thresholds of 0.09 and 0.17, respectively), and calculates the direction and magnitude of the gradient to obtain detailed information about the edge position.

[0206] Furthermore, the extracted edge features are passed to image processing software, such as OpenCV, and contour tracking algorithms such as findContours() are used to identify the boundaries of the image, and size measurements are performed to automatically identify the black border width and edge alignment. For example, the black border width is detected to be 2.5 pixels. Subsequently, the automatically detected black border width of 2.5 pixels is compared with the ideal width of 1.8 pixels, and the deviation between the two is evaluated to be 0.7 pixels. If there is an edge alignment deviation, such as an edge deviation of 0.4 mm from the ideal position, a geometric transformation algorithm (such as an affine transformation) is used to calculate the distance to be moved, and the deviation is corrected. By performing the correction operation, the image processing software applies an algorithm to transform the image to achieve precise edge alignment.

[0207] Specifically, the corrected image will be processed again by the same edge detection algorithm Canny to verify the results. At this time, it should be observed that the black border width is close to or equal to the ideal value of 1.8 pixels. If the corrected detection result shows that the black border width is indeed 1.8 pixels, it means that the image correction is successful and the image quality optimization has achieved the expected effect.

[0208] Specifically, the image correction process is automatically updated to ensure that it meets the observation standards. Automated monitoring will record and analyze the data of each detection. If the deviation detected in long-term monitoring exceeds the preset tolerance range (for example, the black edge width deviation exceeds 0.2 pixels), the correction process will be automatically triggered.

[0209] Specifically, models such as regression analysis in machine learning or neural network training are used to learn the relationship between past correction parameters and actual effects, dynamically adjust future correction algorithm parameters, and achieve self-optimization. Through this continuous process, high-precision monitoring and automatic correction mechanisms ensure that image quality is always maintained at a high level to meet the strict standards set.

[0210] The present invention can obtain a display screen image with a consistent black border width by arranging a proper shooting angle and position of a camera, thereby reducing perspective distortion and visual differences caused by the device design itself.

[0211] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A screen size online detection method based on image analysis, characterized in that: The method comprises: S1: constructing a virtual three-dimensional model prototype of a display screen and a camera, simulating a moving trajectory and a zoom effect of the camera, and generating image previews at multiple angles and distances according to the moving trajectory and the zoom effect; S2: Acquire three-dimensional simulation data according to the virtual three-dimensional model prototype, and use an optical measurement tool to acquire real measurement data of the precise width of the actual black border of the display screen, wherein the real measurement data is used to correct the perspective error in the simulation process; S3: comparing the real measurement data with the three-dimensional simulation data, and adjusting the shooting parameters of the camera according to the deviation result between the real measurement data and the three-dimensional simulation data, wherein the shooting parameters include a preset position and a shooting angle; S4: acquiring and processing the pre-shot image, performing quantitative evaluation on the consistency of the black border, and optimizing the actual shooting configuration of the camera according to the result of the quantitative evaluation, wherein the actual shooting configuration includes an actual shooting position and an actual shooting angle; S5: adjusting ambient light parameters in combination with the physical size data and curvature parameters of the display screen to reduce the impact of reflection and shadow on image quality; S6: acquiring image data, and repeatedly testing and finely adjusting the height parameter and tilt angle parameter of the camera according to the image data until obtaining shooting parameters with the best consistency of black border width; S7: Correcting the pre-captured image, including correcting the perspective distortion of the four corners of the display screen using a digital deformation technique, and adjusting the graphic ratio of the image data and the symmetry of the black border according to the edge data collected during the repeated tests; S8: According to the image data of the corrected pre-shot image, the black border edge alignment and the black border width are detected, and compared with the real measurement data, and subsequent image optimization is performed to match the actual observation result.

2. The method for online screen size detection based on image analysis according to claim 1, characterized in that: The method of constructing a virtual three-dimensional model prototype of a display screen and a camera, simulating a moving track and a zoom effect of the camera, and generating image previews at multiple angles and distances according to the moving track and the zoom effect includes: S101: Designing with three-dimensional modeling software, setting model parameters of the display screen and the camera, and constructing a virtual three-dimensional model prototype of the display screen and the camera; S102: after the virtual three-dimensional model prototype is constructed, simulating the movement trajectory of the camera so that the simulated movement trajectory is close to the actual operation scenario, the movement trajectory including movement speed and angle change; S103: adjusting variables in the movement trajectory that affect image quality, wherein the variables include light and light reflection in a simulated environment; S104: simulating the zoom effect of the camera by the three-dimensional modeling software to obtain simulated images of the camera at different focal lengths, wherein the simulated images serve as a preset standard for subsequent testing and error correction; S105: simulating the camera shooting at multiple angles and multiple distances, determining the imaging performance of the camera at different angles and different distances, and obtaining an image preview effect according to the imaging performance; S106: comparing and analyzing the deviation between the image preview effect obtained under each set of angles and distances and the preset standard, and adjusting the virtual three-dimensional model prototype; S107: Verify the results of the entire simulation process and adjust the simulation parameters through feedback to optimize the image generation process.

3. The method for online screen size detection based on image analysis according to claim 1, characterized in that: The method of acquiring three-dimensional simulation data according to the virtual three-dimensional model prototype and using an optical measurement tool to acquire real measurement data of the precise width of the actual black border of the display screen, wherein the real measurement data is used to correct the perspective error in the simulation process, includes: S201: extracting three-dimensional simulation data of the visual effect of the virtual three-dimensional model prototype using image processing technology to obtain expected width information of the black border; S202: using a high-precision optical measurement tool to measure the actual black border, obtaining real measurement data of the precise width of the actual black border, and recording the actual width information of the actual black border; S203: Compare and analyze the extracted expected width information with the actual width information recorded by the high-precision optical measurement tool to determine whether a deviation occurs; if a deviation occurs, calibrate the high-precision optical measurement tool; S204: After the device is calibrated, the actual black border is measured to obtain new actual width information, and the actual width information is compared with the expected width information again; S205: Determine the deviation by comparing the data in S204, optimize the image processing technology, and use the three-dimensional simulation data extracted by the optimized image processing technology to guide the precise setting of the camera; S206: Determine the placement position of the camera to capture a complete image of the display screen; S207: Obtaining guidance data through the device calibration and the placement position of the camera, and fine-tuning the position of the camera according to the guidance data; S208: After fine-tuning, a final test is performed to measure the width of the black border using the high-precision optical measurement tool to obtain final actual width information, ensuring that the final actual width information is consistent with the expected width information of the virtual three-dimensional model prototype.

4. The method for online screen size detection based on image analysis according to claim 1, characterized in that: The comparing the real measurement data with the three-dimensional simulation data and adjusting the shooting parameters of the camera according to the deviation result between the real measurement data and the three-dimensional simulation data, wherein the shooting parameters include a preset position and a shooting angle, comprises: S301: Analyze the difference between the three-dimensional simulation data and the real measurement data to obtain a deviation result; S302: Formulate data adjustment strategies through deviation result analysis; S303: After obtaining the data adjustment strategy, automatically adjust the preset position of the camera so that the spatial positioning of the camera during shooting meets the model adjustment standard; S304: determining whether the preset position of the camera meets the requirements of the three-dimensional simulation data, and if so, adjusting the shooting angle of the camera to obtain the best shooting posture; S305: After adjusting the shooting angle, continue to evaluate the effect of the new shooting parameters through comparative analysis; S306: When the effect brought by the new shooting parameters in S305 still has deviations, the correction process of steps S302-S305 is looped until the preset model adjustment standard is reached; S307: When the deviation is controlled within an acceptable range, the new shooting parameter setting is saved as a part of the shooting plan.

5. The method for online detection of screen size based on image analysis according to claim 1, characterized in that: The acquiring and processing of the pre-shot image, performing a quantitative evaluation on the consistency of the black border, and optimizing the actual shooting configuration of the camera according to the result of the quantitative evaluation, wherein the actual shooting configuration includes an actual shooting position and an actual shooting angle, comprises: S401: loading a pre-shot photo, identifying the outline of the black edge through a specially designed algorithm, and obtaining edge information of the black edge; S402: performing geometric transformation operator processing on the obtained edge information to determine the actual width of the black edge and obtain an edge width value set; S403: Calculating the standard deviation and the average value of the edge width value set by a statistical method, and presetting a quantization parameter, thereby quantitatively evaluating the consistency of the black edge and obtaining a quantization consistency result before adjustment; S404: using machine learning technology to analyze the correlation between the quantization parameters and the actual visual effect, obtaining a predicted shooting configuration through a training model, simulating the predicted shooting configuration through a simulator, and simulating a simulated visual effect for the predicted shooting configuration, wherein the predicted shooting configuration includes a shooting position and an angle configuration; S405: Generate an optimization report by comparing the simulated visual result and the actual shooting visual effect, wherein the optimization report includes recommended parameters for shooting configuration; S406: Automatically adjust the shooting position and shooting angle of the camera according to the recommended parameters in the optimization report, and accurately perform the adjustment process through a control algorithm; S407: After the adjustment is completed, retake the photo of the display screen to obtain a new photo, and use image analysis software to evaluate the consistency of the black border of the new photo to obtain a quantitative consistency result after adjustment; S408: Compare the quantization consistency result before the adjustment with the quantization consistency result after the adjustment, determine whether the continuity of the visual effects of the pre-taken photo and the new photo is improved, and update the machine learning model according to the comparison result.

6. The method for online detection of screen size based on image analysis according to claim 1, characterized in that: The step of adjusting the ambient light parameters in combination with the physical size data and the curvature parameters of the display screen to reduce the influence of reflection and shadow on the image quality includes: S501: Obtaining physical size data of the display screen, and calculating the optimal viewing distance and viewing angle of the display screen in combination with curvature parameters; S502: Detecting image quality problems existing on the current display screen, where the image quality problems include reflection problems and shadow problems; S503: Acquire ambient light parameters and perform preliminary evaluation, wherein the ambient light parameters include light source type, brightness and color temperature; S504: adjusting the ambient light parameter according to the curvature parameter and the physical size data of the display screen, and the data fed back by the image analysis software, to obtain an adjusted image effect; S505: Detect the adjusted image effect again to determine whether there are still reflection problems and shadow problems. If yes, dynamically adjust the ambient light parameters until the predetermined standard is met; S506: Setting the ambient light parameters when the shadow problem and the reflection problem are minimized as ideal ambient light parameters, and applying the ideal ambient light parameters to the ambient light source; S507: Continuously monitor the image quality problem of the display screen and the ambient light source, and dynamically adjust the ambient light parameters according to the image quality problem and the ambient light source to maintain an optimal display effect of the display screen.

7. The method for online detection of screen size based on image analysis according to claim 1, characterized in that: The acquiring of image data, and repeated testing and finely adjusting the height parameter and tilt angle parameter of the camera according to the image data until the shooting parameters with the best consistency of black border width are obtained include: S601: Setting the height parameters and tilt angle parameters of the initial camera, and automatically adjusting the camera to a preset starting position; S602: Obtain image data through an image acquisition sensor, determine whether the width of the black border displayed by the image data is uneven, and if so, gradually adjust the height of the camera until the black border width is close to a preset range, thereby obtaining image data after the first adjustment; S603: judging whether the best consistency of black border width is achieved according to the image data after the first adjustment, and if not, automatically adjusting the tilt angle parameter according to the image data to obtain the image data after the second adjustment; S604: evaluating again whether the optimal consistency of the black border width is achieved according to the image data after the second adjustment, and if so, setting the height parameter and the tilt angle parameter as shooting parameters; if not, entering the next round of adjustment; S605: Using a statistical method, recording and analyzing multiple sets of image data, and establishing a correlation model between different height parameters of the camera and different tilt angle parameter adjustments; S606: In a subsequent shooting task, shooting parameters when different height parameters and different tilt angle parameters are combined are set in advance according to the correlation model; S607: According to the shooting parameters, the tilt angle parameter of the camera is automatically adjusted to a value that matches the height parameter, and the camera automatically executes the shooting parameters each time shooting is performed.

8. The method for online detection of screen size based on image analysis according to claim 1, characterized in that: The correcting of the pre-captured image includes correcting the perspective distortion of the four corners of the display screen using digital deformation technology, and adjusting the graphic ratio of the image data and the symmetry of the black border according to the edge data collected during the repeated experiments, including: S701: Acquire perspective distortion parameters in the captured image, use digital deformation technology to perform perspective distortion correction operations, correct the distorted graphics at the four corners of the display screen, and obtain a corrected image; S702: Based on the corrected image, an enhancement algorithm is called to refine the edge, improve the edge clarity of the image, and obtain an edge-enhanced image effect; S703: combining the edge data collected in the repeated test process, performing graphic proportion adjustment on the edge enhanced image effect to obtain a proportion adjusted graphic; S704: determining whether the symmetry of the proportion-adjusted graphic meets the expected requirement, and if not, performing detailed correction on the asymmetric part of the black border; S705: If there is still a slight error in the perspective of the image corrected in S701 after adjustment, the digital deformation supplementation technology is used for adjustment again, and the perspective distortion parameters are optimized according to the feedback of the adjustment result; S706: Reapply the optimized parameters in S701-S705 to ensure that the image effect after each correction is balanced and consistent; S707: Determine whether the size of the scale-adjusted graphic in S703 meets the standard size. If not, reacquire the size data of the graphic, adjust the scale calibration algorithm and execute the graphic scale adjustment step again.

9. The method for online detection of screen size based on image analysis according to claim 1, characterized in that: The detecting the black border edge alignment and the black border width according to the image data of the corrected pre-shot image, comparing with the real measurement data, and performing subsequent image optimization to match the actual observation result, includes: S801: extracting image edge features using an edge detection algorithm according to the corrected image data of the pre-shot image, and obtaining position information and accuracy requirements of the image edge; S802: Automatically detecting the black edge width and the black edge alignment in the pre-shot image according to the extracted image edge features, and obtaining black edge width data and black edge alignment deviation data; S803: Obtaining the black border edge alignment and black border width under ideal conditions according to the real measurement data; S804: comparing the black border width data and the black border edge alignment deviation data with the real measurement data to determine whether there is a deviation, and if so, using an algorithm to calculate an image correction amount; S805: adjusting the image data according to the image correction amount and performing edge alignment correction; S806: After the correction, the black edge width is detected again using a high-precision edge detection algorithm to obtain corrected image data; judging whether the corrected image data is consistent with the real measurement data according to the detection result, and if so, determining that the image has met the optimization requirements; S807: continuously monitor the comparison deviation value between the black border width data and the black border edge alignment deviation data of the image and the real measurement data, and execute S804-S806 in a loop.

Images