Visual multi-screen splicing control method for multiple scenes

Abstract

The invention relates to the technical field of multi-screen splicing, in particular to a visual multi-screen splicing control method for multiple scenes, and the method comprises the steps: obtaining parameters before splicing; scene matching decision making; acquiring post-splicing parameters; diagnosing degradation parameters; generating an optimization strategy set; strategy execution and re-evaluation; adjusting a threshold value; and outputting a result. According to the method and the device, the multi-dimensional scene core parameters are intelligently matched through the preset scene template library before splicing, the core picture quality indexes are quantitatively utilized to perform parameter intelligent diagnosis and optimization strategy execution after splicing, and a dynamic feedback mechanism is started, so that the continuous self-optimization capability is improved; a complete technical system from scene perception and intelligent configuration to quality evaluation and closed-loop optimization is constructed, the intelligent level and the stability of the display effect in different complex application scenes are improved, and the problems of poor scene adaptability and poor splicing effect caused by configuration solidification and lack of intelligent regulation and control are effectively solved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of multi-screen splicing, and in particular to a visual multi-screen splicing control method for multiple scenarios. BACKGROUND

[0002] With the improvement of informatization level, multi-screen splicing display systems have become the core carriers of key information presentation and decision support. Most existing multi-screen splicing systems adopt static configuration strategies, and their parameters and processing logic are fixed after deployment, which cannot be dynamically adapted according to actual needs, and the scene switching flexibility is insufficient. In addition, the evaluation of display quality mostly relies on the lack of quantitative standards, which affects the accuracy and efficiency of information transmission. Under this background, how to autonomously complete configuration decision and performance optimization when facing complex and variable input and task requirements has become a technical challenge in the field of multi-screen splicing.

[0003] Chinese patent application publication No. CN120428939A discloses a visual multi-screen splicing control method for a conference room, which includes the following steps: step S1, multi-source input content dynamic segmentation: real-time recognition and segmentation of input content calculation characteristic value, analysis of content characteristics of each display area, and recognition of element distribution; step S2, multi-screen state matrix construction: using a distributed light sensor to collect the brightness value matrix L, color temperature matrix C, and response time matrix T of each screen, and constructing a real-time digital twin model of physical display characteristics; step S3, dynamic fusion weight calculation: based on the results of multi-source content segmentation and the data of the multi-screen state matrix, calculating the fusion weight, and determining the fusion area adjustment priority according to the content motion speed and screen difference; step S4, non-linear gradual correction: based on the dynamic fusion weight, using an improved Sigmoid transition function for correction, while maintaining smooth transition and suppressing the moire caused by the traditional S curve; step S5, distributed synchronous rendering: after correction, designing a time delay compensation algorithm, executing synchronous rendering through the time delay compensation algorithm, and ensuring multi-screen display synchronization; and step S6, closed-loop feedback optimization: establishing a parameter optimization model based on PID control, optimizing parameters using the PID model, and continuously iterating.

[0004] It can be seen that the visual multi-screen splicing control method for a conference room has the following problems: the method only performs content segmentation and processing based on low-level visual features, and cannot recognize higher-dimensional application scenario intentions; the method uses a classic PID control model for closed-loop optimization, which is difficult to effectively handle the highly nonlinear and multi-variable coupled dynamic characteristics in the multi-screen splicing system; and the synchronous rendering mechanism of the method mainly compensates for the downstream processing and transmission time delay, and it is difficult to completely eliminate the tearing or stuttering phenomenon of cross-screen dynamic content. SUMMARY

[0005] To this end, the application provides a multi-scene visual multi-screen splicing control method for overcoming the poor scene adaptability and poor splicing effect caused by configuration solidification and lack of intelligent regulation in the prior art by multi-parameter perception and intelligent matching of scenes and adaptive optimization.

[0006] To achieve the above-mentioned object, the application provides a multi-scene visual multi-screen splicing control method, comprising: real-time acquisition of resolution density values, scene switching rates of pictures and bit depth values of signal source main color spaces of a main signal source of a current scene to be spliced; determination of a plurality of scene matching degrees according to the resolution density values, the bit depth values, the scene switching rates and a preset scene template library, and determination of a target scene mode according to the scene matching degrees and a preset matching degree threshold value; multi-screen splicing based on the target scene mode, acquisition of adjacent screen color difference gradients of physical display units after splicing, multi-screen synchronization offset amounts of picture dynamic content, timestamp dispersion degrees of scene signal source image frame generation and pixel misregistration amounts of splicing seams; calculation of real-time splicing fidelity according to the adjacent screen color difference gradients, the multi-screen synchronization offset amounts, the timestamp dispersion degrees and the pixel misregistration amounts, and determination of a plurality of degraded display parameters according to a comparison result of the real-time splicing fidelity and a preset fidelity threshold value; determination of a composite optimization strategy set according to a type of the degraded display parameters; reacquisition of the real-time splicing fidelity by executing the composite optimization strategy set; adjustment of the preset matching degree threshold value according to a frequency characteristic of the real-time splicing fidelity reacquired within a preset adjustment time length; output of the target scene mode and the real-time splicing fidelity re-determined after adjustment of the preset matching degree threshold value.

[0007] Further, the process of determining a plurality of scene matching degrees according to the resolution density values, the bit depth values, the scene switching rates and a preset scene template library, and determining a target scene mode according to the scene matching degrees and a preset matching degree threshold value comprises: calculation of a standard deviation of the resolution density values within a preset observation time length to obtain a density fluctuation value; calculation of a standard deviation of the bit depth values within a preset observation time length to obtain a bit depth fluctuation value; calculation of a standard deviation of the scene switching rates within a preset observation time length to obtain a switching fluctuation value; obtaining of a parameter determination result according to a comparison result of the density fluctuation value and a preset density threshold value, a comparison result of the bit depth fluctuation value and a preset bit depth threshold value and a comparison result of the switching fluctuation value and a preset switching threshold value; Based on the parameter determination results, the resolution density value, the bit depth value, and the scene switching rate are input into the preset scene template library to determine several candidate applicable scenes, so as to obtain several scene matching degrees. Based on the comparison between the scene matching degree and the preset matching degree threshold, the target scene mode is determined from all the candidate applicable scenes.

[0008] Furthermore, the target scene mode is determined from all the candidate applicable scenes when the scene matching degree is greater than the preset matching degree threshold.

[0009] Furthermore, the process of calculating real-time splicing fidelity based on the adjacent screen color difference gradient, the multi-screen synchronization offset, the timestamp dispersion, and the pixel misalignment, and determining several degraded display parameters based on the comparison between the real-time splicing fidelity and a preset fidelity threshold, includes: The gradient degradation degree is calculated based on the adjacent screen color difference gradient, the preset gradient threshold, and the preset gradient tolerance upper limit; the offset degradation degree is calculated based on the multi-screen synchronization offset, the preset offset threshold, and the preset offset tolerance upper limit; the discrete degradation degree is calculated based on the timestamp dispersion, the preset dispersion threshold, and the preset discrete tolerance upper limit; and the misalignment degradation degree is calculated based on the pixel misalignment amount, the preset misalignment threshold, and the preset misalignment tolerance upper limit. The gradient degradation degree, the offset degradation degree, the discrete degradation degree, and the misalignment degradation degree are weighted and summed to obtain the initial comprehensive degradation value. Calculate the product of the maximum value among the gradient degradation degree, the offset degradation degree, the discrete degradation degree, and the misalignment degradation degree with a preset penalty coefficient to obtain the product penalty factor; The real-time splicing fidelity is determined based on the product penalty factor and the initial integrated degradation value; Based on the comparison results between the real-time splicing fidelity and the preset fidelity threshold, several degraded display parameters are determined according to the initial comprehensive degradation value, the adjacent screen color difference gradient, the multi-screen synchronization offset, the timestamp dispersion, the pixel misalignment amount, and their respective corresponding degradation degrees.

[0010] Furthermore, the real-time splicing fidelity is determined based on the product penalty factor and the initial comprehensive degradation value.

[0011] Furthermore, based on the comparison results between the real-time splicing fidelity and the preset fidelity threshold, the process of determining several degraded display parameters according to the initial comprehensive degradation value, the adjacent screen color difference gradient, the multi-screen synchronization offset, the timestamp dispersion, the pixel misalignment, and their respective corresponding degradation degrees includes: When the real-time splicing fidelity is less than the preset fidelity threshold, the parameter degradation diagnosis process is triggered. Based on the parameter degradation diagnosis process, the difference between the adjacent screen color difference gradient and the preset gradient threshold is calculated to obtain the gradient deviation, and the difference between the multi-screen synchronization offset and the preset offset threshold is calculated to obtain the synchronization deviation, and the difference between the timestamp dispersion and the preset dispersion threshold is calculated to obtain the dispersion deviation, and the difference between the pixel misalignment amount and the preset misalignment threshold is calculated to obtain the misalignment deviation. When the gradient deviation is positive, the ratio of the product of the gradient degradation degree and its corresponding weight to the initial comprehensive degradation value is calculated to obtain the gradient degradation contribution. When the synchronization deviation is positive, the ratio of the product of the offset degradation degree and its corresponding weight to the initial comprehensive degradation value is calculated to obtain the offset degradation contribution. When the discrete deviation is positive, the ratio of the product of the discrete degradation degree and its corresponding weight to the initial comprehensive degradation value is calculated to obtain the discrete degradation contribution. When the misalignment deviation is positive, the ratio of the product of the misalignment degradation degree and its corresponding weight to the initial comprehensive degradation value is calculated to obtain the misalignment degradation contribution. Several degradation display parameters are determined based on the gradient degradation contribution, the offset degradation contribution, the discrete degradation contribution, and the misalignment degradation contribution.

[0012] Furthermore, the process of determining several degradation display parameters based on the gradient degradation contribution, the offset degradation contribution, the discrete degradation contribution, and the misalignment degradation contribution includes: When the gradient degradation contribution is greater than a preset gradient contribution threshold, the adjacent screen color difference gradient is determined to be the degraded display parameter; when the offset degradation contribution is greater than a preset offset contribution threshold, the multi-screen synchronization offset is determined to be the degraded display parameter; when the discrete degradation contribution is greater than a preset discrete contribution threshold, the timestamp discreteness is determined to be the degraded display parameter; and when the misalignment degradation contribution is greater than a preset misalignment contribution threshold, the pixel misalignment is determined to be the degraded display parameter.

[0013] Furthermore, the process of determining the composite optimization strategy set based on the type of the degradation display parameters includes: When the color difference gradient between adjacent screens is equal to the degraded display parameter, the optimization strategy is determined to be a color mapping recalibration strategy. When the multi-screen synchronization offset is equal to the degraded display parameter or the timestamp dispersion is equal to the degraded display parameter, the optimization strategy is determined to be the frame buffer adjustment strategy. When the pixel misalignment is equal to the degraded display parameter, the optimization strategy is determined to be a geometric deformation compensation strategy. The optimization strategies are combined into a composite optimization strategy set.

[0014] Furthermore, the process of adjusting the preset matching degree threshold based on the frequency characteristics of the real-time stitching fidelity re-acquired within the preset adjustment period includes: The frequency at which the real-time stitching fidelity is reacquired within the preset adjustment period is counted to obtain the acquisition frequency; When the acquisition frequency is greater than a preset frequency threshold, the preset matching degree threshold is increased according to the relative deviation between the acquisition frequency and the preset frequency threshold.

[0015] Furthermore, the process of outputting the target scene mode and the real-time stitching fidelity after adjusting the preset matching threshold includes: The redefined target scene mode and the real-time stitching fidelity are output as the current adapted scene and scene display quality score.

[0016] Compared with existing technologies, the advantages of this invention lie in its ability to rapidly and automatically identify multiple scenarios by acquiring three directly quantifiable signal source metadata parameters—resolution density, scene switching rate, and color bit depth—of the scene to be spliced ​​in real time and intelligently matching them with a preset scene template library. Simultaneously, by quantifying the core quality indicators after splicing and intelligently diagnosing degradation parameters based on fidelity scores, generating and executing targeted optimization strategies, a complete technical system is constructed, encompassing scene perception, intelligent configuration, quality assessment, and closed-loop optimization. Through a dynamic feedback mechanism, the system possesses continuous self-optimization capabilities, improving the intelligence level, display consistency and stability, and overall operational efficiency of the multi-screen splicing system in various complex application scenarios. This effectively solves the problems of poor scene adaptability and unsatisfactory splicing effects caused by fixed configurations and a lack of intelligent control.

[0017] Furthermore, by calculating the standard deviation of each feature parameter within the preset observation window and comparing it with the preset fluctuation threshold, a dynamic data confidence assessment layer is constructed. This layer can effectively filter out non-steady-state noise caused by transient signal interference, interactive random fluctuations, or scene transition periods, ensuring that the data input to the template library matching engine represents a statistically stable scene state. This significantly improves the confidence of scene pattern recognition and the timeliness of decision-making from the data source, avoiding pattern misjudgment and oscillation switching caused by brief parameter jumps. This provides highly consistent contextual input for subsequent fidelity assessment and adaptive optimization, enhancing the robustness of closed-loop control and overall decision-making efficiency.

[0018] Furthermore, by introducing a preset matching degree threshold as a decision threshold, the scene matching is improved from relatively optimal to absolutely reliable, ensuring that the determined target scene pattern has sufficient confidence, effectively avoiding configuration errors caused by low-quality matching, providing a stable and reliable decision basis for subsequent splicing parameter optimization, and significantly enhancing the determinism and reliability of the overall output.

[0019] Furthermore, by comparing the four display quality parameters with preset thresholds and tolerance limits, and normalizing them into a unified degradation index, the quantitative characterization and calibration of splicing defects in each dimension were achieved. Based on this, through dynamic weight allocation and weighted fusion based on scene modes, the comprehensive evaluation results can accurately reflect the quality preferences under different application scenarios, such as real-time synchronization in command scenarios and static image quality in exhibition scenarios. Overall, through a quantitative evaluation framework based on multi-dimensional normalized degradation and context-aware weight adaptation, accurate and interpretable comprehensive performance status indicators are provided, effectively improving the intelligence level and user experience of splicing displays in complex application environments.

[0020] Furthermore, by quantifying multi-dimensional degradation parameters into quality assessment indicators, accurate evaluation and visualization of splicing quality are achieved, providing data support for the formulation of subsequent optimization strategies. In addition, combined with the dynamic adjustment mechanism of the product penalty factor C, the evaluation weights can be adaptively adjusted according to the core needs of different scenarios, ensuring that key quality indicators are prioritized in all types of scenarios. This improves the intelligence level and scenario adaptability of the multi-screen splicing system, effectively solving the problem that traditional static evaluation methods struggle to cope with dynamic changes in complex scenarios.

[0021] Furthermore, a two-tiered design of deviation triggering and contribution quantification enables precise attribution of display quality defects. By calculating the positive deviation between the measured values ​​of each quality parameter and the ideal threshold, objectively deteriorating indicators are identified, eliminating non-critical interference. Based on this, by quantifying the contribution of each defect, the diagnosis simultaneously reflects the objective severity of the defect and its relative impact on the current scenario evaluation. This allows for accurate identification of the dominant factors causing the overall fidelity decline, rather than simply listing all anomalies. This provides a basis for generating highly targeted optimization strategies, ensuring that optimization resources are prioritized for the core defects that have the most significant impact on overall perceived quality. This achieves an intelligent decision-making closed loop from anomaly detection to root cause localization, significantly improving the efficiency and convergence of the correction process.

[0022] Furthermore, by establishing differentiated decision thresholds for each type of display quality defect, refined and configurable management of degradation parameter identification is achieved. By transforming the originally comprehensive and continuous contribution values ​​into explicit binary diagnostic conclusions, independent judgment criteria that conform to the physical characteristics and scene tolerance are established for each quality dimension. This avoids misjudgments that may be caused by using a single global threshold, ensuring that the diagnostic results can capture core defects that have a substantial impact on the perceived quality of the current scene, while filtering out secondary fluctuations that have some contribution but have not yet reached the point where optimization intervention is necessary. This enhances the certainty, interpretability, and predictability of decision-making, making the final output set of degraded display parameters a precise, reliable, and easy-to-execute input instruction for the subsequent generation of composite optimization strategy sets, thereby ensuring the pertinence, efficiency, and overall stability of the adaptive optimization process.

[0023] Furthermore, by establishing a precise correspondence between defect types and optimization actions, upstream diagnostic results are directly converted into executable control commands, achieving a seamless intelligent decision-making closed loop from problem identification to correction implementation. By assigning the most targeted underlying optimization methods based on the physical nature and causes of different degradation parameters, it ensures that each optimization action precisely targets the root cause of a specific quality defect, rather than employing a general global adjustment, thereby significantly improving the correction efficiency and signal-to-noise ratio. Building upon this, by combining multiple single strategies into a composite optimization strategy set, it possesses the ability to process multiple concurrent defects in parallel or sequentially, enabling it to cope with complex mixed quality degradation scenarios in real-world applications, significantly enhancing the robustness and comprehensive problem-solving capabilities of the entire method in real-world, ever-changing environments.

[0024] Furthermore, by monitoring the frequency of fidelity reassessment, the overall stability is dynamically perceived. When the frequency is too high, it is determined that the current matching threshold may be too lenient, causing the scene mode to oscillate near the critical state. Accordingly, the matching threshold is automatically increased to enhance the strictness of the decision, thereby suppressing frequent mode switching caused by small fluctuations in parameters. This achieves an adaptive balance between rapid responsiveness and decision stability. It can adjust in a timely manner to maintain the splicing effect when the scene changes significantly, and avoid unnecessary decision oscillations and resource consumption when the scene is stable. This significantly improves the operating efficiency, adaptability and overall robustness in different application environments.

[0025] Furthermore, by re-executing scenario matching and quality assessment based on the updated preset matching threshold, it is ensured that the output scenario mode and fidelity score are the optimal results calculated based on the latest calibration parameters. This not only guarantees a high degree of consistency between the current adaptive scenario and actual requirements, but more importantly, it concretizes the dynamically adjusted intelligent decision results into an operable and monitorable explicit output. This allows the final effect of the entire adaptive control process to be accurately perceived and applied, thus completing the complete intelligent control link from "perception, analysis, optimization" to "decision, output," and consolidating the credibility and usability of the overall self-optimization. Attached Figure Description

[0026] Figure 1 This is a flowchart of the visualization multi-screen splicing control method for multiple scenarios in this embodiment; Figure 2 This is the logic diagram for determining whether the scene parameters are normal in this embodiment; Figure 3 This is a logic diagram for determining the target scene mode in this embodiment; Figure 4 The logic diagram for adjusting the preset matching degree threshold in this embodiment is shown. Detailed Implementation

[0027] To make the objectives and advantages of the present invention clearer, the present invention will be further described below with reference to embodiments; it should be understood that the specific embodiments described herein are merely for explaining the present invention and are not intended to limit the present invention.

[0028] Preferred embodiments of the present invention will now be described with reference to the accompanying drawings. Those skilled in the art should understand that these embodiments are merely illustrative of the technical principles of the present invention and are not intended to limit the scope of protection of the present invention.

[0029] Please see Figure 1 As shown, this is a flowchart of a visual multi-screen splicing control method for multiple scenarios in this embodiment. This embodiment provides a visual multi-screen splicing control method for multiple scenarios, including: The resolution density value of the main signal source of the current scene to be stitched, the scene switching rate of the image, and the bit depth value of the main color space of the signal source are obtained in real time. Several scene matching degrees are determined based on the resolution density value, the bit depth value, the scene switching rate, and the preset scene template library, and the target scene mode is determined based on the scene matching degree and the preset matching degree threshold. Based on the target scene mode, multi-screen splicing is performed to obtain the color difference gradient between adjacent screens of the spliced ​​physical display unit, the multi-screen synchronization offset of dynamic content of the picture, the timestamp dispersion of the scene signal source image frame generation, and the pixel misalignment of the splicing seam. The real-time splicing fidelity is calculated based on the adjacent screen color difference gradient, the multi-screen synchronization offset, the timestamp dispersion, and the pixel misalignment. Several degraded display parameters are determined based on the comparison between the real-time splicing fidelity and the preset fidelity threshold. Determine the set of composite optimization strategies based on the type of the degradation display parameters; Execute the composite optimization strategy set to re-acquire the real-time stitching fidelity; The preset matching degree threshold is adjusted based on the frequency characteristics of the real-time splicing fidelity re-acquired within the preset adjustment period. The output is the target scene mode and the real-time stitching fidelity that are redefined after adjusting the preset matching threshold.

[0030] In this embodiment, the visualization multi-screen splicing control method for multiple scenarios is applicable to intelligent scene adaptation before splicing and quality closed-loop evaluation after splicing, and is primarily geared towards three typical scenarios: command and control, conference presentations, and exhibitions. Each scenario has fundamentally different selections of key parameters due to its different core tasks and requirements. Command and monitoring scenarios require extremely high resolution density to handle massive amounts of information, including multiple video feeds, maps, and data dashboards. Furthermore, the screen may display multiple real-time monitoring feeds simultaneously, and frequent events cause view switching (e.g., camera switching or alarm screen switching), resulting in the highest scene switching rate. To prioritize the stability of concurrent processing of multiple high-bitrate signals, a medium bit depth is used. Conference scenarios, designed for clear display of the desktop, PPT slides, and video windows, also employ a medium resolution density. Screen changes primarily involve speaker switching or PPT page turning, which are relatively gentle and spaced far apart. Therefore, the scene switching rate is lower than that of command and monitoring. The use of a medium bit depth also balances image quality with broad compatibility with various devices. Exhibitions and displays, aiming to perfectly reproduce high-definition content and create an immersive experience, require a resolution density lower than command and monitoring but higher than conference scenarios. Additionally, their content often consists of long shots or smooth transitions, resulting in the lowest scene switching rate among the three scenarios to ensure a stable viewing experience. To achieve professional-grade visual performance, the bit depth is the highest among the three scenarios. By transforming these differentiated needs into quantifiable signal features, a closed-loop management system was achieved, encompassing the entire process from identification and configuration to optimization. This enabled precise adaptation to various types of image stitching and solved the core problem of poor scene adaptability.

[0031] This invention achieves intelligent scene perception and precise quantification of display quality by acquiring multi-dimensional parameters. The resolution density value of the main signal source refers to the total number of pixels in the image output by the signal source, which serves as the core content to be displayed. It is used to quantify the information density of the required display content and the fundamental requirements for image detail. It can be obtained by reading the EDID data block of the signal source or through the operating system's graphics card driver interface. The scene switching rate refers to the number of significant jumps in video content per unit time. The structural similarity index (SSIM) between grayscale images is calculated for consecutive video frames. When the SSIM value of two consecutive frames is lower than a fixed global value, a scene switching event is determined to have occurred. The number of such events occurring per second is counted, which is the scene switching rate within the window. In this embodiment, the global value is set to 0.7. The bit depth value of the main color space of the signal source refers to the number of data bits used by the signal source in each color channel of the main color space. It can be obtained by parsing the EDID data block of the signal source or directly reading it from the current color format configuration of the graphics card driver. The adjacent screen color difference gradient is a precise measurement of the smoothness of color and brightness transitions in overlapping or boundary areas between two adjacent physical display units. It is used by deploying on the screen edge. Color data is acquired using a high-precision imaging colorimeter or a calibrated industrial camera. The color difference between corresponding pixels on adjacent screens is calculated and their spatial gradient is analyzed. Multi-screen synchronization offset quantifies the temporal or spatial misalignment of the same dynamic content on different display units. This can be achieved by using a high-speed camera to capture the entire splicing screen and using feature point tracking algorithms such as KLT optical flow tracking to accurately measure the positional offset of feature points when they cross physical gaps in the screen. Frame generation timestamp dispersion measures the synchronization accuracy of the image frames generated from each signal source at the time source. It is one of the root causes of a sense of disjointedness or asynchrony and can be obtained through the graphics card driver interface or a professional acquisition card that supports precise time protocols. Pixel misalignment at the splicing seam detects the pixel-level image content misalignment at the splicing seam caused by physical installation deviations or image processing errors. This is achieved by displaying a precise checkerboard or line grid test pattern, using a calibrated high-resolution industrial camera to capture the splicing area, and using digital image correlation or sub-pixel edge detection algorithms to analyze the continuity of the test pattern lines on both sides of the splicing seam and calculate the sub-pixel level misalignment.

[0032] The preset matching threshold is a similarity threshold used to determine whether scene recognition is successful. It depends on the discrimination design of the scene template library and the requirements for matching accuracy in actual applications. It is usually set between 0.7 and 0.9. In this embodiment, it is set to 0.85, which can effectively filter out low-confidence mismatches and ensure the accuracy of target scene pattern determination. The preset fidelity threshold is the minimum quality standard for judging whether the real-time splicing effect is qualified. It depends on the tolerance and quality requirements of the display effect in the specific application scenario. It is usually set between 0.6 and 0.8. In this embodiment, it is set to 0.75, which can provide a clear and reasonable benchmark target for splicing quality optimization. The preset adjustment time is the length of the observation time window on which the effect of the optimization strategy is evaluated and the parameters are adaptively adjusted. It depends on the frequency of scene switching and the time required for stability convergence. It is usually set between 10 minutes and 30 minutes. In this embodiment, it is set to 15 minutes, which can ensure that there is enough data to evaluate the execution effect of the composite optimization strategy set, while avoiding excessively slow adjustment.

[0033] In this embodiment, the preset scene template library is a key knowledge base used by this method to intelligently map the real-time collected multi-dimensional scene feature parameters—resolution density value, bit depth value, and scene switching rate—to specific display modes. For the three types of feature parameters, the template library integrates multiple heterogeneous data sources during its construction, including but not limited to industry standards, historical sample data, device characteristics, and expert experience summaries. For resolution density values, the template library includes multiple ultra-high density parameter ranges from 4K×2K to 8K×4K for command and monitoring scenarios. Each range corresponds to the concurrent carrying capacity of different number of signal sources. Through the correspondence between resolution density and the number of carrying sources, it ensures that the content of all concurrent signal sources can be displayed clearly and synchronously, meeting the scenario's requirements for information density and real-time performance. The conference scenario covers the mainstream density range from 1080P to 4K, and is further subdivided into single-screen presentation (4K) and multi-window split-screen (1080P). The density thresholds for sub-modes such as [list of sub-modes] are designed to avoid wasting system resources by using excessively high resolution in single-screen presentations or causing blurry content due to insufficient resolution in multi-window split-screen presentations, thus achieving precise matching of resources and needs. The resolution density value range for exhibition and display scenarios is 4K (3840×2160) to 6K×3K (6144×3072) to adapt to its core needs of creating an immersive experience and achieving professional-grade visual performance, ensuring optimal clarity at different viewing distances. For bit depth values, a standard range of 8-10 bits is set for command and monitoring scenarios to meet basic color reproduction requirements while ensuring the efficiency of multi-channel signal processing. The conference scenario uniformly adopts an 8-bit bit depth reference value, compatible with the output formats of various laptops, projectors, and other access devices. The exhibition and display scenario uses 10-12 bit depth as the core parameter, achieving accurate presentation of HDR content by expanding the color gamut space. Regarding scene switching rates, command and control monitoring primarily involves frequent camera switching, while conference presentations mainly involve PPT slide turning and speaker switching. Therefore, the switching rate for command and control monitoring is higher than that for conference presentations. Exhibitions and displays mostly involve long shots and smooth transitions, resulting in the lowest switching rate among the three scenarios. Therefore, the scene switching rate for command and control monitoring is set to greater than 2 times / second, for conferences to 0.2-2 times / second, and for exhibitions to less than 0.2 times / second. Furthermore, the template library includes built-in parameter correlation verification rules. For example, when the resolution density value exceeds 6K×3K, it automatically triggers dynamic adjustment of the upper limit of the bit depth value to prevent the combination of high resolution and high bit depth from exceeding the system's processing capacity, ensuring the feasibility and stability of scene matching. The template library matches the current scene to be spliced ​​with the cosine similarity of the feature vectors of each template in the library based on the real-time input feature parameters of the current scene to be spliced. By calculating the directional similarity of the scene feature vectors, the similarity score between the scene to be spliced ​​and each template is obtained to obtain the scene matching degree. This enables the template library to adapt to the personalized usage habits of different users and the ever-changing display task requirements, fundamentally supporting the adaptive capability for multiple scenes.

[0034] By acquiring three directly quantifiable signal source metadata parameters in real time—resolution density, scene switching rate, and color bit depth—of the scene to be spliced, and intelligently matching them with a preset scene template library, rapid and automatic identification of multiple scene types is achieved. Simultaneously, by quantifying the core quality indicators after splicing, and intelligently diagnosing degradation parameters based on fidelity scores, and generating and executing targeted optimization strategies, a complete technical system is constructed, encompassing scene perception, intelligent configuration, quality assessment, and closed-loop optimization. Through a dynamic feedback mechanism, the system possesses continuous self-optimization capabilities, improving the intelligence level, display consistency and stability, and overall operational efficiency of the multi-screen splicing system in various complex application scenarios. This effectively solves the problems of poor scene adaptability and unsatisfactory splicing effects caused by fixed configurations and a lack of intelligent control.

[0035] Please see Figure 2 As shown, this is the logic diagram for determining normal scene parameters in this embodiment. In this embodiment, the process of determining several scene matching degrees based on the resolution density value, the bit depth value, the scene switching rate, and the preset scene template library, and determining the target scene mode based on the scene matching degree and the preset matching degree threshold, includes: Calculate the standard deviation of the resolution density values ​​within the preset observation period to obtain the density fluctuation value; Calculate the standard deviation of the depth values ​​within the preset observation period to obtain the depth fluctuation value; Calculate the standard deviation of the scene switching rate within the preset observation period to obtain the switching fluctuation value; When the density fluctuation value is less than a preset density threshold, the bit depth fluctuation value is less than a preset bit depth threshold, and the switching fluctuation value is less than a preset switching threshold, the parameter determination result is obtained. Based on the parameter determination results, the resolution density value, the bit depth value, and the scene switching rate are input into the preset scene template library to determine several candidate applicable scenes, so as to obtain several scene matching degrees. Based on the comparison between the scene matching degree and the preset matching degree threshold, the target scene mode is determined from all the candidate applicable scenes.

[0036] The preset observation duration is the length of the time window used to evaluate the stability of scene feature parameters. It depends on the minimum time interval between scene switching and the requirement for sensitivity to parameter fluctuations, and is usually set between 10 and 30 seconds. In this embodiment, it is set to 15 seconds, which can effectively capture the short-term fluctuation trend of parameters and avoid misjudgment due to instantaneous noise. The preset density threshold is the critical value for determining whether the resolution density value fluctuation is too large. It depends on the physical accuracy of the display unit and the common step of the signal source resolution switching, and is usually set between 0.1 PPI / DPP and 0.5 PPI / DPP. In this embodiment, it is set to 0.3 PPI / DPP, which can filter out small density jumps caused by instantaneous signal disturbances or measurement noise, and ensure that the resolution parameters input to the matching engine are within acceptable limits. The data is stable and reliable. The preset bit depth threshold is the critical value for determining whether the bit depth value fluctuation is too large. It depends on the bit width of the color processing channel and the bit depth difference of common color format switching. It is usually set between 0.5 bits and 2 bits. In this embodiment, it is set to 1.2 bits, which can effectively shield the slight bit depth fluctuation caused by automatic color space adaptation or instantaneous signal distortion, and ensure the input quality of color depth parameters. The preset switching threshold is the critical value for determining whether the scene switching rate fluctuation is too large. It depends on the system's tolerance to the stability of the signal source frame rate and the measurement accuracy. It is usually set between 2 fps and 6 fps. In this embodiment, it is set to 3.5 fps, which can prevent the scene mode switching caused by instantaneous frame rate jitter or non-steady signals during the scene transition period from affecting the data quality.

[0037] By calculating the standard deviation of each feature parameter within a preset observation window and comparing it with a preset fluctuation threshold, a dynamic data confidence assessment layer is constructed. This layer effectively filters out non-steady-state noise caused by transient signal interference, interactive random fluctuations, or scene transition periods. It ensures that the data input to the template library matching engine represents a statistically stable scene state, significantly improving the confidence of scene pattern recognition and the timeliness of decision-making from the data source. It avoids pattern misjudgment and oscillation switching caused by brief parameter jumps, thus providing highly consistent contextual input for subsequent fidelity assessment and adaptive optimization, and enhancing the robustness of closed-loop control and overall decision-making efficiency.

[0038] Please see Figure 3 As shown, it is a logic diagram for determining the target scene mode in this embodiment. In this embodiment, the target scene mode is determined from all the candidate applicable scenes when the scene matching degree is greater than the preset matching degree threshold.

[0039] By introducing a preset matching degree threshold as a decision threshold, the scene matching is improved from relatively optimal to absolutely reliable, ensuring that the determined target scene pattern has sufficient confidence, effectively avoiding configuration errors caused by low-quality matching, providing a stable and reliable decision basis for subsequent splicing parameter optimization, and significantly enhancing the determinism and reliability of the overall output.

[0040] Specifically, the process of calculating real-time splicing fidelity based on the adjacent screen color difference gradient, the multi-screen synchronization offset, the timestamp dispersion, and the pixel misalignment, and determining several degraded display parameters based on the comparison between the real-time splicing fidelity and a preset fidelity threshold, includes: The gradient degradation degree is calculated based on the adjacent screen color difference gradient, a preset gradient threshold, and a preset gradient tolerance upper limit, where D x =max(0, X-X1) / X0, where D x X is the gradient degradation degree, X is the color difference gradient between adjacent screens, X0 is the preset gradient threshold, and X1 is the preset gradient tolerance upper limit. The offset degradation degree is calculated based on the multi-screen synchronization offset, the preset offset threshold, and the preset offset tolerance upper limit, where D... y =max(0, Y-Y1) / Y0, where D y Y is the offset degradation degree, Y is the multi-screen synchronization offset, Y0 is the preset offset threshold, Y1 is the preset offset tolerance upper limit, and the discrete degradation degree is calculated based on the timestamp discreteness, the preset discrete threshold, and the preset discrete tolerance upper limit, where D z =max(0, Z-Z1) / Z0, where D z Z is the discrete degradation degree, Z is the timestamp discreteness, Z0 is the preset discrete threshold, Z1 is the preset discrete tolerance upper limit, and the misalignment degradation degree is calculated based on the pixel misalignment amount, the preset misalignment threshold, and the preset misalignment tolerance upper limit, where D U =max(0, U-U1) / U0, where D U U is the degree of misalignment degradation, U is the amount of pixel misalignment, U0 is the preset misalignment threshold, and U1 is the preset upper limit of misalignment tolerance. The gradient degradation degree, the offset degradation degree, the discrete degradation degree, and the misalignment degradation degree are weighted and summed to obtain the initial comprehensive degradation value. Calculate the product of the maximum value among the gradient degradation degree, the offset degradation degree, the discrete degradation degree, and the misalignment degradation degree with a preset penalty coefficient to obtain the product penalty factor; The real-time splicing fidelity is determined based on the product penalty factor and the initial integrated degradation value; Based on the comparison results between the real-time splicing fidelity and the preset fidelity threshold, several degraded display parameters are determined according to the initial comprehensive degradation value, the adjacent screen color difference gradient, the multi-screen synchronization offset, the timestamp dispersion, the pixel misalignment amount, and their respective corresponding degradation degrees.

[0041] The preset gradient threshold is the critical value for determining whether the color difference gradient between adjacent screens has reached an ideal state. It depends on the minimum perceptible difference in color between the human eye and the color gamut coverage capability of the display device, and is usually set between ΔE2 and ΔE3. In this embodiment, it is set to ΔE2.5, which ensures that the color transition between spliced ​​screens is perceived as natural and continuous under most viewing conditions. The preset gradient tolerance upper limit is the maximum value that the color difference gradient between adjacent screens is allowed to deteriorate. Exceeding this value will be considered as serious color distortion. It depends on the highest tolerance for color consistency in the application scenario, and is usually set between ΔE6 and ΔE10. In this embodiment, it is set to ΔE8, which can clearly define the boundary of unacceptable color deviation in extreme cases. The preset offset threshold is used to determine the multi-screen synchronization offset. Whether the ideal state is reached depends on the human eye's visual persistence time and the display refresh rate, and is usually set between 0.5 milliseconds and 2 milliseconds. In this embodiment, it is set to 1 millisecond to ensure that dynamic content moves across screens without any visually perceptible stuttering or jumping. The preset offset tolerance upper limit is the maximum value that multi-screen synchronization offset deterioration is allowed. Exceeding this value will cause obvious visual tearing. It depends on the maximum speed of the dynamic content and the viewing distance, and is usually set between 8 milliseconds and 16 milliseconds. In this embodiment, it is set to 12 milliseconds to define the threshold at which severe synchronization loss occurs and immediate intervention is required. The preset dispersion threshold is the critical value for determining whether the frame generation timestamp dispersion has reached the ideal state. It depends on the master clock accuracy and the performance of the clock synchronization protocol of each signal source. The preset timeout is typically set between 0.1 and 0.5 milliseconds; in this embodiment, it is set to 0.3 milliseconds to ensure that all signal sources remain highly synchronized on the underlying clock, providing a stable foundation for upper-layer synchronization. The preset discrete tolerance upper limit is the maximum allowable value for frame generation timestamp discreteness degradation. Exceeding this value will lead to synchronization problems that are difficult to correct through upper-layer buffering. It depends on the depth of the frame buffer and the capability of the synchronization correction algorithm, and is typically set between 5 and 15 milliseconds; in this embodiment, it is set to 10 milliseconds to identify a severe state where the underlying clock synchronization has failed. The preset misalignment threshold is the critical value for determining whether the pixel misalignment at the splicing seam has reached an ideal state. It depends on the pixel density of the display screen and the precision of the installation process, and is typically set between 0.1 and 0. Between 0.3 pixels, set to 0.2 pixels in this embodiment, ensures that the image continuity at the seam is not disrupted at standard viewing distances. The preset misalignment tolerance upper limit is the maximum allowable degradation of pixel misalignment at the seam. Exceeding this value will result in obvious image breakage or ghosting. It depends on the characteristics of the displayed content and the user's requirements for geometric accuracy, and is usually set between 1 and 2 pixels. In this embodiment, it is set to 1.5 pixels, which clearly delineates the boundary where geometric alignment is severely compromised and physical calibration or significant software compensation is necessary. The preset penalty coefficient is a multiplier factor used to amplify the impact of severe degradation of a single parameter on the overall score when calculating the final fidelity. It depends on the degree of emphasis this method places on the weakest link effect and is usually set between 0.3 and 0.Between 7 and 7, this embodiment differentiates the preset penalty coefficients to meet the needs of different scenarios: 0.7 for command and monitoring scenarios; 0.5 for meeting scenarios; and 0.3 for exhibition and display scenarios. This allows for precise highlighting of key indicator issues based on the differentiated needs of each scenario, enabling optimization strategies to address core quality defects in each scenario.

[0042] In this embodiment, during the weighted summation calculation of the gradient degradation, offset degradation, discrete degradation, and misalignment degradation, the weights used for each parameter are a set of coefficients dynamically retrieved from a preset weight configuration table according to the target scene mode. This set of coefficients includes gradient weight, offset weight, discrete weight, and misalignment weight, thereby matching the comprehensive evaluation result with the actual perception requirements of the scene. The gradient weight is denoted as t, the offset weight as r, the discrete weight as j, and the misalignment weight as w. Specifically, in command and monitoring mode, priority is given to ensuring the real-time and synchronous nature of information; therefore, the offset weight and discrete weight are relatively high, typically between 0.3 and 0.4, while the gradient weight and misalignment weight are relatively low, typically between 0.1 and 0.2. In this embodiment, the gradient weight is set to 0.15, the offset weight to 0.35, the discrete weight to 0.35, and the misalignment weight to 0.15. In exhibition and display mode, priority is given to ensuring the static image quality and geometric accuracy of the image; therefore, the gradient weight and misalignment weight are relatively high, typically between t and r. The values ​​are between 0.3 and 0.4, while the offset weight and discrete weight are lower, usually between 0.1 and 0.2 respectively. In this embodiment, the gradient weight is set to 0.35, the offset weight to 0.15, the discrete weight to 0.15, and the misalignment weight to 0.35. For the conference presentation mode, a more balanced weight is allocated based on its characteristics of balancing dynamic content and static details. In this embodiment, the gradient weight is set to 0.25, the offset weight to 0.25, the discrete weight to 0.25, and the misalignment weight to 0.25, or slightly biased on the basis of balance.

[0043] By comparing four display quality parameters with preset thresholds and tolerance limits, and normalizing them into a unified degradation index, the quantitative characterization and calibration of splicing defects in each dimension are achieved. Based on this, dynamic weight allocation and weighted fusion based on scene modes enable the comprehensive evaluation results to accurately reflect the quality preferences under different application scenarios, such as real-time synchronization in command scenarios and static image quality in exhibition scenarios. Overall, through a quantitative evaluation framework based on multi-dimensional normalized degradation and context-aware weight adaptation, accurate and interpretable comprehensive performance status indicators are provided, effectively improving the intelligence level and user experience of splicing displays in complex application environments.

[0044] Specifically, the real-time splicing fidelity is determined based on the product penalty factor and the initial comprehensive degradation value, where B = 1 / [1 + (C × V)], where B is the real-time splicing fidelity, C is the product penalty factor, and V is the initial comprehensive degradation value.

[0045] By quantifying multi-dimensional degradation parameters into quality assessment indicators, precise evaluation and visualization of splicing quality are achieved, providing data support for the formulation of subsequent optimization strategies. Furthermore, combined with a dynamic adjustment mechanism of the product penalty factor C, the evaluation weights can be adaptively adjusted according to the core needs of different scenarios, ensuring that key quality indicators are prioritized in all types of scenarios. This improves the intelligence level and scenario adaptability of the multi-screen splicing system, effectively solving the problem that traditional static evaluation methods struggle to cope with dynamic changes in complex scenarios.

[0046] Specifically, based on the comparison results between the real-time splicing fidelity and the preset fidelity threshold, the process of determining several degraded display parameters according to the initial comprehensive degradation value, the adjacent screen color difference gradient, the multi-screen synchronization offset, the timestamp dispersion, the pixel misalignment amount, and their respective corresponding degradation degrees includes: When the real-time splicing fidelity is less than the preset fidelity threshold, the parameter degradation diagnosis process is triggered. Based on the parameter degradation diagnosis process, the difference between the adjacent screen color difference gradient and the preset gradient threshold is calculated to obtain the gradient deviation, and the difference between the multi-screen synchronization offset and the preset offset threshold is calculated to obtain the synchronization deviation, and the difference between the timestamp dispersion and the preset dispersion threshold is calculated to obtain the dispersion deviation, and the difference between the pixel misalignment amount and the preset misalignment threshold is calculated to obtain the misalignment deviation. When the gradient deviation is positive, the ratio of the product of the gradient degradation degree and its corresponding weight to the initial overall degradation value is calculated to obtain the gradient degradation contribution, N. t =(t×D x ) / H, where N t This is the gradient degradation contribution, where H is the initial overall degradation value. When the synchronization deviation is positive, the ratio of the product of the offset degradation degree and its corresponding weight to the initial overall degradation value is calculated to obtain the offset degradation contribution. Where N... d =(r×D y ) / H, where N d This is the offset degradation contribution, and when the discrete deviation is positive, the ratio of the product of the discrete degradation degree and its corresponding weight to the initial comprehensive degradation value is calculated to obtain the discrete degradation contribution, where N L =(j×D z ) / H, where N LTo determine the discrete degradation contribution, and when the misalignment deviation is positive, the ratio of the product of the misalignment degradation degree and its corresponding weight to the initial comprehensive degradation value is calculated to obtain the misalignment degradation contribution, N. x =(w×D U ) / H, where N x It is the contribution of misalignment and degradation; Several degradation display parameters are determined based on the gradient degradation contribution, the offset degradation contribution, the discrete degradation contribution, and the misalignment degradation contribution.

[0047] Through a two-tiered design of deviation triggering and contribution quantification, accurate attribution of display quality defects is achieved. By calculating the positive deviation between the measured values ​​of each quality parameter and the ideal threshold, objectively deteriorating indicators are identified, eliminating non-critical interference. Furthermore, by quantifying the contribution of each defect, the diagnosis simultaneously reflects the objective severity of the defect and its relative impact on the current scenario evaluation. This allows for precise identification of the dominant factors causing the overall decrease in fidelity, rather than simply listing all anomalies. This provides a basis for generating highly targeted optimization strategies, ensuring that optimization resources are prioritized for the core defects that have the most significant impact on overall perceived quality. This achieves an intelligent decision-making closed loop from anomaly detection to root cause localization, significantly improving the efficiency and convergence of the correction process.

[0048] Specifically, the process of determining several degradation display parameters based on the gradient degradation contribution, the offset degradation contribution, the discrete degradation contribution, and the misalignment degradation contribution includes: When the gradient degradation contribution is greater than a preset gradient contribution threshold, the adjacent screen color difference gradient is determined to be the degraded display parameter; when the offset degradation contribution is greater than a preset offset contribution threshold, the multi-screen synchronization offset is determined to be the degraded display parameter; when the discrete degradation contribution is greater than a preset discrete contribution threshold, the timestamp discreteness is determined to be the degraded display parameter; and when the misalignment degradation contribution is greater than a preset misalignment contribution threshold, the pixel misalignment is determined to be the degraded display parameter.

[0049] The preset gradient contribution threshold is the lower limit for determining whether the color difference gradient between adjacent screens constitutes a major degradation parameter. It depends on the sensitivity of the current scene to color consistency and the optimization resource allocation strategy, and is typically set between 10% and 25%. In this embodiment, it is set to 18%, which effectively filters out color deviations that exceed the ideal value but have a negligible actual impact, ensuring that optimization resources focus on major color issues that significantly affect the viewing experience. The preset offset contribution threshold is the lower limit for determining whether the multi-screen synchronization offset constitutes a major degradation parameter. It depends on the scene's requirements for the real-time performance of dynamic content and the destructive nature of synchronization defects, and is typically set between 15% and 30%. In this embodiment, it is set to 22%, which accurately identifies core synchronization defects that cause tearing or stuttering in dynamic scenes, avoiding secondary timing jitter. Overreaction; The preset discrete contribution threshold is the lower limit for determining whether timestamp discreteness constitutes a major degradation parameter. It depends on the stability requirements of the underlying clock synchronization mechanism and the potential amplification effect on the upper-layer synchronization error. It is usually set between 12% and 28%. In this embodiment, it is set to 20%, which can effectively identify fundamental timing disorders that are sufficient to cause systemic synchronization loss and guide the underlying calibration. The preset misalignment contribution threshold is the lower limit for determining whether pixel misalignment constitutes a major degradation parameter. It depends on the scene's requirements for image geometric integrity and the compensable range of physical installation accuracy. It is usually set between 10% and 25%. In this embodiment, it is set to 16%, which can reliably locate the key geometric misalignments that cause image breakage or ghosting and drive targeted deformation compensation or physical adjustment.

[0050] By establishing differentiated decision thresholds for each type of display quality defect, refined and configurable management of degradation parameter identification is achieved. By transforming the originally comprehensive and continuous contribution values ​​into explicit binary diagnostic conclusions, independent judgment criteria conforming to the physical characteristics and scene tolerance are established for each quality dimension. This avoids misjudgments that may result from using a single global threshold, ensuring that the diagnostic results can capture core defects that substantially affect the perceived quality of the current scene, while filtering out secondary fluctuations that, although contributing to some extent, do not yet warrant optimization intervention. This enhances the certainty, interpretability, and predictability of decision-making, making the final output set of degraded display parameters a precise, reliable, and easily executable input instruction for subsequently generating a set of composite optimization strategies. This ensures the targeted nature, efficiency, and overall stability of the adaptive optimization process.

[0051] Specifically, the process of determining the composite optimization strategy set based on the type of the degradation display parameters includes: When the color difference gradient between adjacent screens is equal to the degraded display parameter, the optimization strategy is determined to be a color mapping recalibration strategy. When the multi-screen synchronization offset is equal to the degraded display parameter or the timestamp dispersion is equal to the degraded display parameter, the optimization strategy is determined to be the frame buffer adjustment strategy. When the pixel misalignment is equal to the degraded display parameter, the optimization strategy is determined to be a geometric deformation compensation strategy. The optimization strategies are combined into a composite optimization strategy set.

[0052] In this embodiment, when display degradation parameters are detected in the multi-screen splicing system, corresponding optimization strategies are executed according to different display problems. Specifically, the color mapping recalibration strategy first maps the color response of the display devices using a color correction algorithm (such as a 3D LUT), corrects color deviations, and ensures that the display effect of each screen is consistent with the target color space. Then, the actual display effect is measured by deploying colorimeters or color sensors to obtain the color data of each display unit and compare it with the target color space to ensure color consistency in the splicing area. In addition, when the color difference in the splicing area is detected to be too large, a color mapping algorithm is applied for correction, that is, the color value of each screen is adjusted by using a color correction matrix or by applying the lookup table (LUT) again, especially in the color transition area at the splicing seam, to ensure that the color difference between screens is minimized visually and to avoid obvious color inconsistencies. The frame buffer adjustment strategy first uses frame synchronization technology to adjust the timing of the video signal, ensuring that the content of the frame buffers (i.e., video signal storage areas) of each display unit can be displayed at the same time. Then, dynamic buffer adjustment is used to fine-tune the timestamps of the frame buffers, synchronizing the buffers of each display unit to prevent increased timestamp dispersion caused by rendering delays of different display units. Additionally, when there are differences in hardware performance between different display units, or image processing delays in certain devices, a delay compensation mechanism is used to monitor data in real time and dynamically adjust the image rendering time in the frame buffer. By fine-tuning the display delay of each display unit, the rendering of each display unit is kept consistent with other display units, thus avoiding time differences in displayed content and ensuring image synchronization. The geometric deformation compensation strategy uses geometric correction algorithms to compensate for the splicing seam area, typically including perspective transformation or affine transformation, to eliminate image misalignment caused by physical installation deviations. Alternatively, a high-precision camera can be used to capture images of the splicing area, detect the degree of pixel misalignment, and image processing algorithms can be used to perform pixel-level alignment of the splicing seams.

[0053] By establishing a precise correspondence between defect types and optimization actions, upstream diagnostic results are directly converted into executable control commands, achieving a seamless intelligent decision-making closed loop from problem identification to correction implementation. Based on the physical nature and causes of different degradation parameters, the most targeted underlying optimization methods are assigned, ensuring that each optimization action precisely targets the root cause of a specific quality defect, rather than employing general global adjustments. This significantly improves the signal-to-noise ratio of correction efficiency and effectiveness. Furthermore, by combining multiple single strategies into a composite optimization strategy set, the system possesses the ability to process multiple concurrent defects in parallel or sequentially. This enables it to handle complex mixed quality degradation scenarios in real-world applications, significantly enhancing the robustness and comprehensive problem-solving capabilities of the entire method in real-world, dynamic environments.

[0054] Please see Figure 4 As shown, this is the logic diagram for adjusting the preset matching threshold in this embodiment. In this embodiment, the process of adjusting the preset matching threshold based on the frequency characteristics of the real-time splicing fidelity re-acquired within the preset adjustment period includes: The frequency at which the real-time stitching fidelity is reacquired within the preset adjustment period is counted to obtain the acquisition frequency; When the acquisition frequency is greater than the preset frequency threshold, the preset matching degree threshold is increased according to the relative deviation between the acquisition frequency and the preset frequency threshold, where F'=F×(1+k×|P-P0| / P0), where F' is the adjusted preset matching degree threshold, F is the original preset matching degree threshold, P is the acquisition frequency, P0 is the preset frequency threshold, and k is the preset adjustment coefficient.

[0055] The preset frequency threshold is a critical frequency value used to determine whether the scene matching conditions need to be adjusted due to frequent re-acquisition of real-time stitching fidelity. It depends on the design requirements for scene change sensitivity and hardware processing capabilities, and is usually set between 1 and 5 times. In this embodiment, it is set to 3 times, which can increase the preset matching degree threshold when the acquisition frequency exceeds this threshold, thereby improving the stability of stitching quality and the adaptability of the selected scene. The preset adjustment coefficient is a sensitivity factor used to control the adjustment of the preset matching degree threshold as the acquisition frequency exceeds the limit. It depends on the system's requirements for balancing matching stability and adjustment aggressiveness, and is usually set between 0.1 and 1.0. In this embodiment, it is set to 0.4, which can provide a moderate threshold increase when the frequency exceeds the limit.

[0056] By monitoring the frequency of fidelity reassessment, the overall stability is dynamically perceived. When the frequency is too high, it is determined that the current matching threshold may be too lenient, causing the scene mode to oscillate near the critical state. Accordingly, the matching threshold is automatically increased to enhance the strictness of the decision, thereby suppressing frequent mode switching caused by small fluctuations in parameters. This achieves an adaptive balance between rapid responsiveness and decision stability. It can adjust in time to maintain the splicing effect when the scene changes significantly, and avoid unnecessary decision oscillations and resource consumption when the scene is stable. This significantly improves the operating efficiency, adaptability and overall robustness in different application environments.

[0057] Specifically, the process of outputting the target scene mode and the real-time stitching fidelity after adjusting the preset matching threshold includes: The process of re-triggering the determination of the target scene mode and the real-time stitching fidelity based on adjusting the preset matching degree threshold is used to obtain the re-determined target scene mode and the real-time stitching fidelity. The redefined target scene mode and the real-time stitching fidelity are output as the current adapted scene and scene display quality score.

[0058] By re-executing scene matching and quality assessment based on the updated preset matching threshold, it is ensured that the output scene mode and fidelity score are the optimal results calculated based on the latest calibration parameters. This not only guarantees a high degree of consistency between the current adaptive scene and actual requirements, but more importantly, it concretizes the dynamically adjusted intelligent decision results into an operable and monitorable explicit output. This allows the final effect of the entire adaptive control process to be accurately perceived and applied, thus completing the entire intelligent control link from "perception, analysis, optimization" to "decision, output," and consolidating the credibility and usability of the overall self-optimization.

[0059] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A multi-scene visual multi-screen splicing control method, characterized in that, The method comprises the following steps: obtaining the resolution density value, the scene switching rate and the bit depth value of the main signal source of the current scene to be spliced in real time; determining the scene matching degree according to the resolution density value, the bit depth value, the scene switching rate and the preset scene template library, and determining the target scene mode according to the scene matching degree and the preset matching degree threshold; performing multi-screen splicing based on the target scene mode, obtaining the adjacent screen color difference gradient, the multi-screen synchronization offset of the picture dynamic content, the timestamp dispersion of the scene signal source image frame generation and the pixel misplacement amount of the splicing seam; calculating the real-time splicing fidelity according to the adjacent screen color difference gradient, the multi-screen synchronization offset, the timestamp dispersion and the pixel misplacement amount, and determining the deterioration display parameter according to the comparison result of the real-time splicing fidelity and the preset fidelity threshold; determining the composite optimization strategy set according to the type of the deterioration display parameter; re-obtaining the real-time splicing fidelity by executing the composite optimization strategy set; adjusting the preset matching degree threshold according to the frequency characteristics of the real-time splicing fidelity re-obtained within the preset adjustment time length; outputting the target scene mode and the real-time splicing fidelity re-determined after adjusting the preset matching degree threshold.

2. The method for multi-scene visual multi-screen splicing control according to claim 1, characterized in that, The process of determining the scene matching degree according to the resolution density value, the bit depth value, the scene switching rate and the preset scene template library, and determining the target scene mode according to the scene matching degree and the preset matching degree threshold comprises: calculating the standard deviation of the resolution density value within the preset observation time length to obtain the density fluctuation value; calculating the standard deviation of the bit depth value within the preset observation time length to obtain the bit depth fluctuation value; calculating the standard deviation of the scene switching rate within the preset observation time length to obtain the switching fluctuation value; obtaining the parameter determination result according to the comparison result of the density fluctuation value and the preset density threshold, the comparison result of the bit depth fluctuation value and the preset bit depth threshold, and the comparison result of the switching fluctuation value and the preset switching threshold; inputting the resolution density value, the bit depth value and the scene switching rate into the preset scene template library to determine a plurality of candidate applicable scenes based on the parameter determination result, so as to obtain a plurality of scene matching degrees; determining the target scene mode from all the candidate applicable scenes according to the comparison result of the scene matching degree and the preset matching degree threshold.

3. The method for multi-scene visual multi-screen splicing control according to claim 2, characterized in that, The target scene mode is determined from all the candidate applicable scenes when the scene matching degree is greater than the preset matching degree threshold.

4. The method for multi-scene visual multi-screen splicing control according to claim 3, characterized in that, The process of calculating the real-time splicing fidelity according to the adjacent screen color difference gradient, the multi-screen synchronization offset, the timestamp dispersion and the pixel misplacement amount, and determining the deterioration display parameter according to the comparison result of the real-time splicing fidelity and the preset fidelity threshold comprises: calculating a gradient degradation degree according to the adjacent screen color difference gradient, a preset gradient threshold and a preset gradient tolerance upper limit value, calculating a shift degradation degree according to the multi-screen synchronous shift amount, a preset shift threshold and a preset shift tolerance upper limit value, calculating a dispersion degradation degree according to the timestamp dispersion degree, a preset dispersion threshold and a preset dispersion tolerance upper limit value, and calculating a dislocation degradation degree according to the pixel dislocation amount, a preset dislocation threshold and a preset dislocation tolerance upper limit value; performing weighted summation calculation on the gradient degradation degree, the shift degradation degree, the dispersion degradation degree and the dislocation degradation degree to obtain an initial comprehensive degradation value; calculating a product of a maximum value among the gradient degradation degree, the shift degradation degree, the dispersion degradation degree and the dislocation degradation degree and a preset penalty coefficient to obtain a product penalty factor; determining the real-time splicing fidelity according to the product penalty factor and the initial comprehensive degradation value; based on a comparison result of the real-time splicing fidelity and a preset fidelity threshold, determining a plurality of degradation display parameters according to the initial comprehensive degradation value and the adjacent screen color difference gradient, the multi-screen synchronous shift amount, the timestamp dispersion degree, the pixel dislocation amount and respective degradation degrees thereof.

5. The method for multi-scene visual multi-screen splicing control according to claim 4, characterized in that, The real-time splicing fidelity is determined according to the product penalty factor and the initial comprehensive degradation value.

6. The method for multi-scene visual multi-screen splicing control according to claim 5, characterized in that, The process of determining a plurality of degradation display parameters based on a comparison result of the real-time splicing fidelity and a preset fidelity threshold according to the initial comprehensive degradation value and the adjacent screen color difference gradient, the multi-screen synchronous shift amount, the timestamp dispersion degree, the pixel dislocation amount and respective degradation degrees thereof includes: when the real-time splicing fidelity is less than the preset fidelity threshold, triggering a parameter degradation diagnosis process; based on the parameter degradation diagnosis process, calculating a difference between the adjacent screen color difference gradient and the preset gradient threshold to obtain a gradient deviation, calculating a difference between the multi-screen synchronous shift amount and the preset shift threshold to obtain a synchronous deviation, calculating a difference between the timestamp dispersion degree and the preset dispersion threshold to obtain a dispersion deviation, and calculating a difference between the pixel dislocation amount and the preset dislocation threshold to obtain a dislocation deviation; when the gradient deviation is positive, calculating a ratio of a product of the gradient degradation degree and a corresponding weight thereof and the initial comprehensive degradation value to obtain a gradient degradation contribution degree, when the synchronous deviation is positive, calculating a ratio of a product of the shift degradation degree and a corresponding weight thereof and the initial comprehensive degradation value to obtain a shift degradation contribution degree, when the dispersion deviation is positive, calculating a ratio of a product of the dispersion degradation degree and a corresponding weight thereof and the initial comprehensive degradation value to obtain a dispersion degradation contribution degree, and when the dislocation deviation is positive, calculating a ratio of a product of the dislocation degradation degree and a corresponding weight thereof and the initial comprehensive degradation value to obtain a dislocation degradation contribution degree; determining a plurality of degradation display parameters according to the gradient degradation contribution degree, the shift degradation contribution degree, the dispersion degradation contribution degree and the dislocation degradation contribution degree.

7. The method for multi-scene visual multi-screen splicing control according to claim 6, characterized in that, The process of determining a plurality of display degradation parameters according to the gradient degradation contribution degree, the offset degradation contribution degree, the discrete degradation contribution degree and the misplacement degradation contribution degree comprises: When the gradient degradation contribution degree is greater than a preset gradient contribution threshold, determining that the adjacent screen color difference gradient is the display degradation parameter, when the offset degradation contribution degree is greater than a preset offset contribution threshold, determining that the multi-screen synchronous offset is the display degradation parameter, when the discrete degradation contribution degree is greater than a preset discrete contribution threshold, determining that the timestamp discrete degree is the display degradation parameter, and when the misplacement degradation contribution degree is greater than a preset misplacement contribution threshold, determining that the pixel misplacement is the display degradation parameter.

8. The method for multi-scene visual multi-screen splicing control according to claim 7, characterized in that, The process of determining a composite optimization strategy set according to the type of the display degradation parameter comprises: When the adjacent screen color difference gradient is the display degradation parameter, determining that the optimization strategy is a color mapping recalibration strategy; When the multi-screen synchronous offset is the display degradation parameter or the timestamp discrete degree is the display degradation parameter, determining that the optimization strategy is a frame buffer adjustment strategy; When the pixel misplacement is the display degradation parameter, determining that the optimization strategy is a geometric deformation compensation strategy; Combining the optimization strategies to form a composite optimization strategy set.

9. The method for multi-scene visual multi-screen splicing control according to claim 8, characterized in that, The process of adjusting the preset matching degree threshold according to the frequency characteristic of the real-time splicing fidelity reacquired within a preset adjustment time length comprises: Counting the frequency of the real-time splicing fidelity reacquired within the preset adjustment time length to obtain an acquisition frequency; When the acquisition frequency is greater than a preset frequency threshold, increasing the preset matching degree threshold according to the relative deviation between the acquisition frequency and the preset frequency threshold.

10. The method for multi-scene visual multi-screen splicing control according to claim 9, characterized in that, The process of outputting the target scene mode and the real-time splicing fidelity re-determined after adjusting the preset matching degree threshold comprises: Outputting the target scene mode and the real-time splicing fidelity re-determined after adjusting the preset matching degree threshold as the current adaptive scene and scene display quality score.

Images