A Smart City Digital Twin Map Management System for Multiple Terminals

Abstract

This invention relates to the field of display management technology and discloses a smart city digital twin map management system for multiple terminals. The system divides the target map model into K loading levels; determines S constraint factors based on prior knowledge; determines the set of terminal devices for the target map model and obtains performance index parameters for M types of terminal devices in the set; clusters the M types of terminal devices based on the performance index parameters to obtain N types of feature terminal devices; assigns the nth type of feature terminal devices to load the target map model with random loading levels and constraint factors; performs smoothness analysis on the loaded image set to obtain a smoothness score corresponding to the nth type of feature terminal device; obtains a target loading scheme for the nth type of feature terminal device based on an optimization algorithm; and loads the target map model onto the target terminal devices based on the target loading scheme.

Description

Technical Field

[0001] This invention relates to the field of display management technology, and more specifically, to a smart city digital twin map management system for multiple terminals. Background Technology

[0002] With the continuous advancement of smart city construction, the digital twin map management system integrates data from various aspects such as urban infrastructure, transportation, environment, and energy to construct a virtual digital model of the city.

[0003] Existing smart city digital twin map management systems need to be compatible with various display terminals, such as large-screen consoles, desktop computers, tablets, and mobile devices. However, due to significant differences in processor computing power, rendering area, and graphics processing capabilities among these terminals, standardized display processing can cause lag on some devices, thus affecting the user experience. Summary of the Invention

[0004] This invention provides a smart city digital twin map management system for multiple terminals, solving the technical problems mentioned in the background.

[0005] This invention provides a smart city digital twin map management system for multiple terminals, comprising:

[0006] The model classification module is used to divide the target map model into K loading levels; among them, the loading accuracy of the target map model increases in a gradient from the first to the Kth loading level.

[0007] Load the constraint module to determine S types of constraint factors based on prior knowledge;

[0008] The device clustering module is used to determine the set of terminal devices of the target map model, obtain the performance index parameters of M types of terminal devices in the set of terminal devices, and cluster the M types of terminal devices based on the performance index parameters to obtain N types of feature terminal devices.

[0009] The intelligent loading module is used to allocate the nth type of feature terminal device to load the target map model with random loading levels and constraint factors within a preset time period, and synchronously acquire the corresponding loaded image set at fixed time intervals.

[0010] The intelligent recognition module is used to perform smoothness analysis on the loaded image set and obtain the smoothness score corresponding to the nth type of feature terminal device; 1≤n≤N, where n is a positive integer;

[0011] The loading optimization module is used to obtain the target loading scheme for the nth type of feature terminal device based on the fluency score and the optimization algorithm.

[0012] The multi-terminal loading management module is used to load the target map model onto the target terminal device based on the target loading scheme for each type of characteristic terminal device.

[0013] Furthermore, loading accuracy includes: modeling the target city model using triangulation, where each unit area of ​​the target map model includes a number of sampling points; and the difference between the number of sampling points per unit area at the k-th loading level and the (k+1)-th loading level of the target map model is... Where k and k+1 are both indices for K load levels.

[0014] Furthermore, based on prior knowledge, S impact indicators are determined for the computing power requirements of the terminal device to execute the loading of the target city model, and each impact indicator is used as a limiting factor.

[0015] The methods for determining the influencing indicators are as follows:

[0016] Determine the standard parameters for each influencing indicator;

[0017] Set up standard terminal equipment;

[0018] For standard terminal devices, the target map model is loaded, including: adjusting the standard parameters of the s-th influencing indicator with a fixed scaling factor, while keeping the standard parameters of the remaining influencing indicators unchanged;

[0019] The CPU usage ratio, memory usage ratio, disk usage ratio, and GPU usage ratio of a standard terminal device are obtained, and the computing power requirement of the s-th influencing indicator is obtained through weighted fusion. If the computing power requirement of the s-th influencing indicator is greater than or equal to the preset computing power requirement threshold, the s-th influencing indicator is retained; otherwise, the s-th influencing indicator is discarded. Wherein, 1≤s≤S, and s is a positive integer.

[0020] Furthermore, based on performance index parameters, the M types of terminal devices are clustered, including:

[0021] Performance metrics include: CPU computing power, GPU computing power, memory performance, storage performance, and rendering requirements.

[0022] CPU computing power metrics include: CPU clock speed, number of CPU cores, CPU cache size, and CPU floating-point operation parameters;

[0023] GPU computing power metrics include: GPU floating-point operation parameters, memory bandwidth, and memory capacity;

[0024] Memory performance metrics include: memory capacity and memory bandwidth;

[0025] Storage performance metrics include: number of input / output operations per second and data read / load speed;

[0026] Rendering requirements include: monitor resolution, monitor size, monitor refresh rate, and monitor color depth;

[0027] The loading score of the m-th terminal device is obtained by fusing the performance index parameters of the m-th terminal device; the M-th terminal devices are then K-clustered based on the loading score to obtain N types of characteristic terminal devices; where 1≤m≤M and m is a positive integer.

[0028] Furthermore, the nth type of feature terminal device is assigned to load the target map model with random loading levels and constraint factors, including: for the nth type of feature terminal device, loading the target map model is performed with random parameters of random loading levels and S constraint factors.

[0029] Furthermore, a smoothness analysis is performed on the loaded image set, including:

[0030] The set of loaded images includes U images, where U = T / t; where T represents a preset time period and t represents a fixed time interval.

[0031] Edge detection is performed on the u-th loaded image to obtain the feature edges in the u-th loaded image; where 1≤u≤U, and u is a positive integer;

[0032] Determine the intersection points of any number of feature edges in the u-th loaded image, and obtain the color components of the intersection points and the color components of the eight neighboring pixels to construct the matching matrix of the corresponding intersection points;

[0033] Based on the matching matrix, calculate the Euclidean distance between the matching matrices of any intersection point in the u-th loaded image and all intersection points in the (u+1)-th loaded image to obtain the matching score;

[0034] If the matching score is greater than or equal to the preset threshold, then the corresponding intersection point in the u-th loaded image and the (u+1)-th loaded image is successfully matched;

[0035] Obtain the row and column numbers of successfully matched intersection points in the u-th loaded image and the (u+1)-th loaded image, respectively, and calculate the ratio of the Euclidean distance to the number of successfully matched intersection points to obtain the local mobility parameters of the u-th and (u+1)-th loaded images.

[0036] Obtain the local fluidity parameters of each group of adjacent loaded images, and calculate the variance of U-1 local fluidity parameters to obtain the fluidity score of the loaded image set.

[0037] Furthermore, by combining optimization algorithms, the target loading scheme for the nth type of feature terminal device is obtained, including:

[0038] Step 71: For the nth type of feature terminal device, initialize and generate R loading schemes, each loading scheme including random loading level and random parameters of S types of restriction factors;

[0039] The fitness function includes: a smoothness score of the loaded image set obtained by executing the r-th loading scheme based on the n-th type of feature terminal device, which is used as the fitness value of the r-th loading scheme; where 1≤r≤R, and r is a positive integer;

[0040] Step 72: Sort the R loading schemes from smallest to largest based on their fitness values ​​to obtain the feature ranking;

[0041] Step 73: For feature sorting, retain a preset number of loading schemes from front to back, and update the remaining loading schemes as parents through crossover or mutation;

[0042] Step 74: Repeat steps 72 and 73 a preset number of times, and output the loading scheme with the position of 1 in the corresponding feature sorting as the target loading scheme for the nth type of feature terminal device.

[0043] Furthermore, loading the target map model onto the target terminal device includes:

[0044] Obtain the performance index parameters of the target terminal device and calculate the loading score of the target terminal device;

[0045] Based on the loading score of the target terminal device, cluster the target terminal devices to obtain the type of characteristic terminal device to which the target terminal device belongs;

[0046] Based on the type of the target terminal device, a corresponding target loading scheme is obtained, and the target map model is loaded based on the target loading scheme.

[0047] The beneficial effects of this invention are as follows: through the synergistic effect of hierarchical loading, terminal device performance clustering, and intelligent smoothness detection and optimization algorithms, a map loading scheme is dynamically customized according to the performance of each terminal, which effectively alleviates the lag problem of the traditional unified loading method in a multi-terminal environment, thereby significantly improving the display smoothness of the smart city digital twin map under multiple terminals. Attached Figure Description

[0048] Figure 1 This is a module diagram of a smart city digital twin map management system for multiple terminals according to the present invention. Detailed Implementation

[0049] The subject matter described herein will now be discussed with reference to exemplary embodiments. It should be understood that these embodiments are discussed only to enable those skilled in the art to better understand and implement the subject matter described herein, and changes may be made to the function and arrangement of the elements discussed without departing from the scope of this specification. Various processes or components may be omitted, substituted, or added as needed in the examples. Furthermore, features described in some examples may be combined in other examples.

[0050] like Figure 1 As shown, a smart city digital twin map management system for multiple terminals includes:

[0051] The model classification module is used to divide the target map model into K loading levels; among them, the loading accuracy of the target map model increases in a gradient from the first to the Kth loading level.

[0052] Load the constraint module to determine S types of constraint factors based on prior knowledge;

[0053] The device clustering module is used to determine the set of terminal devices of the target map model, obtain the performance index parameters of M types of terminal devices in the set of terminal devices, and cluster the M types of terminal devices based on the performance index parameters to obtain N types of feature terminal devices.

[0054] The intelligent loading module is used to allocate the nth type of feature terminal device to load the target map model with random loading levels and constraint factors within a preset time period, and synchronously acquire the corresponding loaded image set at fixed time intervals.

[0055] The intelligent recognition module is used to perform smoothness analysis on the loaded image set and obtain the smoothness score corresponding to the nth type of feature terminal device; 1≤n≤N, where n is a positive integer;

[0056] The loading optimization module is used to obtain the target loading scheme for the nth type of feature terminal device based on the fluency score and the optimization algorithm.

[0057] The multi-terminal loading management module is used to load the target map model onto the target terminal device based on the target loading scheme for each type of characteristic terminal device.

[0058] In one embodiment of the present invention, the loading accuracy includes: modeling a target city model using triangulation, wherein a unit area of ​​the target map model includes a number of sampling points; and the difference between the number of sampling points per unit area in the k-th loading level and the (k+1)-th loading level of the target map model is... Where k and k+1 are both indices for K load levels.

[0059] In one embodiment of the present invention, the target city model is a three-dimensional model. First, a complete three-dimensional city model is constructed by collecting sampling points from various locations in the city and forming a polygon with every three sampling points. Subsequently, to meet the requirements of different loading accuracy, the model is divided into K loading levels. For example, if 10,000 sampling points are collected per unit area on the facade of a building, it constitutes the highest loading level (Level K); while at lower loading levels (such as Level K-1), the model is reconstructed by uniformly removing some sampling points and retaining only 9,000 sampling points. In this way, the model accuracy under different loading levels presents a progressive change, providing a flexible and precise selection basis for multi-terminal loading management.

[0060] Specifically, a target map model contains a number of sampling points per unit area, and the number of these sampling points determines the loading accuracy of the map model within that area. To achieve a smooth transition from low to high loading accuracy, this invention divides the target map model into K loading levels, each corresponding to a different sampling point density. Specifically, for the k-th and k+1-th loading levels (where k and k+1 are indices of the K loading levels), the increase in the number of sampling points per unit area, i.e., the difference in the number of sampling points, serves as a key parameter for measuring the improvement in loading accuracy.

[0061] For example, in one embodiment, the number of sampling points per unit area for the k-th loading level can be preset to N(k), and the number of sampling points per unit area for the (k+1)-th loading level can be preset to N(k+1), with the following relationship between them: N(k+1) - N(k) = ΔN; where ΔN is a preset sampling point difference used to control the magnitude of detail increase between adjacent loading levels. In this way, lower loading levels use fewer sampling points to reduce data volume and computational burden, suitable for terminal devices with lower performance or lower requirements for display accuracy; while higher loading levels increase the number of sampling points to improve the detail of the map model, suitable for high-performance devices or occasions with higher requirements for display accuracy.

[0062] In one embodiment of the present invention, S impact indicators of the computing power requirement for the terminal device to execute the loading of the target city model are determined based on prior knowledge, and each impact indicator is used as a limiting factor.

[0063] The methods for determining the influencing indicators are as follows:

[0064] Determine the standard parameters for each influencing indicator;

[0065] Set up standard terminal equipment;

[0066] For standard terminal devices, the target map model is loaded, including: adjusting the standard parameters of the s-th influencing indicator with a fixed scaling factor, while keeping the standard parameters of the remaining influencing indicators unchanged;

[0067] The CPU usage ratio amplitude H1, memory usage ratio amplitude H2, disk usage ratio amplitude H3, and GPU usage ratio amplitude H4 of the standard terminal device are obtained, and the computing power requirement of the s-th influencing indicator is obtained by weighted fusion. If the computing power requirement of the s-th influencing indicator is greater than or equal to the preset computing power requirement threshold, the s-th influencing indicator is retained; otherwise, the s-th influencing indicator is discarded. Wherein, 1≤s≤S, and s is a positive integer.

[0068] Specifically, to adapt to the computing power requirements of different terminal devices when loading target city models, this invention proposes a method for determining influence indicators based on prior knowledge. This method, through a comprehensive consideration of various performance indicators of the terminal devices, identifies several indicators (denoted as S) that significantly influence the computing power requirements of the loading process, and uses these as limiting factors for the formulation of subsequent loading strategies. The details are as follows:

[0069] Basis for determining influencing indicators: When a terminal device performs the target city model loading task, the utilization of its hardware resources such as CPU, memory, disk, and GPU directly affects loading efficiency and display smoothness. To quantify the importance of each resource in the loading process, this invention pre-determines standard parameters for each influencing indicator based on prior knowledge (e.g., industry standards, historical data, experimental verification). These standard parameters reflect the reference values ​​that each indicator should achieve under typical working conditions, thus providing basic data for subsequent adjustments and comparisons.

[0070] Setting a Standard Terminal Device: To ensure the comparability of the determined influencing indicators, this invention selects a standard terminal device as the benchmark platform. This standard terminal device represents a common or representative device in practical applications, and its performance parameters serve as the benchmark for all influencing indicator adjustments and tests. By executing the target map model loading process on the standard terminal device, the amplitude data of various indicators (such as CPU usage ratio, memory usage ratio, disk usage ratio, and GPU usage ratio) can be obtained.

[0071] Application of a fixed proportional coefficient: For each influencing indicator (denoted as the s-th item, 1≤s≤S), this invention uses a fixed proportional coefficient to adjust its standard parameters. That is, while keeping the standard parameters of other influencing indicators unchanged, only the s-th influencing indicator is corrected using a fixed proportional coefficient, making its value closer to the changes in demand that may be encountered during actual loading. This process can be described as follows:

[0072] The standard parameter for the s-th influencing indicator is set as follows: ; by a fixed proportional coefficient Adjustments were made to obtain the corrected parameters. × The remaining influencing indicators will still be tested using their respective standard parameters.

[0073] Calculation and selection of computing power requirements: After loading the target city model on a standard terminal device, the utilization rates of CPU, memory, disk, and GPU are obtained respectively. Then, a weighted fusion method is used to combine the above utilization data with the corresponding adjustment parameters to calculate the computing power requirement value corresponding to the s-th influencing indicator. The weighted fusion method here can use a preset weighting factor to reflect the contribution ratio of each indicator to the overall computing power requirement.

[0074] The calculated computing power requirement of the s-th influencing indicator is compared with the preset computing power requirement threshold. If the computing power requirement is greater than or equal to the threshold, the indicator is considered to have a significant impact during the loading process and is therefore retained as a limiting factor. Otherwise, the indicator is discarded and is not used as a limiting factor in the subsequent loading strategy formulation.

[0075] In one embodiment of the present invention, the calculation formula for the computing power requirement of the s-th influencing index is as follows:

[0076] ;

[0077] in, This represents the computing power requirement of the s-th influencing indicator. , , and These represent the percentage of CPU usage, memory usage, disk usage, and GPU usage, respectively.

[0078] In one embodiment of the present invention, the limiting factors include, but are not limited to: the scaling, rotation or translation of the target map model performed on the terminal device, the corresponding rate of change of the target map model; and the visualization angle and visualization depth of the target map model on the terminal device.

[0079] In one embodiment of the present invention, when a user rapidly zooms in on a 3D map model using a two-finger pinch-and-drag operation, the standard zoom rate of the map model is used as the standard parameter of the corresponding limiting factor.

[0080] In one embodiment of the present invention, when a user rapidly rotates a 3D map model by sliding their finger, the standard rotation speed of the map model is used as the standard parameter of the corresponding limiting factor.

[0081] In one embodiment of the present invention, when a user rapidly moves a 3D map model by dragging, the standard translation speed of the map model is used as the standard parameter of the corresponding limiting factor.

[0082] In one embodiment of the present invention, the visualization angle of the three-dimensional map model represents the range of allowable rotation angles of the three-dimensional map model on the horizontal plane.

[0083] In one embodiment of the present invention, the visualization depth of the three-dimensional map model represents the range of allowable rotation angles of the three-dimensional map model in the vertical plane.

[0084] In one embodiment of the present invention, clustering M types of terminal devices based on performance index parameters includes:

[0085] Performance metrics include: CPU computing power. GPU computing power metrics Memory performance indicators Storage performance indicators and rendering requirement metrics ;

[0086] CPU computing power metrics include: CPU clock speed Number of CPU cores CPU cache size CPU floating-point operation parameters ;

[0087] GPU computing power metrics include: GPU floating-point operation parameters Memory bandwidth and memory capacity ;

[0088] Memory performance metrics include: memory capacity and memory bandwidth ;

[0089] Storage performance metrics include: number of input / output operations per second and data reading and loading speed ;

[0090] Rendering requirements include: monitor resolution Monitor size Monitor refresh rate and display color depth ;

[0091] The loading score of the m-th terminal device is obtained by fusing the performance index parameters of the m-th terminal device; the M-th terminal devices are then K-clustered based on the loading score to obtain N types of characteristic terminal devices; where 1≤m≤M and m is a positive integer.

[0092] Specifically, to accurately reflect the computing power and display capabilities required by the terminal device during the loading of the target city model, this invention defines five categories of performance index parameters, namely:

[0093] CPU computing power metrics include CPU clock speed, number of CPU cores, CPU cache size, and CPU floating-point operation parameters. CPU clock speed reflects the processor's operating speed, the number of CPU cores determines parallel processing capability, and cache size and floating-point operation parameters affect numerical calculation and data processing efficiency.

[0094] GPU computing power metrics include GPU floating-point operation parameters, memory bandwidth, and memory capacity. These metrics reflect the capabilities of the graphics processing unit in image rendering, graphics computing, and data transmission, and are important indicators for judging the graphics display performance of a device.

[0095] Memory performance metrics include memory capacity and memory bandwidth. Memory capacity directly affects data storage and access, while memory bandwidth determines the data transfer rate, playing a crucial role in real-time loading and image processing.

[0096] Storage performance metrics include the number of input / output operations per second and data read / load speed. The response speed and data transfer capacity of storage devices have a direct impact on the loading of large amounts of data for map models.

[0097] Rendering requirements metrics include monitor resolution, monitor size, monitor refresh rate, and monitor color depth. These metrics are directly related to the display capabilities of the terminal device and determine the visual performance of the loaded model.

[0098] Performance metric fusion and loading score calculation: For each type of terminal device (denoted as the m-th device, where 1 ≤ m ≤ M), a comprehensive evaluation of the device can be performed by acquiring its various performance metric parameters. In specific implementation, this invention uses a weighted fusion method to integrate the above-mentioned metric parameters, thereby calculating a comprehensive loading score. During the fusion process, weight factors can be set for each metric so that the contribution of each metric to the loading score reflects its impact on performance during actual loading. For example, for graphics-intensive tasks, the weights of GPU computing power and rendering requirements can be appropriately increased; while for data processing-intensive tasks, the focus may be on CPU and memory performance metrics. The result after weighted fusion is the loading score of the m-th terminal device, which quantifies the comprehensive performance of the terminal device when loading the target city model, thus providing a basis for subsequent clustering.

[0099] Device clustering based on loading scores: This invention compares the loading scores of all M types of terminal devices and uses a clustering algorithm (such as K-means clustering or other appropriate clustering methods) to group these devices. Through clustering, terminal devices with similar performance can be grouped into one category, forming N representative categories of characteristic terminal devices. Each category of characteristic terminal devices reflects the common characteristics of a group of devices in the loading model, which helps in the subsequent formulation of loading strategies. Based on the performance differences of each category of devices, different loading levels and constraint factors can be assigned, thereby achieving the goal of multi-terminal adaptive loading. By refining multiple indicators such as CPU, GPU, memory, storage, and rendering requirements, the actual computing power and display performance of each terminal device when performing loading tasks are fully reflected.

[0100] In one embodiment of the present invention, the loading score of the m-th terminal device is obtained by fusing performance index parameters of the m-th terminal device, and the calculation formula is as follows:

[0101] ;

[0102] ;

[0103] ;

[0104] ;

[0105] ;

[0106] ;

[0107] in, This represents the loading score of the m-th terminal device. This represents the CPU computing power score. This represents the GPU computing power score. This represents a memory performance score. This indicates a storage performance score. This represents the score for rendering requirement metrics. These represent the first to fifth weights, The sum of all is 1, and none of them are 0; and These represent the sixth and seventh weights, respectively. and The sum of all is 1, and none of them are 0; and These represent the eighth to ninth weights, and The sum of all is 1, and none of them are 0; and These represent the tenth to eleventh weights, and The sum of all is 1, and none of them are 0; These represent the twelfth to fifteenth weights, The sum of all is 1, and none of them are 0; These represent the weights of the CPU computing power score, GPU computing power score, memory performance score, storage performance score, and rendering requirement score, respectively.

[0108] In one embodiment of the present invention, K-clustering is performed on M types of terminal devices based on loading scores to obtain N types of characteristic terminal devices, including:

[0109] Step 100: Randomly select K terminal devices from the loading scores of all terminal devices as initial cluster centers;

[0110] Step 200: For the m-th terminal device, calculate the loading score difference with each cluster center, and assign the m-th terminal device to the cluster center with the smallest distance.

[0111] Step 300: For each cluster Gn, update the cluster center based on the mean of the loading scores of all terminal devices within cluster Gn;

[0112] Step 400: Repeat steps 200 and 300 a preset number of times to obtain N clusters.

[0113] In one embodiment of the present invention, assigning a type n feature terminal device to load a target map model with a random loading level and a constraint factor includes: for the type n feature terminal device, loading the target map model using random parameters of a random loading level and S constraint factors.

[0114] For example, the nth type of feature terminal device loads the target map model based on the third loading level, 40% of the standard zoom change speed, 50% of the standard rotation change speed, 80% of the standard translation change speed, and a visualization angle of 45 degrees to 225 degrees.

[0115] In one embodiment of the present invention, a smoothness analysis is performed on a loaded image set, including:

[0116] The set of loaded images includes U images, where U = T / t; where T represents a preset time period and t represents a fixed time interval.

[0117] Edge detection is performed on the u-th loaded image to obtain the feature edges in the u-th loaded image; where 1≤u≤U, and u is a positive integer;

[0118] Determine the intersection points of any number of feature edges in the u-th loaded image, and obtain the color components of the intersection points and the color components of the eight neighboring pixels to construct the matching matrix of the corresponding intersection points;

[0119] Based on the matching matrix, calculate the Euclidean distance between the matching matrices of any intersection point in the u-th loaded image and all intersection points in the (u+1)-th loaded image to obtain the matching score;

[0120] If the matching score is greater than or equal to the preset threshold, then the corresponding intersection point in the u-th loaded image and the (u+1)-th loaded image is successfully matched;

[0121] Obtain the row and column numbers of successfully matched intersection points in the u-th loaded image and the (u+1)-th loaded image, respectively, and calculate the ratio of the Euclidean distance to the number of successfully matched intersection points to obtain the local mobility parameters of the u-th and (u+1)-th loaded images.

[0122] Obtain the local fluidity parameters of each group of adjacent loaded images, and calculate the variance of U-1 local fluidity parameters to obtain the fluidity score of the loaded image set.

[0123] In one embodiment of the present invention, the matching matrix includes:

[0124] The color space of the loaded image is confirmed; in this embodiment, it is set to RGB color space. The matching matrix is ​​a 3×3 matrix. The RGB color components of the intersection points are obtained respectively, and the RGB color component vectors are constructed. These vectors are used as the center element of the matching matrix. Similarly, the color components of the eight neighboring pixels of the intersection points are obtained, and the corresponding vectors are constructed. Then, they are filled into the matching matrix to obtain the matching matrix.

[0125] Specifically, this method involves performing smoothness analysis on a continuously acquired set of loading images during the loading process of a target map model. Its purpose is to objectively evaluate the smoothness of screen transitions and display effects on the terminal device during loading, providing a quantitative basis for subsequent loading optimization, including:

[0126] Composition of the Loading Image Set: The loading image set contains U loading images, where U is determined by the formula U = T / t. Here, T is a preset monitoring time period, and t is a fixed image acquisition time interval. This determined value of U allows the loading process to be discretely sampled in the time dimension, thereby forming a series of continuous loading images to reflect the dynamic changes during the loading process.

[0127] Edge detection is used to extract feature edges: For each image in the loaded image set (denoted as the u-th image, 1≤u≤U), an edge detection algorithm is used to extract the feature edges in the image. Common edge detection techniques such as the Canny algorithm or other appropriate detection methods can be used to obtain clear feature edge information.

[0128] Intersection point determination and matching matrix construction: In each loaded image, for multiple detected feature edges, the intersection points of these edges are further determined. For each intersection point, its own color component and the color components of its eight neighboring pixels are collected. Using this color information, a corresponding matching matrix is ​​constructed. This matrix reflects the local color features and environmental information of the intersection point in the image, providing a basis for subsequent cross-image matching.

[0129] Cross-image matching and matching score calculation: To compare the smoothness between consecutively loaded images, for any selected intersection point in the u-th image, its matching matrix is ​​compared with the matching matrix formed by all intersection points in the (u+1)-th image. The Euclidean distance between the two is calculated to obtain the matching score. When the matching score is greater than or equal to a preset threshold, the corresponding intersection point is considered to be the same intersection point in both the u-th and (u+1)-th images.

[0130] Calculation of local mobility parameter: For each successfully matched intersection, its row and column coordinates in the u-th and u+1-th images are recorded. Using this coordinate data, the corresponding Euclidean distance is calculated, and the ratio of this distance to the number of successfully matched intersections is defined as the local mobility parameter between the two images.

[0131] Determining the overall smoothness score: Statistical analysis is performed on the local fluidity parameters (a total of U-1) of adjacent image pairs in the loaded image set, and their variance is calculated. This variance serves as the smoothness score for the entire loaded image set: a smaller variance indicates that the local fluidity parameters are relatively consistent, suggesting a smooth loading process and seamless switching; a larger variance reflects significant fluctuations during loading, which may lead to display stuttering or disjointed visual effects.

[0132] The above steps effectively reflect the continuity and smoothness of the loaded image set. This analysis result can serve as input parameters for subsequent loading optimization algorithms, helping the system dynamically adjust its loading strategy and improve the overall smoothness of multi-terminal display.

[0133] In one embodiment of the present invention, for example, if the first loaded image and the second loaded image include three successfully matched intersection points, and the calculated Euclidean distances are 0.2, 0.3 and 0.4 respectively, then the corresponding local flow parameter is local flow parameter 0.3.

[0134] In one embodiment of the present invention, obtaining the target loading scheme for the nth type of feature terminal device using an optimization algorithm includes:

[0135] Step 71: For the nth type of feature terminal device, initialize and generate R loading schemes, each loading scheme including random loading level and random parameters of S types of restriction factors;

[0136] The fitness function includes: a smoothness score of the loaded image set obtained by executing the r-th loading scheme based on the n-th type of feature terminal device, which is used as the fitness value of the r-th loading scheme; where 1≤r≤R, and r is a positive integer;

[0137] Step 72: Sort the R loading schemes from smallest to largest based on their fitness values ​​to obtain the feature ranking;

[0138] Step 73: For feature sorting, retain a preset number of loading schemes from front to back, and update the remaining loading schemes as parents through crossover or mutation;

[0139] Step 74: Repeat steps 72 and 73 a preset number of times, and output the loading scheme with the position of 1 in the corresponding feature sorting as the target loading scheme for the nth type of feature terminal device.

[0140] Specifically, the loading scheme optimization method based on optimization algorithms aims to provide the optimal loading strategy for the nth type of feature terminal device during the target map model loading process. This method utilizes randomly generated candidate loading schemes and, through steps such as fitness evaluation, ranking, elite selection, and crossover mutation, progressively optimizes the loading scheme, ultimately outputting the optimal target loading scheme.

[0141] Initialization of candidate loading schemes: For the nth type of feature terminal device, R loading schemes are first randomly generated. Each loading scheme consists of two parts: loading level and constraint factor parameters. Loading level: refers to the accuracy level used when loading the target map model, which is randomly selected from K preset loading levels; constraint factor parameters: corresponding to S constraint factors, the parameters of each of which are determined randomly. This random generation method can construct diverse candidate schemes within a large parameter space, providing a rich search starting point for subsequent optimization.

[0142] Construction of the fitness function: After execution, each candidate loading scheme calculates a smoothness score based on the set of loaded images generated after loading the target map model onto the nth type of feature terminal device. This smoothness score serves as the fitness value of the loading scheme. The fitness function uses the smoothness score of the loaded image set (e.g., the statistical results of edge matching and local fluidity parameters mentioned above) as an evaluation metric to reflect the quality of the loading scheme in achieving smooth loading effects.

[0143] In this embodiment, an ascending sorting method is adopted, that is, the lower the score (or the better the performance according to the preset evaluation criteria), the more ideal the loading scheme is considered.

[0144] Sorting and Selection of Loading Schemes: The generated R loading schemes are sorted according to their fitness values, forming a feature ranking from best to worst. The ranking result is used to distinguish the quality of loading schemes, providing a basis for retaining some excellent schemes (elite schemes) in subsequent iterations. A certain number of loading schemes are pre-selected as an elite set to ensure that excellent features are preserved during the iteration process.

[0145] Crossover and Mutation Update: For loading schemes that are not retained, they are used as parents, and new loading schemes are generated using crossover and mutation operations. Crossover operation: By combining the loading level and constraint factor parameters from two or more candidate loading schemes, new loading schemes are formed, thus inheriting the excellent characteristics of the parent scheme. Mutation operation: Some parameters in a loading scheme are randomly fine-tuned to increase the diversity of the search space and prevent the algorithm from getting trapped in local optima. This step ensures that new schemes are generated in each generation, while retaining some proven excellent schemes, improving the overall optimization effect.

[0146] Iterative optimization: Repeat steps 72 and 73, iterating a predetermined number of times to gradually converge the candidate loading schemes towards the optimal direction. After each iteration, the schemes are reordered based on the latest calculated fitness values, the elite set is updated, and a new round of crossover and mutation is performed. After several iterations, the overall fitness values ​​of the loading schemes tend to stabilize, thus selecting the optimal loading scheme.

[0147] Determining the target loading scheme: After a preset number of iterations, the loading scheme ranked first (with the best fitness value) from the final feature ranking is selected. This scheme serves as the target loading scheme for the nth type of feature terminal device. This target loading scheme determines the specific loading level and various limiting factor parameters, and is the optimal strategy obtained through automatic search and adaptive adjustment during multiple optimization processes.

[0148] This invention achieves the following effects: it can automatically select the most suitable solution for the current terminal device performance and loading scenario from a large number of candidate loading solutions; it expands the search space through crossover and mutation operations, effectively avoids local optima problems, and improves the overall smoothness of the loading process.

[0149] In one embodiment of the present invention, loading a target map model onto a target terminal device includes:

[0150] Obtain the performance index parameters of the target terminal device and calculate the loading score of the target terminal device;

[0151] Based on the loading score of the target terminal device, cluster the target terminal devices to obtain the type of characteristic terminal device to which the target terminal device belongs;

[0152] Based on the type of the target terminal device, a corresponding target loading scheme is obtained, and the target map model is loaded based on the target loading scheme.

[0153] Specifically, the core idea of ​​the dynamic loading method for loading target map models onto target terminal devices is to calculate the loading score based on the performance index parameters of the target terminal device, determine the feature category to which the device belongs through clustering, and then execute the loading operation according to the pre-optimized loading scheme for the corresponding category. The collection of target terminal device performance parameters and the calculation of the loading score are crucial.

[0154] In this embodiment, to evaluate the actual performance of the target terminal device when loading the target map model, the system first obtains various performance index parameters of the terminal device. Based on the above performance indicators, the system uses a previously set weighted fusion method to comprehensively calculate the various indicators, thereby obtaining the "loading score" of the target terminal device. This loading score can quantitatively represent the device's comprehensive computing power and display capability when performing the target map model loading task, providing a quantitative basis for subsequent clustering. After obtaining the loading score of the target terminal device, the system calculates the difference between the loading score and the average loading score of N clusters, and assigns it to the cluster corresponding to the smallest difference. Each type of feature terminal device corresponds to a target loading scheme obtained through the aforementioned optimization algorithm. The target loading scheme includes: loading level, i.e., the precision or level of detail used when loading the target map model; and specific parameter settings for each limiting factor, reflecting the control of the device's computing power requirements (e.g., scaling, rotation, translation speed limits, etc.). Once the target terminal device is classified, the target loading scheme corresponding to its category is selected as the loading scheme for that device. The advantage of this approach is that it allows for the pre-design of loading strategies best suited to the performance characteristics of devices with different performance levels, thereby optimizing loading performance and smoothness. After obtaining the corresponding target loading scheme, the system then performs the operation of loading the target map model onto the target terminal device according to that scheme.

[0155] The embodiments of this example have been described above. However, this example is not limited to the specific implementation methods described above. The specific implementation methods described above are merely illustrative and not restrictive. Those skilled in the art can make many other forms based on the guidance of this example, and all of them are within the protection scope of this example.

Claims

1. A smart city digital twin map management system for multiple terminals, characterized in that, include: The model classification module is used to divide the target map model into K loading levels; among them, the loading accuracy of the target map model increases in a gradient from the first to the Kth loading level. Load the constraint module to determine S types of constraint factors based on prior knowledge; The device clustering module is used to determine the set of terminal devices of the target map model, obtain the performance index parameters of M types of terminal devices in the set of terminal devices, and cluster the M types of terminal devices based on the performance index parameters to obtain N types of feature terminal devices. The intelligent loading module is used to allocate the nth type of feature terminal device to load the target map model with random loading levels and constraint factors within a preset time period, and synchronously acquire the corresponding loaded image set at fixed time intervals. The intelligent recognition module is used to perform smoothness analysis on the loaded image set and obtain the smoothness score corresponding to the nth type of feature terminal device; 1≤n≤N, where n is a positive integer; The loading optimization module is used to obtain the target loading scheme for the nth type of feature terminal device based on the fluency score and the optimization algorithm. The multi-terminal loading management module is used to load the target map model onto the target terminal device based on the target loading scheme for each type of characteristic terminal device. Loading accuracy includes: modeling the target city model using triangulation, where each unit area of ​​the target map model includes a number of sampling points; and the difference between the number of sampling points per unit area at the k-th loading level and the (k+1)-th loading level of the target map model. Where k and k+1 are both indices for the K load levels; Based on prior knowledge, S impact indicators are determined for the computing power requirements of the terminal device to execute the target city model, and each impact indicator is used as a limiting factor. The methods for determining the influencing indicators are as follows: Determine the standard parameters for each influencing indicator; Set up standard terminal equipment; For standard terminal devices, the target map model is loaded, including: adjusting the standard parameters of the s-th influencing indicator with a fixed scaling factor, while keeping the standard parameters of the remaining influencing indicators unchanged; The CPU usage ratio, memory usage ratio, disk usage ratio, and GPU usage ratio of a standard terminal device are obtained, and the computing power requirement of the s-th influencing indicator is obtained through weighted fusion. If the computing power requirement of the s-th influencing indicator is greater than or equal to the preset computing power requirement threshold, the s-th influencing indicator is retained; otherwise, the s-th influencing indicator is discarded. Wherein, 1≤s≤S, and s is a positive integer.

2. The smart city digital twin map management system for multiple terminals as described in claim 1, characterized in that, Clustering of M types of terminal devices based on performance metrics parameters, including: Performance metrics include: CPU computing power, GPU computing power, memory performance, storage performance, and rendering requirements. CPU computing power metrics include: CPU clock speed, number of CPU cores, CPU cache size, and CPU floating-point operation parameters; GPU computing power metrics include: GPU floating-point operation parameters, memory bandwidth, and memory capacity; Memory performance metrics include: memory capacity and memory bandwidth; Storage performance metrics include: number of input / output operations per second and data read / load speed; Rendering requirements include: monitor resolution, monitor size, monitor refresh rate, and monitor color depth; The loading score of the m-th terminal device is obtained by fusing the performance index parameters of the m-th terminal device; the M-th terminal devices are then K-clustered based on the loading score to obtain N types of characteristic terminal devices; where 1≤m≤M and m is a positive integer.

3. A smart city digital twin map management system for multiple terminals as described in claim 2, characterized in that, Assigning the nth type of feature terminal device to load the target map model with random loading level and constraint factors, including: for the nth type of feature terminal device, performing loading of the target map model with random parameters of random loading level and S constraint factors.

4. A smart city digital twin map management system for multiple terminals as described in claim 3, characterized in that, Fluency analysis was performed on the loaded image set, including: The set of loaded images includes U images, where U = T / t; where T represents a preset time period and t represents a fixed time interval. Edge detection is performed on the u-th loaded image to obtain the feature edges in the u-th loaded image; where 1≤u≤U, and u is a positive integer; Determine the intersection points of any number of feature edges in the u-th loaded image, and obtain the color components of the intersection points and the color components of the eight neighboring pixels to construct the matching matrix of the corresponding intersection points; Based on the matching matrix, calculate the Euclidean distance between the matching matrices of any intersection point in the u-th loaded image and all intersection points in the (u+1)-th loaded image to obtain the matching score; If the matching score is greater than or equal to the preset threshold, then the corresponding intersection point in the u-th loaded image and the (u+1)-th loaded image is successfully matched; Obtain the row and column numbers of successfully matched intersection points in the u-th loaded image and the (u+1)-th loaded image, respectively, and calculate the ratio of the Euclidean distance to the number of successfully matched intersection points to obtain the local mobility parameters of the u-th and (u+1)-th loaded images. Obtain the local fluidity parameters of each group of adjacent loaded images, and calculate the variance of U-1 local fluidity parameters to obtain the fluidity score of the loaded image set.

5. A smart city digital twin map management system for multiple terminals as described in claim 4, characterized in that, The target loading scheme for the nth type of feature terminal device is obtained by combining optimization algorithms, including: Step 71: For the nth type of feature terminal device, initialize and generate R loading schemes, each loading scheme including random loading level and random parameters of S types of restriction factors; The fitness function includes: a smoothness score of the loaded image set obtained by executing the r-th loading scheme based on the n-th type of feature terminal device, which is used as the fitness value of the r-th loading scheme; where 1≤r≤R, and r is a positive integer; Step 72: Sort the R loading schemes from smallest to largest based on their fitness values ​​to obtain the feature ranking; Step 73: For feature sorting, retain a preset number of loading schemes from front to back, and update the remaining loading schemes as parents through crossover or mutation; Step 74: Repeat steps 72 and 73 a preset number of times, and output the loading scheme with the position of 1 in the corresponding feature sorting as the target loading scheme for the nth type of feature terminal device.

6. A smart city digital twin map management system for multiple terminals as described in claim 5, characterized in that, Loading the target map model onto the target terminal device includes: Obtain the performance index parameters of the target terminal device and calculate the loading score of the target terminal device; Based on the loading score of the target terminal device, cluster the target terminal devices to obtain the type of characteristic terminal device to which the target terminal device belongs; Based on the type of the target terminal device, a corresponding target loading scheme is obtained, and the target map model is loaded based on the target loading scheme.

Images