Method, device and medium for measuring stereoscopic space sharing based on spatiotemporal clustering entropy

Abstract

The application discloses a three-dimensional space sharing measurement method and device based on space-time clustering entropy and a medium, comprising the following steps: determining the crowd activity data of multiple behaviors in each time period according to the crowd activity data; performing clustering analysis on the multiple behaviors respectively to obtain the clustering cluster type, the clustering cluster quantity and the clustering cluster coordinates of each layer and each time period; dividing the public space into multiple grid units respectively, determining the behavior occupancy rate of each grid unit in each time period according to the clustering cluster type, the clustering cluster quantity and the clustering cluster coordinates; for each grid unit, determining the time structure entropy of the grid unit according to the behavior occupancy rate of the grid unit in each time period; for each time period, determining the space structure entropy of the time period according to the behavior occupancy rate of each grid unit in the time period; and generating the plane sharing comprehensive score of the public space. The sharing use state of the public space can be accurately evaluated.

Description

Technical Field

[0001] This invention relates to the field of interdisciplinary technology of machine learning and urban planning, specifically to a method, device and medium for measuring three-dimensional spatial sharing based on spatiotemporal clustering entropy. Background Technology

[0002] In densely developed urban areas, three-dimensional spatial units, such as ground-level pedestrian streets and underground passages, have become important carriers for people's activities. Their shared use directly reflects the degree to which the quality of the space matches the needs of the population. How to scientifically and accurately measure the shareability of three-dimensional spatial units is a technical problem that urgently needs to be solved in the field of urban planning.

[0003] In the prior art, application CN114120018A, entitled "A Spatial Vitality Quantification Method Based on Crowd Clustering Trajectory Entropy," discloses a method that extracts crowd trajectories from public space videos, clusters these trajectories, and calculates spatial trajectory entropy to reflect spatial vitality. However, this method only considers trajectory clustering in a two-dimensional plane, failing to distinguish between passage and dwelling behaviors, neglecting the layered processing of ground, above-ground, and underground layers in a three-dimensional space, and failing to quantify the uniformity of behavior distribution over time. Therefore, it is difficult to accurately reflect the true shared usage status of three-dimensional spatial units.

[0004] Furthermore, publication number CN120355552A, entitled "A Method for Measuring the Vitality of Urban Community Public Spaces Based on Spatiotemporal Behavior," discloses a method for recording residents' residence activities via GPS and calculating residence rate, residence density, and activity diversity. However, this method only focuses on residence behavior, neglecting passage behavior, and is only applicable to two-dimensional community spaces, making it unsuitable for multi-level, three-dimensional spaces. Its diversity index is unrelated to the uniformity of behavior distribution in time and space, and it cannot comprehensively measure the shared nature of three-dimensional spaces.

[0005] In summary, existing technologies still lack a complete solution to accurately capture multi-layered behavioral characteristics in three-dimensional space, perform time-segmented quantitative clustering, and measure their sharing based on entropy values. Summary of the Invention

[0006] To address the shortcomings of existing technologies, this application provides a method, device, and medium for measuring the shared use of three-dimensional spaces based on spatiotemporal clustering entropy, which can better reflect the usage of three-dimensional spatial units by people in different time periods and accurately assess the shared use status of public spaces.

[0007] To address the above problems, the present invention provides the following technical solution: In a first aspect, embodiments of this application provide a method for measuring the sharing of three-dimensional space based on spatiotemporal clustering entropy, including: acquiring crowd activity data of public spaces on the ground floor, above-ground floor and underground floor of a target three-dimensional area; Based on the crowd activity data, determine the crowd activity data for various behaviors in each time period. The crowd activity data includes activity type, number of activities, and activity coordinates. Cluster analysis was performed on the crowd activity data to obtain the cluster type, number of clusters, and cluster coordinates for each layer and time period; The public spaces of the ground floor, above-ground floor, and underground floor are divided into multiple grid units. Based on the cluster type, number of clusters, and cluster coordinates, the behavior occupancy rate of each grid unit in each time period is determined. For each grid cell, the temporal structure entropy of the grid cell is determined based on its behavior occupancy rate in each time period. The temporal structure entropy is used to quantify the balance of the distribution of crowd activities in the time dimension within the grid cell. For each time period, the spatial structure entropy of that time period is determined based on the behavior occupancy rate of each grid unit in that time period. The spatial structure entropy is used to quantify the balance of the spatial distribution of crowd activities within that time period. A comprehensive score for the planar shareability of public spaces is generated based on temporal and spatial structure entropy.

[0008] In some embodiments, after acquiring crowd activity data in the public spaces of the ground floor, above-ground floor, and underground floor of the target three-dimensional area, wherein the crowd activity data includes activity type, activity quantity, and activity coordinates, the method further includes: Based on the crowd activity data, extract the characteristic information of each individual; Based on the feature information, construct the behavioral feature vector for each individual; Based on each individual's behavioral feature vector, we can determine whether each individual's behavior type belongs to the stationary behavior or the passage behavior.

[0009] In some implementations, cluster analysis is performed on multiple behaviors separately to obtain cluster types at each level and time period, including: Multiple behaviors are categorized into passage behaviors and dwelling behaviors; Cluster analysis was performed on the traffic behavior to obtain traffic clusters at each level and time period; Cluster analysis was performed on the stay behavior to obtain stay behavior clusters at each level and time period, and the type of each stay behavior cluster was determined. The types of stay behavior clusters include at least one of the following: rest cluster, dining cluster, sports cluster, and children's activity cluster.

[0010] In some implementations, the three-dimensional spatial sharing measure based on spatiotemporal clustering entropy further includes: Cluster analysis was performed on dwell behavior to obtain clusters of dwell behavior at each level and time period; Obtain activity tags for the public spaces on the ground floor, above-ground floor, and underground floor of the target three-dimensional area. Each activity tag corresponds to a location coordinate. Match the coordinates of the dwell behavior clusters with the location coordinates corresponding to the activity labels; Determine the type of cluster for the dwell behavior.

[0011] In some implementations, the behavior occupancy of each grid cell in each time period is determined based on the cluster type, number of clusters, and cluster coordinates, including: Determine the corresponding grid cell based on the coordinates of the cluster; Based on the number of clusters, determine the behavior occupancy of each grid cell in each time period.

[0012] In some implementations, it also includes: Collect the number of different types of clusters and the coordinates of each cluster for each grid cell in each time period; Obtain the average number of different types of clusters and the average coordinate offset for each grid cell across all time periods; The correlation coefficient between clusters and time rhythm in each time period is determined based on the number of different types of clusters, the coordinates of the clusters, the average number of clusters, and the average coordinate offset. The core cluster type and auxiliary cluster type are determined based on the correlation coefficient, wherein the correlation coefficient corresponding to the core cluster type is greater than a preset third threshold. Based on the core cluster type and the auxiliary cluster type, a shared spatiotemporal measurement model is constructed.

[0013] In some implementations, the comprehensive sharing score includes a planar sharing score, which generates a planar sharing score for public spaces based on temporal structure entropy and spatial structure entropy, including: The temporal sharing balance of the common space is determined based on the temporal structure entropy of each grid cell. The spatial sharing balance of the public space is determined based on the spatial structure entropy of each grid cell. Based on the time-sharing balance and the spatial-sharing balance, a comprehensive score for the planar sharing of each grid cell is determined.

[0014] In some implementations, the comprehensive sharing score includes a vertical sharing score, which is generated based on temporal structure entropy and spatial structure entropy to form a comprehensive vertical sharing score for the public space, including: Divide the grid cells of the same two-dimensional plane coordinates of multiple layers into a three-dimensional space cell; For a single three-dimensional spatial unit, the first three-dimensional time structure entropy of the three-dimensional spatial unit is calculated based on the number of its clusters in the ground layer, above-ground layer and underground layer. The average value of the second three-dimensional time structure entropy is calculated based on the time structure entropy of the corresponding grid units of the ground layer, the above-ground layer and the underground layer. The vertical correlation coefficient is determined based on the average of the first and second three-dimensional time structure entropies. Acquire the movement trajectory of each individual and extract the trajectory information of all individuals passing through vertical transportation equipment; Construct a behavior transition matrix based on the trajectory information; Based on the behavior transition matrix, determine the transition ratio of each type of behavior from the starting floor to the next floor; The comprehensive score for vertical sharing is determined based on the vertical correlation coefficient and the transfer ratio.

[0015] Secondly, embodiments of this application provide an electronic device, including: At least one processor; and, A memory that is communicatively connected to at least one processor; wherein, The memory stores instructions that can be executed by at least one processor, such that the at least one processor can perform a spatial sharing measure based on spatiotemporal clustering entropy, as described in the first aspect.

[0016] Thirdly, embodiments of this application provide a computer-readable storage medium storing an executable program, which is executed by a processor to implement the three-dimensional spatial sharing measure based on spatiotemporal clustering entropy as described in the first aspect.

[0017] This application provides a method, device, and medium for measuring the shared nature of three-dimensional space based on spatiotemporal clustering entropy. Through three-dimensional hierarchical data collection, separate clustering of passage and dwelling behaviors, and calculation of spatiotemporal dual-dimensional entropy values, a measurement system for the shared nature of urban three-dimensional spatial units is constructed. This method distinguishes between passage and dwelling behaviors and uses temporal and spatial structural entropy to quantify and score shared nature, providing a decision-making basis for spatial optimization of three-dimensional areas. Attached Figure Description

[0018] Figure 1 This is a flowchart illustrating the three-dimensional spatial sharing measure based on spatiotemporal clustering entropy provided in the embodiments of this application.

[0019] Figure 2 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application.

[0020] Figure 3This is a structural block diagram of a computer-readable storage medium provided in an embodiment of this application. Detailed Implementation

[0021] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of this application.

[0022] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0023] The method for measuring three-dimensional spatial sharing based on spatiotemporal clustering entropy provided in this application will be described in detail below with reference to the accompanying drawings.

[0024] Optionally, this embodiment takes the core area of ​​a city's central business district (CBD) as the research object. This area includes a ground-level commercial pedestrian street, a two-story above-ground connecting corridor system, a two-story underground commercial street, and a sunken plaza. It is a high-intensity development area with typical three-dimensional characteristics.

[0025] Please see Figure 1 , Figure 1 This is a flowchart illustrating the method 1 for measuring the shared spatiality based on spatiotemporal clustering entropy provided in this application. Figure 1 As shown, the three-dimensional spatial sharing measurement method 1 based on spatiotemporal clustering entropy includes steps S100 to S600.

[0026] Step S100: Obtain crowd activity data for the public spaces of the ground floor, above-ground floor and underground floor of the target three-dimensional area. The crowd activity data includes activity type, activity quantity and activity coordinates.

[0027] Optionally, video equipment can be used to continuously film in high-intensity development areas, with the camera range covering the main public spaces on the ground floor, above-ground floor, and underground floor to obtain data on crowd activity in the public spaces on the ground floor, above-ground floor, and underground floor of the target three-dimensional area.

[0028] It is important to note that during video capture, all cameras only film public spaces and do not focus on or track individuals. In the video analysis phase, deep learning-based pedestrian detection algorithms are used, such as YOLOv5, to identify pedestrian regions in the images. However, this algorithm only extracts the bounding boxes and center point coordinates of pedestrians, without extracting facial features or any biometric information that could identify an individual. For situations involving multiple people walking together, only the number of people in the group is counted; the identities of individuals within the group are not identified.

[0029] In some embodiments, steps S110 to S130 are further included after step S100, and the specific steps are as follows: Step S110: Extract the feature information of each individual based on the crowd activity data.

[0030] Optionally, the video data includes a time series, and the feature information includes the consecutive location points of each individual in the time series and the corresponding timestamps.

[0031] Optionally, a deep learning-based multi-object tracking algorithm (DeepSORT) is used to analyze the video, extracting the continuous position coordinate sequence of each individual over time, thereby constructing a spatiotemporal trajectory representing the individual's motion characteristics. For each individual, its position coordinates at each timestamp are recorded. ) and the corresponding timestamp This forms a complete sequence of trajectory features.

[0032] Step S120: Construct the behavioral feature vector for each individual based on the feature information.

[0033] Optionally, construct a behavioral feature vector F that includes spatiotemporal attributes. ,in, Continuous position points of each individual in the time series coordinate, Continuous position points of each individual in the time series coordinate, For the corresponding timestamp, For each individual's instantaneous velocity, The acceleration for each individual, The displacement change for each individual. The neighborhood density for each individual.

[0034] Step S130: Based on the behavioral feature vector of each individual, determine whether the behavior type of each individual belongs to the stationary behavior or the passage behavior.

[0035] according to and The system identifies behavior types, confirms whether each individual's behavior type belongs to a common behavior category, and combines this with... Distinguish between single-person passage and multi-person passage.

[0036] Aside from passage-type behaviors, the rest are defaulted to stay-type behaviors, which are further divided into stay-type behaviors with complete trajectories and stay-type behaviors with interrupted trajectories.

[0037] For stationary behaviors with complete trajectories, according to and Identify behavior types.

[0038] For example, if the moving speed is less than a preset first threshold, or the displacement distance is less than a preset second threshold, then the individual is determined to be in a stationary state during that time period, and the stationary state point is marked as the clustering input for stationary state behavior.

[0039] For example, the first threshold can be set to 0.3 m / s to 0.5 m / s, and the second threshold can be set to 0.5 m to 1 m. When <First threshold or When the speed falls below the second threshold, the individual is considered to be in a stationary state during that time period. The first threshold is set based on the lower limit of normal human walking speed, typically between 0.3 m / s and 0.5 m / s. When the speed is below this range, the individual can be considered to be in a state of stagnation or slight movement.

[0040] For stationary behaviors with interrupted trajectories, such as those caused by an individual entering an indoor area or obstructing a region in video surveillance, if the point of interruption is within the effective service radius of a preset activity label, the continuous location points are marked as clustering inputs for stationary behaviors. The corresponding behavior type is then identified based on the activity label, thereby enabling the identification of stationary behaviors within video blind spots.

[0041] For example, if the moving speed is greater than a preset first threshold, or the displacement distance is greater than a preset second threshold, then the individual is determined to be passing through during that time period.

[0042] when ≥ First threshold and When the value is greater than or equal to the second threshold, the individual is determined to have engaged in passage behavior during that time period. Passage behavior includes single-person passage and multi-person passage. When it is determined to be multi-person passage, the number of people is additionally identified and recorded.

[0043] Specifically, if an individual is in a passage state within an adjacent timestamp and there are no other individuals in its neighborhood (radius 3 meters, time window ± 2 seconds), it is marked as a single person passage point, and its coordinates and time are recorded separately. If there are other individuals in the neighborhood, mark it as a multi-person passage point and record the number of people in the group.

[0044] Step S130 can distinguish between passage behavior and dwelling behavior. For individuals whose trajectory is interrupted, their dwelling behavior type can be determined. For individuals whose trajectory is not interrupted and whose behavior is dwelling, their behavior type can be identified through step S2124.

[0045] Step S200: Perform cluster analysis on the crowd activity data to obtain the cluster type, number of clusters, and cluster coordinates for each layer and time period.

[0046] In some implementations, step S200 includes steps S210 to S230, wherein step S210 is used to determine the cluster type, step S220 is used to determine the number of clusters, and step S230 is used to determine the cluster coordinates.

[0047] In some embodiments, step S210 includes steps S211 to S213, the specific steps of which are as follows: Step S211: Divide the various behaviors into passing behaviors and staying behaviors.

[0048] Specifically, step S130 can be used to classify various behaviors into passage behaviors and stay behaviors.

[0049] Step S212: Perform cluster analysis on the traffic behavior to obtain traffic clusters for each layer and time period.

[0050] Specifically, a density-based spatial clustering algorithm (DBSCAN) is used to cluster traffic behavior points in each layer and time period. The clustering parameters of the DBSCAN algorithm are set as follows: the neighborhood radius ranges from 8 to 12 meters, which can be set according to the minimum aggregation scale of crowd activities in urban public spaces. In this embodiment, 10 meters is used. The minimum number of neighborhood samples ranges from 1 to 8, which can be set according to the data sampling frequency. The minimum number of neighborhood samples also ranges from 1 to 8, and in this embodiment, 5 is used. This means that only when there are 5 or more traffic behavior points within a radius of 10 meters can a traffic cluster be formed. Single-person traffic points (only 1 person in the neighborhood) and groups of 2 to 4 people (less than 5 people in the neighborhood) do not meet the core point condition and are therefore classified as noise points and removed from the clustering results.

[0051] Specifically, step S212 includes steps S2121 to S2125, and the specific steps are as follows: Step S2121: For each passage point, calculate the number of sample points contained in its neighborhood.

[0052] Step S2122: If the number of sample points in the neighborhood of a point is greater than or equal to the number of samples in the minimum neighborhood, then mark the point as a core point; Step S2123: Using each core point as a seed, group all points in its neighborhood into the same cluster; Step S2124: Merge interconnected core points and their associated points into the same common cluster; Step S2125: Noise points that cannot be assigned to any cluster are removed.

[0053] Step S213: Perform cluster analysis on the stay behavior to obtain stay behavior clusters at each level and time period, and determine the type of each stay behavior cluster. The types of stay behavior clusters include at least one of rest cluster, dining cluster, sports cluster, and children's activity cluster.

[0054] In some implementations, step S213 includes steps S2131 to S2134.

[0055] Step S2131: Perform cluster analysis on dwell behavior to obtain dwell behavior clusters for each layer and time period.

[0056] Specifically, the DBSCAN algorithm is used for cluster analysis of dwelling behavior, with parameter settings consistent with those for passage behavior. The difference lies in the clustering input: the complete trajectory point set and the trajectory interruption point set are used for dwelling behavior. The complete trajectory point set includes multiple dwelling behavior points identified in step S230 that maintain a continuous trajectory within the video surveillance range. The trajectory interruption point set includes multiple trajectory interruption points identified in step S230, where the coordinates of their last occurrence are located within the effective service radius of a preset active label. By unifying these two types of point sets as clustering input, collaborative clustering of dwelling behavior in both blind and non-blind spots of video surveillance is achieved, ensuring the completeness and accuracy of dwelling behavior analysis.

[0057] For example, the kernel density analysis of the lingering behavior at the ground level from 16:00 to 17:00 on weekends showed that multiple high-density clusters formed in the plaza area, located in the rest seating area, the dining area, and other shopping areas.

[0058] Step S2132: Obtain activity tags for the public spaces of the ground floor, above-ground floor, and underground floor of the target three-dimensional area. Each activity tag corresponds to a location coordinate.

[0059] Specifically, activity tags can be obtained from planning drawings, site surveys, or map data, including the coordinates, names, and types of restaurants, fitness facilities, indoor playgrounds, children's playgrounds, rest areas, etc. Each activity tag corresponds to a specific location coordinate.

[0060] Step S2133: Match the coordinates of the dwell behavior cluster with the location coordinates corresponding to the activity label.

[0061] For the cluster of dwell behaviors corresponding to complete trajectory points, optionally, a spatial proximity matching algorithm can be used to calculate the distance between the center coordinates of each dwell behavior cluster and the location coordinates of each activity label. If the distance is less than a preset threshold, the cluster is considered to match the activity label. Optionally, the preset threshold can be in the range of 3 to 5 meters.

[0062] For the cluster of dwell behavior corresponding to the trajectory interruption point, optionally, the trajectory interruption point is located within the effective service radius of the preset activity label, and the trajectory interruption point is regarded as the cluster corresponding to the activity label.

[0063] Step S2134: Determine the type of cluster for dwell behavior.

[0064] Optionally, the type of dwell behavior cluster can be determined based on the matching results. For example, if a dwell behavior cluster matches the rest area label, it is determined to be a rest cluster. If a dwell behavior cluster matches the restaurant label, it is determined to be a restaurant cluster. If a dwell behavior cluster matches the fitness facility label, it is determined to be an exercise cluster. If a dwell behavior cluster matches labels such as indoor amusement park or children's playground, it is determined to be a children's activity cluster.

[0065] The above scheme can quickly identify the type of cluster of dwelling behavior and save computing power, but some complete trajectory points and activity labels cannot be matched, so algorithm recognition is required to analyze individual actions.

[0066] Optionally, the behavior of staying can be divided into corresponding individual actions, including rest, such as sitting, lying down, leaning, etc.; eating, such as eating, etc.; exercise, such as running, fitness, square dancing, etc.; and children's activities, such as playing in areas such as children's playgrounds, etc.

[0067] In some implementations, to further improve the accuracy of dwelling behavior recognition, a human pose recognition-based method can be used to determine the type of specific dwelling behavior clusters. Optionally, the YOLOv8 algorithm can be used to detect human bodies and output keypoint coordinates, performing human target detection and pose estimation on video stream data, and extracting the skeletal keypoint coordinates of each individual. The model using the YOLOv8 algorithm can detect multiple keypoints, such as nose, eyes, ears, shoulders, elbows, wrists, palms, hips, knees, and ankles, with each keypoint corresponding to pixel coordinates and its confidence level in the image coordinates. To protect personal privacy, the algorithm only extracts the geometric position information of the keypoints and does not perform facial feature recognition or identity matching. All keypoint data is used only for behavior analysis, and the original image is discarded immediately after keypoint extraction. Based on the angular relationships, distances, and time-series features between keypoints, behavior types such as resting (sitting, lying down, leaning), eating (eating actions), and exercise (square dancing, high knees in place, fitness) can be identified. The identification results can be cross-validated with the activity label matching results: for successfully matched clusters, posture recognition can be used as an auxiliary verification; for clusters without matching labels, posture recognition can be used as the primary identification method. By fusing the two methods, comprehensive and accurate identification of dwelling behavior is achieved.

[0068] Specifically, based on the key features of the hand and mouth, a function of the hand-mouth distance changing over time is calculated. ,like If the distance is less than a preset threshold and the duration exceeds a preset time, it is counted as one hand-mouth proximity event. The number N of proximity events within the preset time window is counted. If the number N is greater than or equal to the preset threshold, the behavior cluster is determined to be a dining cluster. For example, the preset threshold ranges from 0.1 to 0.15 meters, and the preset time ranges from 1 to 2 seconds. When a hand-mouth proximity event is detected, even if the individual is simultaneously in a resting posture, it is still preferentially determined to be a dining behavior. This is because eating is a definitive indicator of dining behavior, while a resting posture only indicates a resting state and does not exclude the possibility of eating.

[0069] Specifically, resting behavior is identified by the angular relationships formed by key points on the human body to determine different posture types. Based on biomechanical principles, the angles between different limb segments exhibit a specific range distribution in different postures such as sitting, lying down, and leaning. Based on the characteristics of the hip, knee, and ankle key points, the angles between the hip-knee vector and the knee-ankle vector are calculated to obtain the angle between the thigh and lower leg. Based on the key features of the shoulder, hip, and knee, the angles between the shoulder-hip vector and the hip-knee vector are calculated to obtain the angle between the upper torso and the thigh. Based on the key features of the shoulder and hip, the angle between the shoulder-hip vector and the horizontal plane is calculated to obtain the angle between the upper torso and the horizontal plane. .

[0070] If the conditions for catering behavior are not met, then according to and First, determine whether it is a seated or reclining type of rest, for example... , The duration must be greater than or equal to a preset time, which can optionally be 20 to 30 seconds. If the rest is not in a seated or reclining position, then according to... To determine whether it is leaning against something for rest, if If so, it is determined to be resting in a lying position.

[0071] The core characteristic of motor behavior is the periodic swinging of limbs. Whether it's running in place or stationary movement (routine movements of the arms or legs), it will exhibit a clear periodic signal in the time series. Specifically, the type of movement is determined based on the key features of the wrist and ankle. A Fast Fourier Transform (FFT) is performed on the positional changes of the wrist and ankle to extract the dominant frequency and amplitude. If the dominant frequency exceeds a preset threshold, it is determined that periodic movement has occurred, thus classifying it as a type of movement.

[0072] It is important to note that periodic limb swinging can occur during commuting activities (running, brisk walking) or stationary exercises (aerobics, weightlifting), and can be distinguished in the following ways: Based on step S120, the displacement distance or movement speed of each individual has been determined. If the displacement distance within a preset time period is less than a preset threshold, or the movement speed is less than a preset threshold, then it is determined to be a moving cluster within the stationary behavior cluster; otherwise, it is a passing cluster. For example, if the preset time is one minute, and the displacement distance within one minute is greater than 1 meter, or the speed is greater than 0.5 meters / s, then it is a passing cluster; otherwise, it is a moving cluster within the stationary behavior cluster.

[0073] Step S220 is used to determine the number of clusters. Specifically, step S220 includes steps S221 to S222, and the specific steps are as follows: Step S221: Count the total number of independent clusters formed in each layer and time period. This indicator is used to characterize the spatial dispersion and distribution breadth of crowd activities in public spaces.

[0074] Step S222: Count the number of people in each independent cluster at each level and time period.

[0075] For clusters with complete trajectories: count the number of individuals contained in the cluster to obtain the real-time number of residents in the cluster.

[0076] For clusters with interrupted trajectories: Since they are in blind spots, calculations are performed based on the law of conservation of traffic flow at entrances and exits in the video capture area. Specifically, the calculation is as follows: Number of people in this cluster at the current time = Number of people in this cluster at the previous time + Total number of individuals entering the area. The total number of individuals who left the area.

[0077] Step S230 is used to determine the coordinates of the clusters. Specifically, step S230 includes step S231.

[0078] Step S231: For clusters with complete trajectories, the centroid calculation method can be used. Specifically, by calculating the average value of these coordinates in the horizontal direction (X-axis) and the vertical direction (Y-axis), the center coordinates of the cluster can be obtained. These center coordinates represent the geometric center of the population activity area and can effectively characterize the spatial distribution of the cluster in the three-dimensional area.

[0079] For clusters with interrupted trajectories, since these clusters are located in blind spots of video surveillance, their centroids cannot be calculated from the trajectory points. Therefore, the coordinates of this cluster are used as the position coordinates of its corresponding active label.

[0080] Step S300: Divide the public spaces of the ground floor, above-ground floor and underground floor into multiple grid units respectively. Calculate the behavior occupancy rate of each grid unit in each time period based on the cluster type, number of clusters and cluster coordinates.

[0081] In some embodiments, step S300 includes steps S310 to S340, the specific steps of which are as follows: Step S310: Divide the common space of the ground floor, above-ground floor and underground floor into multiple grid units according to the preset size, and spatially encode each grid unit.

[0082] In some implementations, the public spaces of the ground floor, above-ground floor, and underground floor are divided into grids according to preset dimensions. Optionally, the preset dimensions can be 50 meters × 50 meters for macro-structural control and functional zoning; alternatively, the preset dimensions can be 10 meters × 10 meters or 5 meters × 5 meters for more detailed design.

[0083] Each grid cell is assigned a unique spatial code, and the coding rule can be in the form of layer code - grid row number - grid column number - scale identifier.

[0084] Optionally, when dividing the common space of the ground floor, above-ground floor, and underground floor into grid units, the grid units of each floor can be made to correspond, that is, have the same two-dimensional planar coordinates, namely the X coordinate and Y coordinate. Grid units with the same two-dimensional planar coordinates on multiple layers form a three-dimensional space unit, which is used for the calculation of vertical sharing score in subsequent steps S610B to S680B.

[0085] Step S320: Determine the corresponding grid cell based on the cluster coordinates.

[0086] The center coordinates of each cluster are imported into the GIS server, and each cluster is matched to its corresponding grid cell based on the coordinates. For clusters located on grid boundaries, their corresponding grid cells are determined by the center point of the cluster.

[0087] Step S330: Determine the behavior occupancy of each grid cell in each time period based on the number of clusters.

[0088] For the i-th grid cell in the j-th time period, the behavior occupancy rate Calculate using the following formula: ; In the formula, Let i be the occupancy rate of the i-th grid in time period j. Let i be the number of clusters in the i-th grid during the j-th time period. Let i be the number of clusters in the i-th grid throughout the entire time period. Total number of time periods satisfy It has been normalized.

[0089] Specifically, For the i-th grid in the th... The number of clusters for a given time period, with the minimum number of neighboring samples ranging from 1 to 8, can be set to 1. This means that there is one behavior point within a 10-meter radius, forming a passage cluster. In this case, the number of clusters includes all passage points and all stopping points. Passage points include single-person passage points, multi-person passage points, and all points within the passage cluster. Stopping points include various types of stopping points such as rest areas, restaurants, sports areas, and children's activity areas.

[0090] Step S400: For each grid cell, determine the temporal structure entropy of the grid cell based on its behavior occupancy rate in each time period. The temporal structure entropy is used to quantify the balance of the distribution of crowd activities in the time dimension within the grid cell.

[0091] In some implementations, the temporal structure entropy for the i-th grid cell is calculated using the following formula: ; In the formula, For the first Temporal structure entropy of each grid cell Total number of time periods For the first The grid cell in the first... The percentage of time-based activities.

[0092] Specifically, Calculation using the formula: In the formula, Let be the total number of clusters for the i-th grid in time period t, including both passage and dwell behaviors. Total number of time periods satisfy It has been normalized.

[0093] Specifically, the range of values ​​for the time structure entropy is 0 ≤ ≤ The larger the entropy value, the more evenly the activity is distributed over time within the grid cell, meaning there is a relatively balanced amount of activity throughout the day; the smaller the entropy value, the more concentrated the activity is in a few specific time periods, such as only during the morning and evening peak hours.

[0094] This is used to calculate the temporal structure entropy of each grid cell.

[0095] For the entire target three-dimensional area, after calculating the temporal structure entropy of each grid cell in the ground layer G, above-ground layer A, and underground layer U respectively, the temporal structure entropy of the entire three-dimensional area can be calculated in the following manner: ; In the formula, For the ground layer G, the first Temporal structure entropy of each grid cell For the ground layer G, the first The area of ​​the common space of each grid cell The first floor of the above-ground layer A Temporal structure entropy of each grid cell The first floor of the above-ground layer A The area of ​​the common space of each grid cell For the underground layer U Temporal structure entropy of each grid cell For the underground layer U The common space area of ​​each grid cell.

[0096] The average temporal structure entropy of the entire three-dimensional area is used to reflect the balance of population activity distribution in the time dimension of the entire area.

[0097] If you want to calculate the temporal structure entropy of the ground layer G, the above-ground layer A, and the underground layer U separately, you can refer to the above formula for calculation.

[0098] Step S500: For each time period, calculate the spatial structure entropy of that time period based on the behavior occupancy rate of each grid cell in that time period. The spatial structure entropy is used to quantify the balance of the spatial distribution of crowd activities within that time period.

[0099] In some implementations, the spatial structure entropy for the j-th time period is calculated using the following formula: ; In the formula, For the first Spatial structure entropy over time period The total number of grid cells. For the first The grid cell in the first... The percentage of behaviors during each time period.

[0100] Specifically, Calculation using the formula: In the formula, For the first The number of clusters in a grid cell at time j, including various behaviors. The total number of grid cells. satisfy It has been normalized.

[0101] It is important to note that when used to calculate temporal structure entropy, the behavior occupancy rate... It refers to the ratio of the number of clusters in the same grid cell during a certain period to the total number of clusters in the same grid cell throughout the entire period, reflecting the distribution weight of activity on the time axis during that period.

[0102] When used to calculate spatial structure entropy, behavior occupancy It refers to the ratio of the number of clusters of the same grid unit in a certain period of time to the total number of clusters of all grid units in that period of time, reflecting the distribution weight of the grid activity on the spatial axis.

[0103] The range of spatial structure entropy is 0 ≤ ≤ The larger the entropy value, the more evenly the activities are distributed in space during that period, meaning the activities are scattered across multiple grid cells; the smaller the entropy value, the more concentrated the activities are in a few specific grid cells.

[0104] The average spatial structure entropy of the entire three-dimensional area can be calculated by referring to the method for calculating the average temporal structure entropy.

[0105] Step S600: Generate a comprehensive score for the sharing of public space based on the temporal structure entropy and the spatial structure entropy.

[0106] In some implementations, the shared comprehensive score includes a planar shared comprehensive score, which only considers the temporal and spatial entropy within a single layer of the grid and is obtained by calculating the balance. Step S600 includes steps S610A to S630A, and the specific steps are as follows: Step S610A: Determine the time-sharing balance of the common space based on the time structure entropy of each grid cell.

[0107] In some implementations, the time structure entropy is normalized to obtain the time sharing balance: ; In the formula, To share the balance, For time structure entropy, It represents the minimum value of the temporal structure entropy across all time periods. This represents the maximum value of the temporal structure entropy across all time periods.

[0108] The above calculation is for the time-sharing balance of the i-th grid cell. To calculate the time-sharing balance of the target 3D patch encompassing all grid cells... It needs to be calculated using the following formula: ; In the formula, N is the total number of grid cells. For the time-sharing balance of the i-th grid, The weights can be determined based on the area of ​​each grid cell. set up.

[0109] Step S620A: Determine the spatial sharing balance of the common space based on the spatial structure entropy of each grid cell.

[0110] In some implementations, for the j-th time period, the spatial structure entropy is normalized to obtain the spatial sharing equilibrium degree.

[0111] ; In the formula, For the spatial sharing equilibrium degree in time period j, Let the spatial structure entropy be the value at time j. This represents the minimum spatial structure entropy among all grid cells. This represents the maximum value of the spatial structure entropy across all grid cells.

[0112] The above calculation is for the spatial sharing balance of the j-th time period. To calculate the spatial sharing balance of the target three-dimensional area encompassing all time periods... It needs to be calculated using the following formula: ; In the formula, The spatial sharing balance is given by N for time period j, where N is the total number of time periods. The weights can be set according to each time period. In this embodiment, the time period is 24 hours. The weight for each time period is 1. .

[0113] Step S630A: Determine the comprehensive planar sharing score for each grid cell based on the time sharing balance and spatial sharing balance.

[0114] Specifically, a weighted summation method is used to calculate the comprehensive score of planar sharing for a region containing all grid cells. :

[0115] In the formula, The weighting coefficient for time-sharing equilibrium. The weighting coefficient for spatial sharing equilibrium. To achieve a balanced distribution of time across the region, To achieve a balanced level of spatial sharing within the area.

[0116] + =1, and It can be configured according to actual needs. In this embodiment, and All are taken as 0.5, for and Normalization was performed.

[0117] like If the score is less than the preset threshold, the score is too low. If the activity level is too low, it indicates a significant difference in activity levels across different time periods. For example, during off-peak hours, limited-time events or special offers could be implemented. If the height is too low, it indicates poor spatial distribution balance. In such cases, the location of the entrances and exits can be changed, directional signs can be added, or the layout can be rearranged.

[0118] In some implementations, the comprehensive sharing score includes a vertical sharing score, which considers cross-layer relationships and is obtained by calculating the vertical correlation coefficient and cross-layer behavior transmission rate. Step S600 includes steps S610B to S680B, and the specific steps are as follows: Step S610B: Divide the multi-layered grid cells of the same two-dimensional plane coordinates into a three-dimensional spatial cell.

[0119] Specifically, for example, the ground layer includes two layers, U1 and U2 from top to bottom, the above-ground layer includes five layers, A1, A2, A3, A4 and A5 from bottom to top, and the ground layer is a single layer G1. So there are a total of eight layers from ground layer U2 to above-ground layer A5. The grid cells of the same two-dimensional plane coordinates of the eight layers are divided into a three-dimensional space unit.

[0120] Step S620B: For a single three-dimensional spatial unit, calculate the first three-dimensional time structure entropy of the three-dimensional spatial unit based on the number of its clusters in the ground layer, above-ground layer and underground layer.

[0121] In some embodiments, step S620B includes steps S621B to S622B, and the specific steps are as follows: Step S621B: For a single 3D spatial unit, determine the number of its clusters in time period t. : ; In the formula, The number of all clusters in the aboveground layer. The total number of clusters in the underground layer. This represents the total number of clusters in the ground layer.

[0122] Step S622B: Based on the number of clusters of a single three-dimensional spatial unit , and determine its first three-dimensional time structure entropy.

[0123] Specifically, it is calculated according to the following formula. : ; In the formula, The number of clusters in a single three-dimensional spatial unit. The temporal structure entropy of a single three-dimensional spatial unit.

[0124] The first three-dimensional temporal structure entropy represents the degree of uniformity in the temporal distribution of the total amount of human activity when the entire vertical space is viewed as a plane.

[0125] Step S630B: Calculate the average value of the second three-dimensional time structure entropy based on the time structure entropy of the corresponding grid units of the ground layer, the above-ground layer and the underground layer.

[0126] In some embodiments, step S630B includes steps S631B to S632B, and the specific steps are as follows: Step S631B: Obtain the temporal structure entropy of the corresponding grid cells of a single three-dimensional spatial unit located in the above-ground layer, underground layer, and ground layer.

[0127] You can refer to step S621B to obtain it.

[0128] Specifically, for example, when calculating the above-ground layers, assuming there are seven layers in total, we calculate the number of clusters of individual three-dimensional spatial units located in the corresponding grid units of these seven layers, using the formula... The calculations are performed similarly for underground and surface layers.

[0129] Step S632B: Calculate the average value of the second three-dimensional time structure entropy of the corresponding grid cells of the ground layer, above-ground layer and underground layer.

[0130] The temporal structure entropy of the corresponding grid cells for the ground layer, above-ground layer, and underground layer is added together and then divided by the total number of layers. The total number of layers is the total number of layers spanned by the three-dimensional spatial cell. For example, if there are 7 above-ground layers + 1 underground layer + 2 underground layers, then L 10.

[0131] The second three-dimensional time structure entropy mean represents the average degree of balance of the entire three-dimensional spatial unit in the time dimension when there is no vertical linkage between the layers and they only act independently.

[0132] Step S640B: Determine the vertical correlation coefficient based on the average of the first and second stereoscopic time structure entropies.

[0133] Specifically, the vertical correlation coefficient of a single three-dimensional spatial unit is calculated according to the following formula: ; In the formula, The vertical correlation coefficient of a single three-dimensional spatial unit. The temporal structure entropy of the corresponding grid cell of a single three-dimensional spatial unit located on the ground level. The temporal structure entropy of the corresponding grid cell of a single three-dimensional spatial unit located on the ground layer. The temporal structure entropy of the corresponding grid cell in the underground layer for a single three-dimensional spatial unit. The first three-dimensional time structure entropy of a single three-dimensional spatial unit. This represents the average entropy of the second three-dimensional time structure.

[0134] like If the value exceeds the preset first threshold, it indicates that activities on each floor are staggered and complementary in terms of timing, resulting in a more balanced overall time utilization in the vertical direction. If the value is less than the preset second threshold, and the second threshold is less than the first threshold, it indicates that the peak activity periods for each floor overlap, which may lead to vertical traffic congestion or resource competition. If the value is between the first and second thresholds, it means that there is no significant complementarity or overlap between the floors.

[0135] Optionally, the first threshold is set to 1.2 and the second threshold is set to 1.

[0136] Specifically, the vertical correlation coefficient of each three-dimensional spatial unit can be calculated in the manner described above. Finally, the vertical correlation coefficient of the target three-dimensional area can be obtained by summing the vertical correlation coefficients and dividing by the total number of three-dimensional spatial units. And normalize it.

[0137] Step S650B: Obtain the movement trajectory of each individual and extract all cross-level trajectory information of all individuals passing through vertical transportation equipment. The cross-level trajectory information includes the departure floor, the arrival floor, and the corresponding behavior type.

[0138] Specifically, it acquires the movement trajectory of each individual within the target three-dimensional area.

[0139] Specifically, vertical transportation equipment includes escalators, elevators, stairs, ramps, etc.

[0140] Specifically, the trajectory information includes the floor at departure, the floor at arrival, the time period of crossing floors, the type of activity on the departure floor before crossing floors, and the type of activity on the arrival floor after crossing floors.

[0141] Step S660B: Based on the cross-floor trajectory information, construct a behavior transition matrix between adjacent floors, wherein the elements of the behavior transition matrix are... , Indicates the number of floors from the starting floor The behavior changes to reaching the floor number. The number of people engaging in this type of behavior.

[0142] Specifically, for each pair of adjacent floors, all cross-floor related trajectory information is collected throughout the entire time period, and a behavior transition matrix is ​​constructed. In this embodiment, there are five types of activities to be collected, which can be set according to requirements. Therefore, the matrix is ​​k. k form, and =5, specifically, the matrix is ​​as follows.

[0143] matrix: ; Corresponding to the five behavior categories of passage, rest, dining, exercise, and children's activities in this application, they can be set as needed, namely 1, 2, 3, 4, and 5, matrix elements. , indicating the number of floors from the starting floor The behavior changes from crossing floors to reaching the next floor. The total number of people exhibiting this type of behavior. The types of behaviors also correspond to the five types of behaviors in this application: passage, rest, dining, exercise, and children's activities, which are respectively 1, 2, 3, 4, and 5.

[0144] For example, This indicates the number of people whose movement from the starting floor to the arriving floor remains a movement activity. This indicates the number of children whose activity shifts from traveling to arriving at the floor.

[0145] This is used to construct the behavior transition matrix for each pair of adjacent floors.

[0146] Step S670B: Based on the behavior transition matrix, determine the transition ratio of each type of behavior from the starting floor to the next floor. .

[0147] Calculate in sequence , ... The proportion, for example, calculating It can be done The number of people engaging in this type of behavior, i.e., the number of people whose common behavior was converted into common behavior, divided by + +…… Total number of people calculation ratio .

[0148] For example, if there are six floors in total, calculate the percentage of people on each type of behavior from the sixth to the fifth floor, the fifth to the fourth floor, and so on, from the second to the first floor, as well as from the first to the second floor, the second to the third floor, and so on, from the fifth to the sixth floor, out of the total number of people. .

[0149] Specifically, each type of behavior is , ... .

[0150] Specifically, calculation Then, it is normalized to obtain the final comprehensive score for vertical sharing. It falls within the [0,1] interval, which facilitates horizontal comparison of different scenarios.

[0151] Step S680B: Determine the comprehensive score for vertical sharing based on the vertical correlation coefficient and the transfer ratio.

[0152] Specifically, the comprehensive score for vertical sharing is calculated according to the following formula. : + ; In the formula, for The total number of activity types for The total number of activity types The percentage of each type of behavior that travels from the starting floor to the next floor. The weight coefficient for each type of behavior, Normalization has been performed, and D is the vertical correlation coefficient. This is the weighting coefficient for the vertical correlation coefficient.

[0153] Specifically, This can be represented as bidirectional or unidirectional cross-floor behavior between adjacent floors. If it is bidirectional, and there are N floors (i.e., from the first floor to the Nth floor plus from the Nth floor to the first floor), it needs to be divided by 2.

[0154] for The possible values, for example, for , , as well as This type of behavior, where the act of passing through is transformed into another behavior of staying, can be given a higher weight, such as 0.3. For , , , , , , , , , , , , , , as well as These types of behaviors—where staying behavior transforms into staying behavior, and shared behavior continues across different floors—are given the highest weight, for example, 0.5. For , , , This type of behavior, where the customer leaves after sharing, can be given a lower weight, such as 0.15. For This type of behavior, where a customer switches from a common behavior to a common behavior, is given the lowest weight, 0.05. The weighting is determined based on the contribution to the continuation of the shared behavior. The value is 0.2.

[0155] right Normalization was performed, if If the value exceeds a preset threshold (optionally, a preset threshold of 0.7 indicates good vertical sharing, good continuity of sharing behavior, users are more willing to move between different floors and continuously use space resources, and activities between floors are relatively complementary). If the value is less than a preset threshold (optionally, the preset threshold is 0.4), it is determined to be poor vertical sharing, indicating that users are less willing to move between different floors and continuously use space resources, and there is a high overlap of activities between floors. The optimization suggestion is to optimize the layout of vertical transportation nodes or optimize the activity types and corresponding commercial facilities on each floor. Between 0.4 and 0.7, users are generally less willing to move between different floors and continuously use space resources.

[0156] Calculate the comprehensive score for planar sharing and / or vertical sharing comprehensive score After that, the comprehensive score for shared access can be determined. The specific steps are as follows: Specifically, a comprehensive score can be given based solely on the degree of shared functionality. Determine the comprehensive score for shared benefits Or based on a comprehensive score for vertical sharing. Determine the comprehensive score for shared benefits .

[0157] In some implementations, a comprehensive score based on planar sharing is used. Comprehensive score for vertical sharing Determine the comprehensive score for shared benefits The overall score for shared accessibility is calculated using the following formula. : Shared comprehensive score ; In the formula, For shared comprehensive scoring, For the overall score of planar sharing, As the first coefficient, A comprehensive score for vertical sharing. This is the second coefficient.

[0158] Specifically, and All have undergone normalization.

[0159] Specifically, and The value is 0.5, and can be set according to requirements.

[0160] In some implementations, the three-dimensional spatial sharing measure method 1 based on spatiotemporal clustering entropy provided in this application further includes step S700: constructing a spatiotemporal measurement model of sharing.

[0161] In some implementations, step S700 includes steps S710 to S750: Step S710: Obtain the number of different types of clusters and the coordinates of each cluster for each grid cell in each time period; Specifically, it can be obtained through step S200.

[0162] Step S720: Obtain the average number and average coordinate offset of different types of clusters for each grid cell across all time periods.

[0163] Specifically, It is the i-th grid cell within the entire time period. The average number of clusters of each type can be calculated using the following formula: ; In the formula, This represents the total number of time periods.

[0164] Specifically, It is the i-th grid cell within the entire time period. The average coordinate offset of each cluster type can be calculated using the following formula: ; In the formula, This represents the total number of time periods.

[0165] Step S730: Determine the correlation coefficient between clusters and time rhythm for each time period based on the number of different types of clusters, cluster coordinates, average number of clusters, and average coordinate offset. .

[0166] ; In the formula, For the i-th grid cell in the j-th time period The correlation coefficients between the cluster types and the time rhythm The i-th grid cell in the j-th time period is... The number of clusters of each type It is the i-th grid cell within the entire time period. The average number of clusters of each type The values ​​are sequentially taken as 1, 2, ..., J, where J is the total number of time periods.

[0167] in, .

[0168] Calculate the Pearson correlation coefficient between the coordinate offset and the average offset for each time period. .

[0169] ; In the formula, For the i-th grid cell in the j-th time period The correlation coefficients between the coordinate offsets of each type of cluster and the time rhythm For the i-th grid cell in the j-th time period Coordinate offsets of various types For the i-th grid cell in the entire time period, the th The average coordinate offset of each type of cluster.

[0170] Step S740: Determine the core cluster type and auxiliary cluster type based on the correlation coefficient, wherein the correlation coefficient corresponding to the core cluster type is greater than a preset third threshold.

[0171] Specifically, the third threshold can be set to 0.7, when Greater than 0.7, and when When the value is greater than 0.7, it is determined that the cluster type has a significant temporal rhythm correlation with the behavior of the grid cell. The cluster type with the highest proportion in that time period is identified as the core cluster type, and the other cluster types are classified as auxiliary cluster types.

[0172] Step S750: Construct a shared spatiotemporal measurement model based on the core cluster type and the auxiliary cluster type.

[0173] In some implementations, based on the above analysis results, a spatiotemporal measurement model is constructed, with spatial coding as the core dimension and incorporating temporal rhythm features. This model is used to output the first shared association total score, the second shared association total score, and the spatial shared basic association value.

[0174] Based on the formula for behavior occupancy rate in step S330, the behavior occupancy rate corresponding to the clusters of the core behavior types can be calculated separately, and the behavior occupancy rate corresponding to the clusters of the auxiliary behavior types can also be calculated separately. The sum of the core behavior occupancy rates is then calculated using the following formula. : ; In the formula, The total percentage of core behaviors. To determine the percentage of different types of behaviors categorized into core behavior types, This represents the total number of time periods.

[0175] ; In the formula, The total percentage of auxiliary behaviors. To determine the percentage of different types of behaviors categorized as auxiliary behaviors, This represents the total number of time periods.

[0176] The first shared association score is obtained by weighted summation of the total share of core behaviors and the total share of auxiliary behaviors, with the weight of the total share of core behaviors being higher than the weight of the total share of auxiliary behaviors, and the sum of the two weights being one.

[0177] The contribution of each type to the mixing degree is calculated using the following formula: In the formula, For the contribution of class k to the degree of mixing, This represents the percentage of clusters of the k-th activity type within the grid during that time period.

[0178] Summing over time periods yields the total core mixing degree of grid i. Total degree of mixing with auxiliary The specific formula is as follows: ; In the formula, For the first The contribution of a class to the degree of mixing can be expressed as follows: The first grid The number of clusters in the th order The time period, representing the proportion of all types of clusters. As the core behavior type, if If it belongs to the core behavior type, then sum it up.

[0179] ; In the formula, For the first The contribution of a class to the degree of mixing can be expressed as the contribution of the first class to the degree of mixing. The first grid The number of clusters in the th order The time period, representing the proportion of all types of clusters. For auxiliary behavior types, i.e., if If it belongs to the auxiliary behavior type, then sum it up.

[0180] The second shared correlation score is obtained by weighted summation of the core mixture score and the auxiliary mixture score, with the weight of the core mixture score being higher than the weight of the auxiliary mixture score, and the sum of the two weights being one.

[0181] In some implementations, a spatiotemporal measurement model is constructed based on the first and second shared correlation total scores, with spatial coding as the core dimension and incorporating temporal rhythm features. This model focuses on core cluster types as the primary analysis object and auxiliary cluster types as the comparative analysis object. By comparing the differences between the two types of clusters in terms of time occupancy and activity type mixing, the behavioral rhythm characteristics of each grid unit or three-dimensional area are identified, providing a reference for the temporal optimization of spatial functions.

[0182] In some implementation methods, grid units with lower sharing scores are selected for renovation potential assessment to obtain renovation and optimization costs and implementation feasibility scores. The renovation and optimization costs can be estimated through renovation plans. Specifically, based on the aforementioned parameters such as vertical sharing comprehensive score, activity adaptability, time sharing balance, spatial sharing balance, renovation and optimization costs, and implementation feasibility scores, corresponding renovation plans can be formulated for each grid unit. Renovation costs are estimated based on the renovation plans. Experts in urban planning, engineering management, and other fields are invited to comprehensively score the renovation plans based on factors such as technical difficulty, construction period, and impact on residents; the average score is used to obtain the implementation feasibility score. A comprehensive evaluation index is calculated based on the renovation and optimization costs and the implementation feasibility score.

[0183] In summary, the three-dimensional spatial sharing measure based on spatiotemporal clustering entropy provided in this application has the following advantages: 1. This application's embodiments achieve comprehensive perception of three-dimensional spatial units by separately acquiring crowd activity data at the ground level, above-ground level, and underground level, and independently performing grid division and cluster analysis. Compared to traditional methods that only focus on a two-dimensional plane or a single level, this solution can accurately capture the vertical distribution of crowd activity, reflecting the shared usage status of the three-dimensional space.

[0184] 2. This application's embodiments clearly distinguish between passage behavior and dwelling behavior, and perform cluster analysis separately, enabling subsequent occupancy calculations and entropy analysis to reflect the differentiated contributions of different types of activities to spatial sharing. Passage behavior reflects the fluidity of spatial sharing, while dwelling behavior reflects the resident nature of spatial sharing, resulting in more comprehensive parameters.

[0185] 3. This application's embodiments calculate temporal structure entropy and spatial structure entropy. Temporal structure entropy quantifies the uniformity of distribution of various activities throughout the day, reflecting spatial utilization efficiency at different times. Spatial structure entropy quantifies the balanced distribution of various activities across different grid cells, identifying whether activities are excessively concentrated in a few areas, reflecting the overall utilization efficiency of spatial resources. The combination of these two forms a dual-dimensional evaluation system of time and space, overcoming the limitations of traditional single-indicator evaluations.

[0186] 4. The embodiments of this application calculate a comprehensive score for vertical sharing. The comprehensive score for vertical sharing is used to quantify the transmission rate of shared behavior between adjacent floors due to the cross-floor flow of people, and can be used to reflect the degree of vertical coupling of the target three-dimensional area.

[0187] Please see Figure 2 , Figure 2 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. For example... Figure 2 As shown, the electronic device 400 includes: one or more processors 410 and a memory 420. Figure 2 Take a processor 410 as an example.

[0188] In some implementations, the processor 410 and the memory 420 may be connected via a bus or other means. Figure 2 Taking the example of a connection between China and Israel via a bus.

[0189] In some implementations, the processor 410 is configured to acquire crowd activity data in the public spaces of the ground floor, above-ground floor, and underground floor of the target three-dimensional area; determine crowd activity data for various behaviors in each time period based on the crowd activity data, wherein the crowd activity data includes activity type, activity quantity, and activity coordinates; perform cluster analysis on the various behaviors to obtain the cluster type, cluster quantity, and cluster coordinates for each floor and time period; divide the public spaces of the ground floor, above-ground floor, and underground floor into multiple grid units respectively, and calculate the behavior occupancy rate of each grid unit in each time period based on the cluster type, cluster quantity, and cluster coordinates; for each grid unit, calculate the temporal structure entropy of the grid unit based on its behavior occupancy rate in each time period, the temporal structure entropy being used to quantify the balance of crowd activity distribution in the temporal dimension within the grid unit; for each time period, calculate the spatial structure entropy of the time period based on the behavior occupancy rate of each grid unit in that time period, the spatial structure entropy being used to quantify the balance of crowd activity distribution in the spatial dimension within that time period; and generate a comprehensive score for the vertical sharing of the public space based on the temporal structure entropy and spatial structure entropy.

[0190] In some implementations, memory 420 serves as a non-volatile computer-readable storage medium, used to store non-volatile software programs, non-volatile computer-executable programs, and modules, such as the program instructions / modules of the three-dimensional spatial sharing measure method based on spatiotemporal clustering entropy in the embodiments of this application. Processor 410 executes various functional applications and data processing of electronic device 400 by running the non-volatile software programs, instructions, and modules stored in memory 420, thereby implementing the three-dimensional spatial sharing measure method based on spatiotemporal clustering entropy described in the above method embodiments.

[0191] In some embodiments, memory 420 may include a program storage area and a data storage area, wherein the program storage area may store the operating system and applications required for at least one function; the data storage area may store data created based on the use of electronic device 400, etc. Furthermore, memory 420 may include high-speed random access memory, and may also include non-volatile memory, such as at least one disk storage device, flash memory device, or other non-volatile solid-state storage device. In some embodiments, memory 420 may optionally include memory remotely located relative to processor 410, and this remote memory may be connected to the controller via a network. Examples of such networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof.

[0192] In some implementations, one or more modules are stored in memory 420 and, when executed by one or more processors 410, perform the spatial sharing measure method based on spatiotemporal clustering entropy in any of the above method embodiments, for example, performing the above-described... Figure 1 The method steps S100 to S600.

[0193] In some implementations, the electronic device can be a chip, such as a data processor (DPU) chip used in a data center, or the electronic device can be a network interface card including a chip and multiple interfaces (such as PCI / PCIE interface, UART interface, USB interface, etc.), or the electronic device can be a traditional server, or a server including a network interface card or chip. The server includes a host and a data processor. The data processor is used to schedule packets to the host or the data processor itself for processing, and the host is used to process the packets scheduled by the data processor.

[0194] Please refer to Figure 3 , Figure 3 This is a structural block diagram of a computer-readable storage medium provided in an embodiment of this application. The computer-readable storage medium 500 stores program code 510, which can be called by a processor to execute the three-dimensional spatial sharing measure based on spatiotemporal clustering entropy described in the above method embodiments.

[0195] The computer-readable storage medium 500 may be an electronic memory such as flash memory, EEPROM (Electrically Erasable Programmable Read-Only Memory), EPROM, hard disk, or ROM. Optionally, the computer-readable storage medium includes a non-volatile computer-readable storage medium. The computer-readable storage medium 500 has storage space for program code that performs any of the method steps of the above-described method for measuring the spatial sharing of space based on spatiotemporal clustering entropy. This program code can be read from or written to one or more computer program products. The program code may, for example, be compressed in an appropriate form.

[0196] In summary, this application provides a method, device, and medium for measuring the shared nature of three-dimensional space based on spatiotemporal clustering entropy. The method includes acquiring crowd activity data from the public spaces of the ground floor, above-ground floor, and underground floor of a target three-dimensional area; determining crowd activity data for various behaviors at each time period based on the crowd activity data, wherein the crowd activity data includes activity type, activity quantity, and activity coordinates; performing cluster analysis on the various behaviors to obtain the cluster type, number of clusters, and cluster coordinates for each floor and time period; and dividing the public spaces of the ground floor, above-ground floor, and underground floor into multiple networks. This application constructs a measurement system for the sharing of urban three-dimensional spatial units by using three-dimensional hierarchical data collection, separate clustering of passage and dwelling behaviors, and calculation of temporal and spatial dual-dimensional entropy values. This method distinguishes between passage and dwelling behaviors and uses temporal and spatial entropy to quantify the sharing of public space, providing a decision-making basis for spatial optimization of three-dimensional areas.

[0197] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application.

Claims

1. A method for measuring the sharing of three-dimensional space based on spatiotemporal clustering entropy, characterized in that, include: Acquire crowd activity data in the public spaces of the ground floor, above-ground floor, and underground floor of the target three-dimensional area. The crowd activity data includes activity type, activity quantity, and activity coordinates. Cluster analysis was performed on the crowd activity data to obtain the cluster type, number of clusters, and cluster coordinates for each layer and time period; The common spaces of the ground floor, above-ground floor, and underground floor are divided into multiple grid units. Based on the cluster type, the number of clusters, and the coordinates of the clusters, the behavior occupancy rate of each grid unit in each time period is determined. For each grid cell, the temporal structure entropy of the grid cell is determined based on its behavior occupancy rate in each time period. The temporal structure entropy is used to quantify the balance of the distribution of crowd activities in the time dimension within the grid cell. For each time period, the spatial structure entropy of that time period is determined based on the behavior occupancy rate of each grid unit in that time period. The spatial structure entropy is used to quantify the balance of the spatial distribution of crowd activities within that time period. Based on the temporal structure entropy and the spatial structure entropy, a comprehensive score for the sharing of public space is generated.

2. The method for measuring three-dimensional spatial sharing based on spatiotemporal clustering entropy according to claim 1, characterized in that, After acquiring crowd activity data in the public spaces of the ground floor, above-ground floor, and underground floor of the target three-dimensional area, wherein the crowd activity data includes activity type, activity quantity, and activity coordinates, the method further includes: Based on the crowd activity data, extract the feature information of each individual; Based on the aforementioned feature information, construct a behavioral feature vector for each individual; Based on the behavioral feature vector of each individual, it is determined whether the behavior type of each individual belongs to the stationary behavior or the passage behavior.

3. The method for measuring three-dimensional spatial sharing based on spatiotemporal clustering entropy according to claim 1, characterized in that, The cluster analysis performed on the various behaviors yields the cluster types for each layer and time period, including: The aforementioned behaviors are categorized into passage behaviors and stay behaviors; Cluster analysis was performed on the aforementioned traffic behavior to obtain traffic clusters for each layer and time period; Cluster analysis is performed on the dwelling behavior to obtain dwelling behavior clusters at each level and time period, and the type of each dwelling behavior cluster is determined. The types of dwelling behavior clusters include at least one of rest clusters, dining clusters, sports clusters, and children's activity clusters.

4. The method for measuring three-dimensional spatial sharing based on spatiotemporal clustering entropy according to claim 3, characterized in that, Also includes: Cluster analysis was performed on the dwelling behavior to obtain clusters of dwelling behavior at each level and time period; Obtain activity tags for the public spaces of the ground floor, above-ground floor, and underground floor of the target three-dimensional area, where each activity tag corresponds to a location coordinate. Match the coordinates of the dwell behavior clusters with the location coordinates corresponding to the activity tags; Determine the type of cluster for the dwell behavior.

5. The method for measuring three-dimensional spatial sharing based on spatiotemporal clustering entropy according to claim 1, characterized in that, The step of determining the behavior occupancy rate of each grid cell in each time period based on the cluster type, the number of clusters, and the cluster coordinates includes: Based on the coordinates of the clusters, determine their corresponding grid cells; Based on the number of clusters, the behavior occupancy of each grid cell in each time period is determined.

6. The method for measuring three-dimensional spatial sharing based on spatiotemporal clustering entropy according to claim 1, characterized in that, Also includes: Obtain the number of different types of clusters and the coordinates of each cluster in each time period for each grid cell; Obtain the average number of different types of clusters and the average coordinate offset for each grid cell across all time periods; The correlation coefficient between clusters and time rhythm in each time period is determined based on the number of different types of clusters, the coordinates of the clusters, the average number of clusters, and the average coordinate offset. The core cluster type and auxiliary cluster type are determined based on the correlation coefficient, wherein the correlation coefficient corresponding to the core cluster type is greater than a preset third threshold. Based on the core cluster type and the auxiliary cluster type, a shared spatiotemporal measurement model is constructed.

7. The method for measuring three-dimensional spatial sharing based on spatiotemporal clustering entropy according to claim 1, characterized in that, The shared access score includes a planar shared access score, which is generated based on the temporal structure entropy and the spatial structure entropy, including: The temporal sharing balance of the common space is determined based on the temporal structure entropy of each grid cell. The spatial sharing balance of the public space is determined based on the spatial structure entropy of each grid cell. Based on the time-sharing balance and the spatial-sharing balance, a comprehensive planar sharing score is determined for each grid cell.

8. The method for measuring three-dimensional spatial sharing based on spatiotemporal clustering entropy according to claim 1, characterized in that, The comprehensive sharing score includes a vertical sharing score. Based on the temporal structure entropy and the spatial structure entropy, a vertical sharing score for the public space is generated, including: Divide the grid cells of the same two-dimensional plane coordinates of multiple layers into a three-dimensional space cell; For a single three-dimensional spatial unit, the first three-dimensional time structure entropy of the three-dimensional spatial unit is calculated based on the number of its clusters in the ground layer, above-ground layer and underground layer. Calculate the average time structure entropy of the second three-dimensional structure based on the time structure entropy of the corresponding grid units of the ground layer, above-ground layer and underground layer. The vertical correlation coefficient is determined based on the average of the first and second three-dimensional time structure entropies. Acquire the movement trajectory of each individual and extract the trajectory information of all individuals passing through vertical transportation equipment; Construct a behavior transition matrix based on the trajectory information; Based on the behavior transition matrix, determine the transition ratio of each type of behavior from the starting floor to the next floor; A comprehensive score for vertical sharing is determined based on the vertical correlation coefficient and the transfer ratio.

9. An electronic device, comprising at least one processor; and a memory communicatively connected to said at least one processor; wherein, The memory stores instructions executable by the at least one processor, which, when executed by the at least one processor, enables the at least one processor to perform the three-dimensional spatial sharing measure based on spatiotemporal clustering entropy as described in any one of claims 1 to 8.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores an executable program, which is executed by a processor to implement the three-dimensional spatial sharing measure based on spatiotemporal clustering entropy as described in any one of claims 1 to 8.

Images