A dynamic region division method based on active coordinate data

Abstract

The present invention discloses a dynamic area division method based on activity coordinate data, which belongs to the field of big data processing and aims to achieve efficient processing of coordinate data and dynamic area identification. The method first collects the activity coordinate data of multiple active subjects in the target area within a certain time period, removes outliers and noise interference through data preprocessing, and ensures data accuracy. Then, the activity area is preliminarily divided into multiple sub-areas based on the distribution characteristics of the coordinates. On this basis, the sub-areas are fused in combination with time and space density factors, and the sub-area boundaries are dynamically updated to adapt to the behavior patterns of the active subjects that change over time. The present invention can flexibly divide the area according to the actual activity situation accurately and in real time, effectively improve the efficiency of space utilization, and can be widely used in many fields such as indoor space optimization layout, providing strong support for the rational allocation of resources.

Description

Technical Field

[0001] The present invention belongs to the field of big data processing, and in particular is a region division method based on activity coordinate data. Background Art

[0002] With the rapid development of indoor positioning technologies (such as Wi-Fi positioning, Bluetooth beacons, UWB positioning, and image analysis), it's becoming increasingly easy to collect coordinate data on people's movements indoors. This data can reflect people's movement trajectories and dwell time in real time within indoor environments, encompassing the characteristics and patterns of social interactions between people. This provides a crucial basis for indoor space management and activity optimization. Therefore, how to effectively utilize this positioning data to accurately characterize activity patterns and analyze the social characteristics embedded in these activity trajectories remains an urgent challenge.

[0003] Generally speaking, analyzing the social patterns of people's activities within an area first requires demarcating activity areas and identifying individuals participating in various group activities, such as the different game items in a children's play session. These areas are often manually defined in advance, which cannot quickly adapt to changes in children's activity hotspots and interests, making automated analysis and processing impossible. Furthermore, manual labeling methods are subject to high subjectivity and low accuracy, making them difficult to accurately reflect actual activity patterns. This is especially true in scenarios where activity hotspots frequently change, where manual labeling errors are further amplified. In recent years, some studies have attempted to perform dynamic area segmentation based on indoor activity data. For example, clustering algorithms have been used to analyze indoor positioning data to generate dynamic areas. However, these methods typically focus only on the spatial dimension, ignoring the temporal dimension and changes in activity density, resulting in inaccurate segmentation results. Furthermore, existing methods often suffer from low computational efficiency and poor real-time performance when processing large-scale indoor data, making them difficult to meet practical application requirements. Especially with rapidly increasing data volumes, algorithm complexity and computational resource consumption become major bottlenecks.

[0004] Therefore, there is an urgent need for a dynamic area division method that can comprehensively consider space, time and activity density to better adapt to the actual needs of indoor scenes. Summary of the Invention

[0005] This paper proposes a dynamic area segmentation method based on activity coordinate data, which has the advantages of efficiently processing coordinate data and identifying activity areas in real time. Through steps such as data preprocessing, preliminary sub-area segmentation, and spatiotemporal density fusion, combined with efficient algorithms and real-time data processing technology, the method can achieve dynamic and precise segmentation of indoor areas.

[0006] Technical solution: A dynamic area division method based on active coordinate data, comprising the following steps:

[0007] S1. Acquire continuous coordinate data of multiple target objects within a certain geographical area and a certain data collection period within a given time interval, and then obtain real-time coordinate data of the target objects at a certain moment;

[0008] S2, preprocessing the acquired real-time coordinate data of the target object;

[0009] S3, dividing the pre-processed coordinate data into dynamic time windows, and preliminarily dividing the geographical area where the coordinate data corresponding to each dynamic time window is located into multiple sub-areas;

[0010] S4, based on the stored preliminary region division results and sub-region overlap, performing a region fusion operation to dynamically update the sub-region boundaries;

[0011] S5. For the results after regional fusion processing, in order to avoid the regional division results being inconsistent with the actual area due to sparse data points in the edge area, deep learning is used to further optimize the regional division according to the density distribution of target objects in each sub-area to improve the division accuracy.

[0012] S6. After completing the above steps, the final dynamic area division result can be obtained.

[0013] Furthermore, in S1, the real-time coordinate data includes the ID, three-dimensional coordinates and time information corresponding to the three-dimensional coordinate position of the target object.

[0014] Furthermore, in S1:

[0015] (1) The method of obtaining continuous coordinate data includes UWB positioning, image processing, sensor positioning or other technologies.

[0016] (2) When obtaining coordinates, a certain point in the geographic area is used as the coordinate origin, and then the real-time coordinate data of the target object at a certain moment is obtained. The obtained real-time coordinate data needs to be able to distinguish the position of a target object at different times.

[0017] (3) The time interval setting depends on the activity type of the target object. For situations with large amounts of activity, the time interval should be reduced. The specific time interval value should be selected based on the results of subsequent data analysis.

[0018] Furthermore, in S2, the data preprocessing method includes removing duplicate values, missing values, and abnormal values in the coordinate data, and screening the real-time coordinate data of the target object that meets the conditions;

[0019] Duplicate values refer to the situation where a target object has multiple position coordinates at the same time due to the influence of the acquisition equipment. It is necessary to retain the correct coordinate data based on the position coordinates of the previous and next moments and perform deduplication processing;

[0020] Missing values refer to situations where some data is not correctly recorded or transmitted due to reasons such as equipment failure, sensor failure, or network problems;

[0021] Outliers refer to coordinate data that exceeds the actual movement capability of the target object within a set time interval, identified in the collected coordinate data through in-depth analysis of the target object's behavioral characteristics.

[0022] Screening for eligible target object data refers to filtering the collected target object coordinate data based on the preset target area boundaries to extract eligible target object coordinate data. The eligible target real-time coordinate data refers to data whose coordinate data consistently appears within the target geographic area during the complete data collection cycle. The target geographic area refers to the entire area to be zoned.

[0023] Furthermore, in S2, to improve the accuracy and reliability of coordinate data, an intelligent coordinate correction module is introduced to further optimize the coordinate data after deduplication for missing values and outliers. The specific method is as follows:

[0024] 1) Build an intelligent coordinate correction model based on a generative adversarial network (GAN), leveraging a large amount of historical coordinate data for learning and training. Through adversarial training of the generator and discriminator, the model automatically scans coordinate data and identifies abnormal coordinates that do not conform to the target object's behavioral characteristics, such as coordinate data that exceeds the target object's actual mobility within a set time interval.

[0025] 2) For identified anomalous coordinates, the model automatically generates more reasonable coordinate values based on contextual information (such as normal coordinate data from the preceding and following moments), effectively optimizing the coordinate data. The corrected coordinate data is then compared and verified with the original data to ensure the accuracy and reliability of the correction results.

[0026] By introducing the intelligent coordinate correction module, not only can the quality of coordinate data be effectively improved and coordinate data closer to the actual situation be generated, thus avoiding complex manual operations, but it also has strong generalization capabilities, providing a reliable basis for subsequent data analysis and application, thereby further enhancing the overall performance and reliability of the system.

[0027] Furthermore, the specific steps of S3 are as follows:

[0028] S3.1. Adopt a dynamic adjustment algorithm based on sliding windows to adaptively adjust the size and step size of the time window to better capture the key features in the time series. The specific method is as follows:

[0029] S3.1.1. Based on the dynamic characteristics of coordinate data, dynamically adjust the time window size by analyzing the target object's behavior patterns and data fluctuations. During periods of frequent target activity or drastic data changes, reduce the time window size to capture more details; during periods of stable target activity or slow data changes, increase the time window size appropriately to reduce the frequency of data processing. The window size adjustment formula is as follows:

[0030] (1)

[0031] in, w is the size of the current sliding time window, w min and w max are the minimum and maximum values of the window size, data_variance ( D [ i : i + w ]) refers to the variance of the data within the window, which indicates the volatility of the data within the window. D [i:i+ w ] is the window from the i Data points to i + w A sequence of data points, θ high is the data high fluctuation threshold, θ low is the low fluctuation threshold of data;

[0032] S3.1.2. Dynamically adjust the step size of the time window based on the continuity and change trend of the data. During periods of rapid data change, use a smaller step size to ensure data continuity; during periods of slow data change, use a larger step size to improve partitioning efficiency. The step size adjustment formula is as follows:

[0033] (2)

[0034] in, s is the step size of the current sliding time window, s min and s max are the minimum and maximum values of the step size, data_rate ( D [ i : i + w ]) refers to the rate of change of data within the window, indicating the speed of change of data within the window. D [ i : i + w] is the window from the i Data points to i + w A sequence of data points, θ rate is the threshold of data change rate.

[0035] S3.2. Perform preliminary geographic division on the coordinate data within each dynamic time window, generate a set of sub-regions corresponding to each time period, and store the preliminary regional division results.

[0036] The preliminary region division methods include, but are not limited to, the K-Means clustering algorithm, the DBSCAN clustering algorithm, the grid division method, and the division method based on the minimum spanning tree. Through the above steps, the present invention can effectively implement the spatiotemporal segmentation processing of coordinate data, dynamically adapt to the changing characteristics of the data, provide more accurate basic data support for subsequent dynamic region division, and further enhance the flexibility and accuracy of the system.

[0037] Furthermore, the specific steps of S4 are as follows:

[0038] S4.1. Perform kernel density estimation on the data points within each subregion. Each preliminary partitioning result contains multiple subregions, each defined by its boundary coordinates (x1, y1, x2, y2), where x1 and y1 are the coordinates of the lower left corner and x2 and y2 are the coordinates of the upper right corner. Perform kernel density estimation (KDE) on the data points within each subregion and calculate the density value of each data point. The kernel density estimation formula is:

[0039] (3)

[0040] in: is a data point x The kernel density estimate of K is the kernel function used to measure the data points x A data point in the neighborhood x i For data points x The contribution weight of h is the bandwidth, which is used to control the smoothness of the kernel function. n is the total number of data points in the current sub-region (i.e. the number of samples involved in density calculation), d It is the data dimension.

[0041] By this formula, for each target point x , calculate all data points in its neighborhood x i The weighted contribution of K bandwidth h and number of samples nNormalize and get the kernel density estimate .

[0042] S4.2. Based on the kernel density estimate, the data points in the sub-region are divided into high-density regions and low-density regions. The kernel density estimate is higher than the preset density threshold. θ high The sub-region is a high-density area; the kernel density estimate is lower than the preset density threshold θ low The sub-region is a low-density area.

[0043] S4.3. Arrange the stored preliminary region division results in chronological order, and compare the coordinates of all sub-regions in the preliminary division results of two adjacent groups in turn. If a sub-region in the preliminary region division results of one group overlaps with a sub-region in the results of another group, calculate the kernel density estimate of the overlapping area. If the overlapping area is greater than or equal to the dynamic fusion threshold, perform the region fusion operation.

[0044] If a sub-region in the preliminary regional division results of one group overlaps with a sub-region in the results of another group, the specific steps to be performed are as follows:

[0045] S4.3.1. Calculate the overlapping area R of two sub-areas overlap :

[0046] (4)

[0047] in, R overlap is the overlapping area of the two sub-areas, (x1, y1, x2, y2) are the boundary coordinates of the first sub-area, (x1, y1) are the lower left corner coordinates, (x2, y2) are the upper right corner coordinates, (x1′, y1′, x2′, y2′) are the boundary coordinates of the second sub-area, (x1′, y1′) are the lower left corner coordinates, and (x2′, y2′) are the upper right corner coordinates.

[0048] S4.3.2. Calculate the area A of the overlapping region overlap :

[0049] (5)

[0050] in, A overlap is the area of the overlapping region of the two subregions, is the left boundary of the overlapping area, is the left boundary of the overlapping area, is the right boundary of the overlapping area, is the upper boundary of the overlapping area.

[0051] S4.3.3. Use formula (3) to calculate the kernel density estimate of the overlapping area, and dynamically adjust the fusion threshold θ according to the kernel density estimate of the overlapping area. fuse :

[0052] (6)

[0053] in, θ fuse is the dynamic fusion threshold, θ high and θ low are the preset high density threshold and low density threshold, respectively, used to classify density levels. Represents the kernel density estimate within the overlapping region.

[0054] S4.3.4, if the overlapping area A overlap Greater than or equal to the dynamic fusion threshold θ fuse , the regional fusion operation is performed; otherwise, the sub-region is directly retained and saved as an independent new sub-region in the new regional division result.

[0055] In S4.3.4, the steps for performing the region fusion operation are:

[0056] (1) For sub-regions that meet the fusion conditions, the adaptive density region fusion clustering algorithm is applied for clustering. According to the kernel density estimation value, the parameters of the adaptive density region fusion clustering are dynamically adjusted. and MinPts:

[0057] (7)

[0058] in, is the neighborhood radius, which is used to define the neighborhood range of the data point. and are the neighborhood radius parameters for high-density and low-density areas, respectively. MinPts is the minimum number of neighborhood points required for a core point. large and MinPts small are the minimum number of points in high-density and low-density areas, respectively. is the kernel density estimate in the overlapping region, θ high is the preset high-density threshold.

[0059] (2) For data points in the overlapping area, determine whether they are core points based on their density value and the number of points in the neighborhood. If a point is a core point, then starting from that point, expand the clustering through the density reachability relationship.

[0060] (3) For the subclusters formed by adaptive density region fusion clustering, determine whether they meet the fusion conditions. If they meet the fusion conditions (such as the density distribution of the subclusters is continuous and the kernel density estimation value is high), these subclusters are merged into a new subregion; if they do not meet the fusion conditions, these subclusters are retained as independent subregions.

[0061] S4.4. For the merged sub-regions, dynamically adjust their boundaries based on the kernel density estimate, and save the newly generated sub-regions to the new region division result. The method for dynamically adjusting the sub-region boundaries is:

[0062] The four corner coordinates (x1, y1, x2, y2) and (x 1’ 、y 1’ 、x 2’ 、y 2’ ), where (x1, y1) and (x2, y2) are the lower left corner and upper right corner coordinates of the first sub-region respectively, (x 1’ ,y 1’ ) and (x 2’ ,y 2’ ) are the lower left corner coordinates and upper right corner coordinates of the second sub-region, respectively, extract the minimum and maximum values in the x and y directions, and generate the new sub-region boundary coordinates (x min 、y min 、x max 、y max ):

[0063] x min =min(x1,x 1’ ), y min =min(y1,y 1’ );

[0064] x max =max(x2,x 2’ ), y max =max(y2,y 2’ );

[0065] Each fused sub-region coordinate contains a unique region number and the fused four-corner coordinates.

[0066] S4.5. After completing the region fusion operation for two adjacent sets of preliminary region division results, further perform a region fusion operation on all subregions in the new region division result. This region fusion operation is performed directly based on the new region division result, eliminating the need to repeatedly save intermediate results, thereby effectively saving storage space while ensuring the accuracy and consistency of the new region division result.

[0067] Furthermore, the specific steps of S5 are as follows:

[0068] S5.1. From all preprocessed coordinate data, filter out the coordinate data that exists in each subregion (the subregion updated in S4) and draw a heat map for each subregion. The color distribution in the heat map reflects the density distribution and its changes in the target object. The heat map will serve as input data for the deep learning model.

[0069] S5.2. Use a deep learning model to perform multi-scale analysis on the heat map to capture local and global features. Multi-scale analysis can effectively address uneven density distribution of coordinate data and accurately identify high- and low-density areas.

[0070] S5.3. Introducing an attention mechanism into the deep learning model enables it to automatically focus on key areas in the heat map, improving the accuracy of density distribution judgment. The attention mechanism optimizes the feature extraction process by assigning weights to highlight important areas and suppress unimportant areas.

[0071] S5.4. Optimize the boundaries of sub-regions based on the processing results of the deep learning model: If there are two or more high-density regions in a sub-region, retain the original region boundaries and make comprehensive adjustments based on the edges of multiple high-density regions; if there is only one high-density region in a sub-region, redivide the region boundaries based on the edges of the high-density region.

[0072] Beneficial Effects: Compared to existing technologies, this invention not only significantly improves computational efficiency but also adapts to changes in activity patterns in real time, avoiding the subjectivity and inaccuracy of manual labeling. For example, in kindergartens, this invention can dynamically adjust the division of play areas based on children's activity hotspots and changing interests, ensuring efficient use of spatial resources while enhancing children's activity experience. Furthermore, this technology can also be widely applied to indoor scenarios such as shopping malls and warehouses, providing strong support for space planning, crowd management, and resource allocation. BRIEF DESCRIPTION OF THE DRAWINGS

[0073] Figure 1 is a flow chart of the method of the present invention;

[0074] Figure 2 A comparison diagram of the preliminary region division results and the manual labeling results of an embodiment of the present invention;

[0075] Figure 3 A density distribution heat map of the preliminary area division result according to an embodiment of the present invention;

[0076] Figure 4 This is a diagram showing the final region division result according to an embodiment of the present invention;

[0077] Figure 5 A comparison diagram of the final region division result and the manual labeling result according to an embodiment of the present invention; DETAILED DESCRIPTION

[0078] The technical solutions of the present invention are described in detail below through examples, but the protection scope of the present invention is not limited to the examples.

[0079] Example 1

[0080] Figure 1 This is a flow chart of a dynamic area division method based on activity coordinate data of the present invention. This embodiment is a case of using this method for dynamic division of game areas in children's role-playing games.

[0081] The specific steps are as follows:

[0082] Step 1: Get the real-time coordinate data of all children during the game time period:

[0083] Multiple UWB base stations were deployed above the classroom, with the front door as the coordinate axis' origin. Each child wore a fully charged UWB tag while playing. Trilateration was used to collect coordinate data of each child's movements at regular intervals (every one second) throughout the play period, ensuring data continuity and accuracy. The data collected included time, child ID, and its corresponding three-dimensional coordinates (x, y, z).

[0084] Step 2: Preprocess all the real-time coordinate data of children, including removing duplicate values, missing values, and abnormal values in the coordinate data, and screening the target object coordinate data that meets the conditions. The specific steps are as follows:

[0085] (1) Ensure that the data format is uniform. All data columns should be named time, id, x, y, and z. At the same time, use the deduplication function to deduplicate records with identical values for the time and id columns and coordinate data (x, y) in the original data. Use zero-filling to handle missing data and ensure that there is still enough data for dynamic area division after deduplication and filling.

[0086] (2) In the collected coordinate data, the coordinate data whose distance difference between adjacent time points exceeds the reasonable movement level of children (movement distance of more than 1.5 meters in 1 second) is defined as an outlier. An intelligent coordinate correction model based on LSTM-GAN is established. The generator uses a 3-layer LSTM network to learn the movement pattern of children, and the discriminator is a convolutional neural network. The dynamic anomaly threshold is defined: when the displacement speed between two adjacent points exceeds 1.5m / s (considering the maximum running speed of children), anomaly detection is triggered. The model generates a smooth trajectory that conforms to human kinematics by analyzing the trajectory context before and after 10 seconds, and verifies the rationality of the correction result through the residual network.

[0087] (3) Filter out the target object coordinate data that meets the conditions according to the target area boundary. Due to the limited classroom space, only the area within the classroom is divided into game areas. All the children's coordinate data are filtered by giving the x and y value ranges. Finally, all the children's coordinate data within the classroom are filtered out as the data used for subsequent area division.

[0088] Step 3: Divide the pre-processed coordinate data into multiple groups of fixed time periods according to the time series, and perform preliminary regional division on each group. The specific steps are as follows:

[0089] (1) The children spend 40 minutes in the game area. The pre-processed coordinate data are sorted by time, and the initial window size is set to 4 minutes and the initial step size is set to 2 minutes. A dynamic adjustment algorithm based on a sliding window is used to adaptively adjust the size and step size of the time window.

[0090] During periods when children are active or data changes dramatically, the window size is automatically adjusted to 3 minutes and the step length is adjusted to 1 minute; during periods when children are active or data changes slowly, the window size is automatically increased to 5 minutes and the step length is adjusted to 2 minutes.

[0091] (2) Customize the neighborhood radius (eps) and minimum number of points (min_samples) of the DBSCAN clustering algorithm. In this embodiment, the neighborhood radius selected for the division of the children's play area is 0.6, and the minimum number of points is 180.

[0092] (3) For the coordinate data in each dynamic time window, the DBSCAN clustering algorithm is applied to perform preliminary regional division, merge clusters with overlap > 65%, and remove small clusters with less than 50 points. The preliminary dynamic sub-region results corresponding to each time period are generated, and the coordinates of each divided sub-region are saved as a JSON file.

[0093] Step 4: Based on the coordinate data in each set of JSON files saved from the initial regional division, perform regional fusion and dynamically update the sub-region boundaries. The specific steps are as follows:

[0094] (1) Parse the JSON file that stores the preliminary area division results, extract the coordinates of each group of divided sub-areas, group the coordinate data by time point, and generate a dictionary indexed by time point.

[0095] (2) A high kernel density estimation with a bandwidth of 0.45 m (based on the average shoulder width of young children and the safety margin) is used for each sub-region to divide the data points in the sub-region into high-density areas and low-density areas.

[0096] If a sub-region in the preliminary regional division results of a certain group overlaps with a sub-region in the results of another group, calculate the kernel density estimate of the overlapping area;

[0097] If the overlapping area is greater than or equal to the dynamic fusion threshold (dynamically adjusted according to the kernel density estimate in the overlapping area, the threshold is 30% for high-density areas (≥1.2 people / m2), 35% for medium-density areas (0.8-1.2 people / m2), and 40% for low-density areas) and the overlap is continuous for ≥3 groups of windows (about 5-7 minutes), the regional fusion operation is performed; otherwise, the sub-region is directly retained and saved as an independent new sub-region in the new regional division result.

[0098] For the sub-regions that meet the fusion conditions, the adaptive density region fusion clustering algorithm is applied to cluster them.

[0099] Based on the kernel density estimate, the algorithm parameters ϵ and MinPts are dynamically adjusted. Key parameters are first dynamically calculated based on the kernel density distribution of the overlapping area: the neighborhood radius ϵ is set to 0.5-0.7 meters (base value 0.6 meters, dynamically scaled by ±20% based on the density range), and the minimum number of points MinPts is set to 150-200 (based on a weighted calculation of the area and average density). Core points with a density above 60% of the average are identified, and the area is then expanded if the density gradient is less than 20%. Finally, a rectangular boundary is generated and saved.

[0100] (4) For the merged sub-regions, dynamically adjust their boundaries according to the kernel density estimation value, and save the newly generated sub-regions to the new region division results. Finally, save the fused region division results as a new JSON file. Each fused region coordinate contains a unique region number, the fused four corner coordinates (x n 、y n 、x m 、y m ).

[0101] S5. Analyze the density distribution of target objects in each sub-region of the region division result after region fusion processing, and use deep learning to optimize the region division. The specific steps are as follows:

[0102] (1) Extract the coordinates of the four corners of each sub-region divided in the JSON file (x n 、y n 、x m 、y m ), define the corresponding area coordinate range and name.

[0103] (2) Use the pre-processed coordinate data of all children to draw a hexagonal heat map, where the color scale is set to 100 and the hexagon size is set to 30.

[0104] (3) A dual-channel deep network is constructed. Channel 1 (local features) is a three-layer convolutional network with a convolution kernel size of 3×3 to capture microscopic distributions; channel 2 (global features) is a dilated convolutional network with a dilation rate of 5 to perceive macroscopic patterns; the feature fusion layer realizes feature interaction through a cross-attention mechanism, and the weights of local features are dynamically adjusted by global features.

[0105] (4) The drawn heat map is used as the input data of the deep learning model, and the heat map is analyzed at multiple scales to capture the local and global features in the heat map and accurately identify high-density and low-density areas.

[0106] (5) Based on the processing results of the deep learning model, the boundaries of the sub-regions are optimized: if there are two or more high-density regions in a sub-region, the original region boundaries are retained and comprehensively adjusted according to the edges of multiple high-density regions; if there is only one high-density region in a sub-region, the region boundaries are re-divided according to the edges of the high-density region.

[0107] Figure 4 This is a diagram showing the final region division result according to an embodiment of the present invention; Figure 5 This is a comparison chart of the final region division result and the manual labeling result according to an embodiment of the present invention. Figure 4 and Figure 5 The solid line frame represents the sub-area after the operation of Example S5, and the dotted line frame represents the activity area divided manually based on the boundaries of the kindergarten classroom facilities such as tables, chairs, cabinets, etc. Figure 5 The comparison in shows that the area division result after the operation of Example S5 is similar to the result of manual division, which proves the effectiveness of using this method to divide the area. At the same time, this method takes into account human activity and is more accurate in dividing the activity area than manual marking according to the placement of facilities.

[0108] Through the above steps, the present invention can realize dynamic area division of children's role-playing games and provide technical support to optimize children's behavior analysis and game experience.

[0109] Figure 2 This is a comparison chart of the preliminary area division results and the manually labeled results of an embodiment of the present invention; the solid-line box represents the sub-area after the operation in embodiment S4. The dotted-line box represents the activity area manually divided based on the boundaries of the kindergarten classroom facilities such as desks, chairs, and cabinets. Figure 3 This is a density distribution heat map of the preliminary area division result of an embodiment of the present invention; the solid line box represents the sub-area after the operation of embodiment S4. Figure 2This shows that the regional fusion effect of S4 is already relatively ideal. However, considering that after S4 performs the regional fusion operation, the distribution of edge scattered points (data point density distribution problem) may cause the divided space of each sub-region to be quite different from the actual activity area, so S5 is further performed to divide the region more accurately. Figure 3 It is a heat map showing the execution of step S5 (process display).

[0110] The method first collects coordinate data for multiple active individuals within a target area over a specific time period. Data preprocessing removes outliers and noise to ensure data accuracy. Next, the activity area is initially divided into multiple sub-areas based on the coordinate distribution characteristics. This sub-area fusion is then performed, incorporating temporal and spatial density factors. Sub-area boundaries are dynamically updated to accommodate the changing behavior patterns of the active individuals over time. This method enables precise, real-time, and flexible area division based on actual activity patterns, effectively improving space utilization efficiency. It can be widely applied to numerous fields, including optimizing indoor space layout, providing strong support for the rational allocation of resources.

[0111] As described above, although the present invention has been shown and described with reference to specific preferred embodiments, it should not be construed as limiting the present invention itself. Various changes may be made to the form and details without departing from the spirit and scope of the present invention.

Claims

1. A dynamic area division method based on active coordinate data, characterized in that: The steps include: S1. Acquire continuous coordinate data of multiple target objects within a certain geographical area and within a data collection period within a given time interval, thereby obtaining real-time coordinate data of the target objects at a certain moment; S2. Preprocessing the acquired real-time coordinate data of the target object; S3, dividing the pre-processed coordinate data into dynamic time windows, and preliminarily dividing the geographical area where the coordinate data corresponding to each dynamic time window is located to obtain multiple sub-areas; The steps for S3 are as follows: S3.

1. Adopt a dynamic adjustment algorithm based on a sliding window to adaptively adjust the size and step size of the time window. The specific steps of S3.1 are as follows: S3.1.

1. Dynamically adjust the time window size based on the dynamic characteristics of the coordinate data. The window size adjustment formula is as follows: （1） in, w is the size of the current sliding time window, w min and w max are the minimum and maximum values of the window size, data_variance ( D [ i : i + w ]) refers to the variance of the data within the window, which indicates the volatility of the data within the window. D [i:i+ w ] is the window from the i Data points to i + w A sequence of data points, θ high is the data high fluctuation threshold, θ low is the low fluctuation threshold of data; S3.1.

2. Dynamically adjust the step size of the time window based on the continuity and change trend of the data. The step size adjustment formula is as follows: （2） in, s is the step size of the current sliding time window, s min and s max are the minimum and maximum values of the step size, data_rate ( D [ i : i + w ]) refers to the rate of change of data within the window, indicating the speed of change of data within the window. D [ i : i + w ] is the window from the i Data points to i + w A sequence of data points, θ rate is the threshold of data change rate; S3.

2. Perform preliminary geographic division of the coordinate data within each dynamic time window, generate a set of sub-regions corresponding to each time period, and store the preliminary regional division results; S4. Based on the stored preliminary region division results and sub-region overlap, perform a region fusion operation to dynamically update the sub-region boundaries.

2. The dynamic area division method according to claim 1, characterized in that: In S1, the real-time coordinate data includes the ID, three-dimensional coordinates and time information corresponding to the three-dimensional coordinate position of the target object.

3. The dynamic area division method according to claim 1, characterized in that: In S2, the data preprocessing method includes removing duplicate values, missing values, and abnormal values in the coordinate data, and screening out the real-time coordinate data of the target object that meets the conditions; The qualified real-time coordinate data of the target object refers to data in which the coordinate data of the target object always appears within the scope of the target geographical area during a complete data collection cycle. The target geographical area refers to the entire area that needs to be divided into regions.

4. The dynamic area division method according to claim 1, characterized in that: In S2, an intelligent coordinate correction module is introduced to optimize the coordinate data after deduplication for missing values and outliers. The specific method is as follows: An intelligent coordinate correction model based on a generative adversarial network is constructed; the intelligent coordinate correction model automatically scans coordinate data, identifies abnormal coordinates that do not conform to the behavioral characteristics of the target object, and automatically generates more reasonable coordinate values for the identified abnormal coordinates to replace them.

5. The dynamic area division method according to claim 1, characterized in that: The specific operation steps of S4 are as follows: S4.

1. Perform kernel density estimation on the data points in each sub-region. The kernel density estimation formula is: （3） in: is a data point x The kernel density estimate of K is the kernel function used to measure the data points x A data point in the neighborhood x i For data points x The contribution weight of h is the bandwidth, which is used to control the smoothness of the kernel function. n is the total number of data points in the current sub-region, that is, the number of samples involved in density calculation, d is the data dimension; S4.

2. Divide the data points in the sub-region into high-density regions and low-density regions based on the kernel density estimate; if the kernel density estimate is higher than the preset density threshold, θ high The sub-region is a high-density area; the kernel density estimate is lower than the preset density threshold θ low The sub-region of is a low-density area; S4.

3. Arrange the stored preliminary region division results in chronological order, and compare the coordinates of all subregions in the preliminary region division results of two adjacent groups. If a subregion in one group's preliminary region division results overlaps with a subregion in the other group's results, calculate the kernel density estimate of the overlapping area. If the overlapping area is greater than or equal to the dynamic fusion threshold, perform the region fusion operation. S4.

4. For the merged sub-regions, dynamically adjust their boundaries based on the kernel density estimate, and save the newly generated sub-regions to the new region division result; S4.

5. After completing the regional fusion operation on two adjacent groups of preliminary regional division results, further perform regional fusion operation on all sub-regions in the new regional division result.

6. The dynamic area division method according to claim 5, characterized in that: In S4.3, if a sub-region in the preliminary regional division results of one group overlaps with a sub-region in the results of another group, the specific steps to be performed are: S4.3.

1. Calculate the overlapping area R of two sub-areas overlap : （4） in, R overlap is the overlapping area of the two sub-areas, (x1, y1, x2, y2) are the boundary coordinates of the first sub-area, (x1, y1) are the lower left corner coordinates, (x2, y2) are the upper right corner coordinates, (x1′, y1′, x2′, y2′) are the boundary coordinates of the second sub-area, (x1′, y1′) are the lower left corner coordinates, (x2′, y2′) are the upper right corner coordinates; S4.3.

2. Calculate the area A of the overlapping region overlap : （5） in, A overlap is the area of the overlapping region of the two subregions, x left overlap is the left boundary of the overlapping area, y bottom overlap is the left boundary of the overlapping area, x right overlap is the right boundary of the overlapping area, y topoverlap is the upper boundary of the overlapping area; S4.3.

3. Calculate the kernel density estimate of the overlapping area and dynamically adjust the fusion threshold θ based on the kernel density estimate of the overlapping area. fuse : （6） in, θ fuse is the dynamic fusion threshold, θ high and θ low are the preset high density threshold and low density threshold, respectively, used to classify density levels. represents the kernel density estimate within the overlapping region; S4.3.4, if the overlapping area A overlap Greater than or equal to the dynamic fusion threshold θ fuse , then perform the regional fusion operation; otherwise, retain the sub-region directly and save it as an independent new sub-region to the new regional division result; the steps for performing the regional fusion operation are: (1) For sub-regions that meet the fusion conditions, the adaptive density region fusion clustering algorithm is applied for clustering; according to the kernel density estimation value, the parameters ϵ and MinPts of the adaptive density region fusion clustering are dynamically adjusted: （7） in, is the neighborhood radius, which is used to define the neighborhood range of the data point. and are the neighborhood radius parameters for high-density and low-density areas, respectively. MinPts is the minimum number of neighborhood points required for a core point. large and MinPts small are the minimum number of points in high-density and low-density areas, respectively. is the kernel density estimate in the overlapping region, θ high is the preset high-density threshold; (2) For data points in the overlapping area, determine whether they are core points based on their density value and the number of points in the neighborhood; if a point is a core point, start from the core point and expand the clustering through the density reachability relationship; (3) For the subclusters formed by adaptive density region fusion clustering, determine whether they meet the fusion conditions; if they meet the fusion conditions, merge these subclusters into a new subregion; if they do not meet the fusion conditions, retain these subclusters as independent subregions.

7. The dynamic area division method according to claim 5, characterized in that: The method for dynamically adjusting sub-region boundaries in S4.4 is: The four corner coordinates (x1, y1, x2, y2) and (x 1’ 、y 1’ 、x 2’ 、y 2’ ), where (x1, y1) and (x2, y2) are the lower left corner and upper right corner coordinates of the first sub-region respectively, (x 1’ ,y 1’ ) and (x 2’ ,y 2’ ) are the lower left corner coordinates and upper right corner coordinates of the second sub-region, respectively, extract the minimum and maximum values in the x and y directions, and generate the new sub-region boundary coordinates (x min 、y min 、x max 、y max ): x min =min(x1,x 1’ ), y min =min(y1,y 1’ ); x max =max(x2,x 2’ ), y max =max(y2,y 2’ )。 8. The dynamic area division method according to claim 1, characterized in that: It also includes S5, which uses deep learning to further optimize the region division based on the density distribution of target objects in each sub-region after the region fusion processing.

Images