A spatiotemporal semantic constraint trajectory data compression method

Abstract

The application discloses a spatiotemporal semantic constraint trajectory data compression method, which comprises the following steps: S1, setting a time-synchronized Euclidean distance SED threshold and a trajectory point semantic similarity threshold; S2, acquiring a trajectory T={P1, P2, L, P n} and an approximate trajectory simtr={P1, P n}; S3, compressing the trajectory based on a trajectory compression criterion; and S4, outputting a compressed trajectory. The application solves the technical problem that a trajectory compression result lacks interpretability in a conventional trajectory data compression method, and can be used for spatiotemporal trajectory data semantic representation and deep data mining, such as semantic information mining in human activity trajectory big data, hurricane activity feature mining, and animal migration rule and habit analysis.

Description

Technical Field

[0001] The present invention belongs to the technical field of data management, and in particular relates to a trajectory data compression method with spatiotemporal semantic constraints. Background Art

[0002] With the rapid development and popularization of global positioning systems and mobile terminal devices, massive amounts of spatiotemporal trajectory data have been generated. How to effectively manage, analyze and mine massive amounts of spatiotemporal trajectory data has become a hot research area. Chinese invention patent CN112035873A discloses a method for desensitizing spatiotemporal trajectory data; CN111190891A discloses a method for segmented storage of multi-semantic trajectory data. Spatiotemporal trajectory data, such as timestamps, device numbers, base station numbers, and time types, will generate massive amounts of data over time, posing a huge challenge to storage. Therefore, spatiotemporal trajectory data compression has become an urgent problem to be solved. The compression of spatiotemporal trajectory data belongs to the field of data management. Spatiotemporal trajectory data compression technology can reduce the data size while maintaining the similarity between the original trajectory and the compressed trajectory as much as possible, achieving efficient storage and representation of spatiotemporal trajectory data, and providing strong support for spatiotemporal trajectory data query, analysis, and mining. Chinese invention patent CN112988759A provides a processing method for spatiotemporal trajectory data compression, solving the problem of low data compression efficiency caused by the inability to accurately extract features after compression. The method includes: splitting the access data and inferring a complete base station information table; grouping the data by device number and sorting the multiple data copies within each group; generating a first trajectory compressed data block and establishing parameter information; marking the latitude and longitude of the trajectory compressed data block; iteratively processing each data copy within each group; creating multiple trajectory compressed data blocks and determining the type of each trajectory compressed data block in turn; calculating the time difference between two adjacent static data blocks to assign a value to them; and repeating the steps to complete the compression of each group of trajectory compressed data blocks in turn.

[0003] Traditional spatiotemporal trajectory data compression techniques primarily focus on the positional features of trajectory points during the compression process, aiming to ensure geometric similarity between the compressed and original trajectories. However, they ignore the crucial semantic information of spatiotemporal trajectory data, making the compressed trajectories lacking interpretability and difficult to understand and apply. From a data mining perspective, the semantic information hidden in spatiotemporal trajectory data is of great value, reflecting the essential properties and behavioral patterns of moving objects. Therefore, researching trajectory compression methods with spatiotemporal semantic constraints and considering the semantic features of sampling points during the compression process is crucial for enhancing the interpretability and utility of compressed trajectories. Summary of the Invention

[0004] This invention aims to address the problem that traditional trajectory data compression methods focus solely on geometric features while ignoring the semantic characteristics of spatiotemporal trajectories, resulting in a lack of interpretability in the resulting trajectory compression. This invention enables spatiotemporal trajectory compression that incorporates semantic information, enabling applications in semantic representation of spatiotemporal trajectory data and deep data mining.

[0005] To solve the above technical problems, the specific technical solution of the trajectory data compression method with spatiotemporal semantic constraints of the present invention is as follows:

[0006] S1: Set the time-synchronized Euclidean distance SED threshold and the trajectory point semantic similarity threshold;

[0007] S2: Get trajectory T = {P1, P2, L, P n} and its approximate trajectory simtr={P1,P n};

[0008] S3: compressing the trajectory based on a trajectory compression criterion;

[0009] S4: Output compression trajectory.

[0010] Furthermore, the trajectory compression criterion method is as follows:

[0011] S31: Calculate the trajectory point P in the current trajectory i Normalized spatiotemporal semantic distance of ;

[0012] S32: Find the point corresponding to the maximum normalized spatiotemporal semantic distance in the current trajectory, and determine whether the SED value or semantic similarity of the point is greater than a threshold. If so, go to step S33; if not, go to step S35;

[0013] S33: The trajectory point P i Mark it as a feature point and divide the current trajectory into two sub-trajectories from this point;

[0014] S34: Repeat step S3 for each sub-track in S33;

[0015] S35: Replace the current trajectory with an approximate trajectory of the current trajectory.

[0016] Furthermore, the method for calculating the normalized spatiotemporal semantic distance of the trajectory points in the current trajectory in step S31 is as follows:

[0017] S311: Calculate trajectory point P i The position point P obtained by mapping it in time and space proportion on the approximate trajectory simtr i SED between ′;

[0018] S312: Calculate the trajectory point P iSemantic similarity of trajectory points;

[0019] S313: Calculate the trajectory point P i Normalized spatiotemporal semantic distance.

[0020] Furthermore, the calculation method of step S311 is as follows:

[0021]

[0022]

[0023]

[0024] The method for calculating the semantic similarity of trajectory points in step S312 is as follows:

[0025]

[0026] Among them, w k represents the classification attribute weight, M c Indicates the number of classification attributes, H(A k ) represents the original approximate trajectory attribute A k The information entropy of H(A k ′) represents the original approximate trajectory adding the trajectory point P i Then the information entropy of the attribute;

[0027] The normalized spatiotemporal semantic distance calculation method in step S313 is as follows:

[0028]

[0029] Among them, w1 and w2 correspond to the spatiotemporal weight and semantic weight respectively, SED_max is the maximum SED value on the sub-trajectory segment, and Semantic_Sim_max is the maximum semantic similarity of the trajectory points on the sub-trajectory segment.

[0030] The present invention also provides an application of the above method, wherein the trajectory data compression method with spatiotemporal semantic constraints is applied to semantic information mining in human activity trajectory big data, hurricane activity feature mining, and animal migration pattern and habit analysis.

[0031] The present invention's spatiotemporal semantically constrained trajectory data compression method offers the following advantages: It calculates the trajectory point corresponding to the maximum normalized spatiotemporal semantic distance in the trajectory and compresses the trajectory using a trajectory compression criterion. This method addresses the technical issue of traditional spatiotemporal trajectory data compression techniques, which ignore the important semantic information in spatiotemporal trajectory data. This method makes the compressed trajectory highly interpretable, making it easier for people to understand and apply. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] Figure 1 Schematic diagram of the method of the present invention;

[0033] Figure 2 Schematic diagram for defining the original trajectory, sub-trajectory, spatiotemporal projection trajectory points and approximate trajectory simtr of the present invention;

[0034] Figure 3 is the trajectory point P of the present invention i The position point P obtained by mapping the approximate trajectory simtr according to the time and space ratio i Schematic diagram of SED between ′;

[0035] Figure 4 Schematic diagram of a detailed implementation method of the present invention;

[0036] Figure 5 This is a graph showing the experimental results of the single-dimensional semantic trajectory compression of the present invention;

[0037] Figure 6 This is a graph showing the experimental results of the two-dimensional semantic trajectory compression of the present invention;

[0038] Figure 7 It is the two-dimensional semantic joint semantic retention result graph of the present invention;

[0039] Figure 8 This is a graph showing the experimental results of the three-dimensional semantic trajectory compression of the present invention;

[0040] Figure 9 It is the three-dimensional semantic joint semantic retention result graph of the present invention;

[0041] Figure 10 The following are trajectory data compression results using the method of the present invention and the trajectory data compression results using the TD-TR algorithm. DETAILED DESCRIPTION

[0042] In order to better understand the purpose, structure and function of the present invention, the following is a further detailed description of a trajectory data compression method with spatiotemporal semantic constraints of the present invention with reference to the accompanying drawings.

[0043] The technical solution of the present invention is described in detail below with reference to the accompanying drawings:

[0044] The overall process of the present invention is as follows Figure 1 As shown, the concepts are Figure 2As shown in the figure, we first set the time synchronous Euclidean distance (SED) threshold α and the trajectory point semantic similarity threshold β; read the trajectory data to obtain the trajectory T and its approximate trajectory simtr; find the trajectory point corresponding to the maximum normalized spatiotemporal semantic distance in the trajectory T by calculation, and compress the trajectory based on the trajectory compression criterion to achieve the purpose of trajectory compression.

[0045] This paper constructs a spatiotemporal semantic similarity measurement model based on SED and information entropy. Based on this model, it achieves the goal of spatiotemporal trajectory data compression by marking and retaining the spatiotemporal semantic feature points of the original spatiotemporal trajectory and discarding trajectory points with unclear spatiotemporal semantic information. The innovation of this paper lies in comprehensively considering the spatiotemporal and semantic characteristics of spatiotemporal trajectories, providing a spatiotemporal semantically constrained trajectory compression method. This method organically combines the spatiotemporal and semantic information of spatiotemporal trajectory data and proposes a normalized spatiotemporal semantic distance. This method takes into account the spatiotemporal and semantic similarity between the compressed trajectory and the original trajectory, removes redundant data, and achieves spatiotemporal trajectory data compression.

[0046] The specific steps are as follows:

[0047] S1: Set the SED threshold α and the trajectory point semantic similarity threshold β;

[0048] S2: Get trajectory T = {P1, P2, L, P n} and its approximate trajectory simtr={P1,P n};

[0049] S3: Compress the trajectory according to the trajectory compression criterion. The specific steps are as follows:

[0050] S31: Calculate the normalized spatiotemporal semantic distance of all trajectory points P in the current trajectory. The specific steps are as follows:

[0051] S311: If Figure 3 As shown, calculate the trajectory point P i The position point P obtained by mapping it in time and space proportion on the approximate trajectory simtr i The SED between ′ is calculated as follows:

[0052]

[0053]

[0054]

[0055] S312: Calculate the trajectory point P i Semantic similarity of trajectory points:

[0056] The trajectory point P i The semantic similarity of trajectory points is defined as P i The change in the attribute entropy of the approximate trajectory simtr caused by adding the approximate trajectory simtr is calculated as follows:

[0057]

[0058] Among them, w k represents the classification attribute weight, M c Indicates the number of classification attributes, H(A k ) represents the original approximate trajectory attribute A k The information entropy of H(A k ′) represents the original approximate trajectory adding the trajectory point P i Then the information entropy of the attribute;

[0059] S313: Calculate the trajectory point P i Normalized spatiotemporal semantic distance of:

[0060] In order to unify the spatiotemporal and semantic data, the present invention defines the trajectory point P i The normalized spatiotemporal semantic distance of is as follows:

[0061]

[0062] Among them, w1 and w2 correspond to the spatiotemporal weight and semantic weight respectively, SED_max is the maximum SED value on the sub-trajectory segment, and Semantic_Sim_max is the maximum semantic similarity of the trajectory points on the sub-trajectory segment;

[0063] S32: Find the point P with the maximum normalized spatiotemporal semantic distance on the current trajectory k , and judge P k Corresponding SED Pk Is it greater than the threshold α or Semantic_Sim(P i ) is greater than the threshold β, such as Figure 4 If yes, go to step S33; if no, go to step S35;

[0064] S33: The P k Mark it as a feature point and move the current sub-trajectory T from the P k The treatment is divided into sub-trajectory tr1 = {P1, P2, L, P k} and sub-trajectory tr2 = {P k ,P k+1 ,L,P n};

[0065] S34: Repeat step S3 for sub-trajectory tr1 and sub-trajectory tr2 respectively;

[0066] S35: Replace the current trajectory with an approximate trajectory of the current trajectory.

[0067] This invention relates to a trajectory data compression method with spatiotemporal semantic constraints. By preserving the spatiotemporal semantic features of the spatiotemporal trajectory, the compressed spatiotemporal trajectory becomes interpretable, thus facilitating the analysis and mining of large-scale trajectory data. This method is primarily applicable to analyzing and uncovering the motion states and activity patterns of individual mobile objects in spatiotemporal environments, for example, in semantic information mining from human activity trajectory data, hurricane activity feature mining, and animal migration patterns and behavior analysis.

[0068] The present invention is based on three different sets of spatiotemporal trajectory data (Trajectory 1, Trajectory 2, and Trajectory 3), and assigns single-dimensional, two-dimensional, and three-dimensional semantics to each set of spatiotemporal trajectory data, so as to observe the ability of the spatiotemporal trajectory data compressed by the present invention to retain the semantic information of the original spatiotemporal trajectory data. The experimental results are compared with the traditional top-down compression algorithm based on time ratio (Top-Down Time-Ratio, TD-TR) under the same compression rate. The results are as follows: Figures 5 to 10 As shown in the figure, " / " represents the original trajectory, "★" represents the trajectory compressed by the present invention, and "." represents the trajectory compressed by the traditional compression method TD-TR. It is also worth noting that semantic value 1 represents an ordinary semantic point, while semantic values ​​2 to 10 represent important semantic points. Undoubtedly, if the compressed trajectory retains more important semantic points from the original trajectory, more ordinary semantic points will be discarded.

[0069] Figure 5 The figure shows the compression of spatiotemporal trajectory data with only single-dimensional semantics. The number of spatiotemporal trajectory points after compression by the present invention is greater at semantic values ​​2 to 10 than that after compression by the traditional TD-TR algorithm, indicating that the present invention is more capable of retaining important semantic points of the original trajectory.

[0070] Figure 6 The figure shows the compression of spatiotemporal trajectory data with two-dimensional semantics. Each set of trajectories corresponds to two semantic-preserving graphs. It can be seen that the proposed method still has a good ability to preserve the two-dimensional semantics of the original trajectory, and the retention of important semantics in each dimension is greater than that of the traditional TD-TR algorithm.

[0071] Figure 7 The figure shows the result of two-dimensional semantic joint semantic preservation. This figure is based on the fact that if a trajectory point contains two important semantic attributes in two dimensions at the same time, it means that the point has more important semantic features than the spatiotemporal trajectory point with only single-dimensional semantics, and thus should be preserved. Figure 7It can be seen that the present invention greatly retains the important semantic joint points of the original spatiotemporal trajectory, further indicating that the important spatiotemporal trajectory semantic information retained after compression by the present invention is more similar to the original trajectory semantic information.

[0072] Figure 8 The figure shows the compression of spatiotemporal trajectory data with three-dimensional semantics, where each group of trajectories corresponds to three semantic preservation graphs. From the overall perspective, it can be seen that the present invention can still well preserve the important semantic information of each dimensional semantic feature of the spatiotemporal trajectory data.

[0073] Figure 9 The figure shows the result of 3D semantic joint semantic preservation. Since the spatiotemporal trajectory data in this group of experiments has 3D semantics, the joint semantic points can be divided into 2D joint semantic points and 3D joint semantic points, which correspond to 2D important semantic trajectory points and 3D important semantic trajectory points respectively. Obviously, 3D joint semantic points are more important than 2D joint semantic points in terms of semantics. Figure 9 It can be seen that the compressed spatiotemporal trajectory of the present invention greatly retains the two-dimensional joint semantic points of the original trajectory and recognizes almost all the three-dimensional joint semantic points of the original trajectory. Under the same compression rate, the ability to retain important semantic features of the original trajectory is far greater than that of the traditional TD-TR algorithm, verifying the effectiveness of the present invention.

[0074] In addition, in order to intuitively see the difference between the present invention and the traditional algorithm in the selection of spatiotemporal trajectory points, the present invention takes a partial trajectory compression result graph for display. Figure 10 The figure is a one-dimensional semantic compression result diagram, in which the dots represent the original trajectory points, the rectangles represent the compressed trajectory points, and the triangles represent the important semantic points retained after compression. Since the one-dimensional semantics contains 9 different important semantic information, it is difficult to classify and display them. Therefore, the present invention uses triangles to represent them uniformly. However, it should be known that the important semantic information contained in the same triangle may not be the same. Figure 10 It can be seen that under the same compression rate, the present invention ( Figure 10 (a)) and the traditional TD-TR algorithm ( Figure 10 (b)) can better preserve the basic shape features of the original trajectory. From the circled details, it can be seen that due to the use of normalized spatiotemporal semantic distance as a metric, the present invention is better able to identify important semantic feature points of the original trajectory and is more inclined to retain important semantic feature points in places where trajectory points are similar. Therefore, compared with the traditional TD-TR algorithm, the trajectory compressed by the present invention has a higher semantic similarity with the original trajectory.

[0075] In summary, under the premise of the same compression rate, compared with the traditional TD-TR algorithm, the compressed trajectory of the present invention retains the spatiotemporal characteristics of the original trajectory while having better semantic feature similarity, which is conducive to analysis and mining based on trajectory data.

[0076] It will be understood that the present invention is described by way of some embodiments, and it will be appreciated by those skilled in the art that various changes or equivalent substitutions may be made to these features and embodiments without departing from the spirit and scope of the present invention. In addition, under the teachings of the present invention, these features and embodiments may be modified to adapt to specific circumstances and materials without departing from the spirit and scope of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed herein, and all embodiments falling within the scope of the claims of this application are intended to be protected by the present invention.

Claims

1. A trajectory data compression method with spatiotemporal semantic constraints, characterized in that: The following steps are involved: S1: Set the Euclidean distance for time synchronization threshold and trajectory point semantic similarity threshold; S2: Get trajectory and its approximate trajectory ; S3: compressing the trajectory based on a trajectory compression criterion; S4: output compression trajectory; The trajectory compression criterion method described in S3 comprises the following steps: S31: Calculate the trajectory point P in the current trajectory i Normalized spatiotemporal semantic distance of ; i=1, 2, ...n: S311: Calculate trajectory points Rather than being on an approximate trajectory The position points obtained by mapping the time and space ratio between ; S312: Calculate the trajectory points Semantic similarity of trajectory points; S313: Calculate the trajectory points Normalized spatiotemporal semantic distance of ; The calculation method of step S311 is as follows: （1）； （2）； （3）； The method for calculating the semantic similarity of trajectory points in step S312 is as follows: （4）； in, represents the classification attribute weight, Indicates the number of classification attributes, Represents the original approximate trajectory attributes The information entropy of Indicates that the original approximate trajectory is added to the trajectory point Then the information entropy of the attribute; The normalized spatiotemporal semantic distance calculation method in step S313 is as follows: （5）； in, 、 Corresponding to spatiotemporal weight and semantic weight respectively, For sub-trajectory segments Maximum value, is the maximum semantic similarity of the trajectory points on the sub-trajectory segment; S32: Find the point P corresponding to the maximum normalized spatiotemporal semantic distance in the current trajectory k , judge the point Is the value or semantic similarity greater than the threshold? If so, go to step S33; if not, go to step S35; S33: The trajectory point P k Mark it as a feature point and divide the current trajectory into two sub-trajectories from this point; S34: Repeat step S3 for each sub-track in S33; S35: Replace the current trajectory with an approximate trajectory of the current trajectory.

2. The trajectory data compression method with spatiotemporal semantic constraints according to claim 1, characterized in that: The spatiotemporal semantically constrained trajectory data compression method is applied to semantic information mining in human activity trajectory big data, hurricane activity feature mining, and animal migration pattern and habit analysis.

Images