Spatial-temporal trajectory data character representation and feature extraction method

Abstract

The invention relates to a spatio-temporal trajectory data character representation and feature extraction method, and belongs to the technical field of trajectory data processing. Through the steps of preprocessing, standard vectorization, unit circle character processing and feature extraction, efficient characterization and depth feature mining of spatio-temporal trajectory data are realized. According to the method, data missing is effectively filled, data processing efficiency is improved, data interpretability is enhanced through character representation, meanwhile, the feature extraction process is optimized, calculation complexity is reduced, and data analysis precision is improved. The method is suitable for various spatio-temporal trajectory data, has wide applicability and high flexibility, is easy to implement, and provides a novel and effective technical means for the fields of trajectory data analysis, clustering, pattern recognition and the like.

Description

Technical Field

[0001] The invention belongs to the technical field of trajectory data processing and relates to a character representation and feature extraction method for spatiotemporal trajectory data. Background Art

[0002] With the popularity of the Internet of Things, smart phones and in-car navigation systems, spatiotemporal trajectory data, as a data type that records the movement path of objects, has become a valuable resource in many fields such as geographic information systems, intelligent transportation, urban planning, and environmental monitoring. In-depth analysis of these data can reveal valuable information such as travel patterns, causes of traffic congestion, and population mobility patterns.

[0003] Spatiotemporal trajectory data are often affected by various factors during the collection process, such as signal interference, equipment failure, etc., resulting in noise and missing values ​​in the data. Traditional data preprocessing techniques such as simple elimination and mean filling cannot effectively maintain the integrity and authenticity of the data, affecting the accuracy of subsequent analysis results.

[0004] Traditional spatiotemporal trajectory data representation methods, such as coordinate-based point sequence representation, cannot fully express the dynamic characteristics and complex structure of the data. In addition, these methods have limitations in data compression and feature extraction, which is not conducive to large-scale data processing and in-depth analysis.

[0005] Feature extraction is a key step in spatiotemporal trajectory data analysis, but existing feature extraction techniques often have high computational complexity and are difficult to achieve efficient processing on large-scale data sets. In addition, how to extract discriminative and robust features to support high-precision pattern recognition and predictive analysis is a problem that has not yet been fully solved.

[0006] Existing spatiotemporal trajectory data processing methods are often designed for specific types of data or application scenarios and lack versatility. In practical applications, spatiotemporal trajectory data from different sources and formats require different processing methods, which limits the scope of application of existing technologies.

[0007] As the amount of data continues to grow, higher requirements are placed on the processing speed and real-time performance of spatiotemporal trajectory data. Traditional data processing methods are difficult to meet the needs of real-time analysis and online decision support in terms of computational efficiency.

[0008] Spatiotemporal trajectory data often involve personal privacy. How to conduct effective data analysis while ensuring data security is another challenge facing current technology. Summary of the invention

[0009] In view of this, the purpose of the present invention is to provide a character representation and feature extraction method for spatiotemporal trajectory data. The purpose is to solve the problems of low efficiency, waste of storage space and high computational complexity in existing trajectory data storage and processing technologies. The present invention greatly reduces the amount of data storage by characterizing the offset of a trajectory point relative to the previous trajectory point, while retaining the integrity and analysis capabilities of the trajectory.

[0010] In order to achieve the above object, the present invention provides the following technical solutions:

[0011] A method for characterizing and extracting features of spatiotemporal trajectory data, comprising the following steps:

[0012] S1: Obtain the spatiotemporal trajectory dataset to be processed;

[0013] S2: Preprocessing the spatiotemporal trajectory dataset, including:

[0014] S21: sampling the spatiotemporal trajectory dataset at a preset time interval;

[0015] S22: interpolating the sampled spatiotemporal trajectory data to fill in the missing data;

[0016] S3: Perform standard vectorization processing on the preprocessed spatiotemporal trajectory data, including:

[0017] S31: normalizing the longitude and latitude information in the spatiotemporal trajectory data so that it falls within a preset range;

[0018] S32: Calculate the vector between adjacent trajectory points and express the vector in a two-dimensional coordinate form;

[0019] S33: Calculate the normalized radian between the vector and the reference axis;

[0020] S4: performing unit circle characterization on the normalized radians, including:

[0021] S41: Divide the unit circle into a preset number of layers, and divide each layer into a preset number of equal parts;

[0022] S42: determining the position of the vector on the unit circle according to the normalized radian, and expressing the position using character code;

[0023] S5: combining the character code with the length information of the vector to obtain a character representation of the vector;

[0024] S6: Extracting features from the character representation, including:

[0025] S61: Segment the character representation into sequences of preset length and step size;

[0026] S62: Using the sequence as feature information of the spatiotemporal trajectory data.

[0027] Furthermore, the preset time interval is T seconds.

[0028] Furthermore, the interpolation process adopts linear interpolation, Lagrange interpolation or spline interpolation.

[0029] Furthermore, the preset interval is (-1, 1).

[0030] Furthermore, the reference axis is any one of the x-axis, the y-axis and the z-axis.

[0031] Furthermore, the character code starts with the lowercase letter 'a' of the ASCII code.

[0032] Furthermore, in the feature extraction, the preset length is L and the preset step size is S.

[0033] A spatiotemporal trajectory data characterization and feature extraction system, comprising:

[0034] A data acquisition module, used to acquire the spatiotemporal trajectory data set to be processed;

[0035] The preprocessing module is used to preprocess the spatiotemporal trajectory dataset, including:

[0036] A sampling module, used to sample the spatiotemporal trajectory dataset at a preset time interval;

[0037] An interpolation module, used for performing interpolation processing on the sampled spatiotemporal trajectory data to fill in missing data;

[0038] A vectorization module is used to perform standard vectorization processing on the preprocessed spatiotemporal trajectory data, including:

[0039] A standardization module, used for standardizing the latitude and longitude information in the spatiotemporal trajectory data;

[0040] A vector calculation module, used for calculating the vectors between adjacent trajectory points and expressing the vectors in a two-dimensional coordinate form;

[0041] A radian calculation module, used for calculating the normalized radian between the vector and the reference axis;

[0042] A characterization module, used for performing unit circle characterization on the normalized radian, comprises:

[0043] A unit circle division module, used for dividing the unit circle into a preset number of layers, and dividing each layer into a preset number of equal parts;

[0044] A position determination module, used to determine the position of the vector on the unit circle according to the normalized radian;

[0045] An encoding module, used for representing the position using a character encoding;

[0046] A combining module, used for combining the character code with the length information of the vector to obtain a character representation of the vector;

[0047] A feature extraction module, used for extracting features from the character representation, comprises:

[0048] A segmentation module, used for segmenting the characterized representation into sequences of preset length and step size;

[0049] The sequence generation module is used to use the sequence as feature information of the spatiotemporal trajectory data.

[0050] Furthermore, the data acquisition module is connected to an external data source and is used to receive a spatiotemporal trajectory data set provided by the external data source.

[0051] Furthermore, the feature extraction module is connected to the subsequent analysis and processing module to transmit the feature information to the subsequent analysis and processing module.

[0052] The beneficial effects of the present invention are:

[0053] (1) Through sampling and interpolation processing at preset time intervals, the present invention can effectively fill in the missing values ​​in the spatiotemporal trajectory data, improve the quality and integrity of the data, and thus improve the efficiency of subsequent processing.

[0054] (2) By performing standard vectorization on the longitude and latitude information and unit circle characterization on the angle between the vector and the reference axis, the present invention achieves a compact and intuitive representation of the spatiotemporal trajectory data and enhances the interpretability and representation capabilities of the data.

[0055] (3) The present invention effectively extracts the deep features of the spatiotemporal trajectory data by dividing the character representation into sequences and using them as feature information, providing a rich feature basis for subsequent data analysis and application.

[0056] (4) Compared with the traditional spatiotemporal trajectory data processing method, the characterization method of the present invention simplifies the data representation, reduces the computational complexity, and improves the computational efficiency.

[0057] (5) Through the method of the present invention, the feature extraction of spatiotemporal trajectory data is more accurate, which helps to improve the accuracy and reliability of subsequent data analysis, such as applications in trajectory clustering, pattern recognition, etc.

[0058] (6) The method of the present invention is applicable to spatiotemporal trajectory data from different sources and formats, has wide applicability and flexibility, and can meet data processing requirements in different scenarios.

[0059] (7) The system structure of the present invention is clear, and the modular design makes the implementation process simple and convenient for deployment and maintenance in practical applications.

[0060] Other advantages, objectives and features of the present invention will be described in the following description to some extent, and to some extent, will be obvious to those skilled in the art based on the following examination and study, or can be taught from the practice of the present invention. The objectives and other advantages of the present invention can be realized and obtained through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0061] In order to make the purpose, technical solutions and advantages of the present invention more clear, the present invention will be described in detail below in conjunction with the accompanying drawings, wherein:

[0062] Figure 1 A schematic diagram of the standard vectorization between two points of the spatiotemporal trajectory data provided by the present invention;

[0063] Figure 2 A schematic diagram of the characterization of the unit circle provided by the present invention;

[0064] Figure 3 A schematic diagram of vector characterization provided by the present invention;

[0065] Figure 4 This is a schematic diagram of feature extraction provided by the present invention. DETAILED DESCRIPTION

[0066] The following describes the embodiments of the present invention by specific examples, and those skilled in the art can easily understand other advantages and effects of the present invention from the contents disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and the details in this specification can also be modified or changed in various ways based on different viewpoints and applications without departing from the spirit of the present invention. It should be noted that the illustrations provided in the following embodiments only illustrate the basic concept of the present invention in a schematic manner, and the following embodiments and features in the embodiments can be combined with each other without conflict.

[0067] Among them, the drawings are only used for illustrative explanations, and they only represent schematic diagrams rather than actual pictures, and should not be understood as limitations on the present invention. In order to better illustrate the embodiments of the present invention, some parts of the drawings may be omitted, enlarged or reduced, and do not represent the size of actual products. For those skilled in the art, it is understandable that some well-known structures and their descriptions in the drawings may be omitted.

[0068] The same or similar numbers in the drawings of the embodiments of the present invention correspond to the same or similar parts; in the description of the present invention, it should be understood that if the terms "upper", "lower", "left", "right", "front", "rear", etc. indicate the orientation or position relationship, they are based on the orientation or position relationship shown in the drawings, which is only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operate in a specific orientation. Therefore, the terms describing the position relationship in the drawings are only used for illustrative purposes and cannot be understood as limiting the present invention. For ordinary technicians in this field, the specific meanings of the above terms can be understood according to specific circumstances.

[0069] The technical solution adopted by the present invention is to perform character expression dimensionality reduction processing and feature extraction on the spatiotemporal trajectory data.

[0070] Assume that a characterized spatiotemporal trajectory dataset Tras = (Tra1, Tra2, ..., Tra n ,), where Tra i is a space-time trajectory segment of a voyage, Tra i ={p1,p2,p3,...,p m}, p i =(lon i ,lat i ,v i ,h i ,ins i ,t i ) indicates that at t i The location information at the moment, including longitude information lon i , latitude information lat, speed information v i , ship heading information h i , other location information ins i The steps of characterization processing and feature extraction of spatiotemporal trajectory data are as follows:

[0071] S1: Get the characterized spatiotemporal trajectory dataset Tras to be processed;

[0072] S2: Preprocessing of spatiotemporal trajectory data. The obtained spatiotemporal trajectory data is filled with T-second intervals as basic time points using linear interpolation or other equivalent interpolation methods, including but not limited to linear interpolation, Lagrange interpolation and spline interpolation;

[0073] S3: Standard vectorization of spatiotemporal trajectory data. For each spatiotemporal trajectory data, the part between two adjacent trajectory points is first standardized and then vectorized. This step includes calculating the Euclidean distance between adjacent trajectory points and the normalized arc between their constituent vectors and the reference axis (usually the x-axis). Figure 1shown.

[0074] S31: Read all spatiotemporal trajectory data, perform minimum-maximum normalization, and scale the longitude and latitude data to a specific range. For example, scale to (-1,1), or other applicable ranges;

[0075] S32: Vectorization of the spatiotemporal trajectory data of two adjacent points, assuming that the two adjacent points A = (lon a ,lat a ) and B=(lon b ,lat b ), whose expression is:

[0076]

[0077] In the formula, is the vector of two adjacent points in the empty trajectory, the first component lon b -lon a Indicates the change in longitude from point A to point B. The second component lat b -lat a Indicates the change in latitude from point A to point B.

[0078] S34: Calculate the normalized radian θ between the two point vectors and the positive direction of the x-axis, and its expression is:

[0079]

[0080] In the formula, θ represents the vector The normalized angle formed along the positive direction of the x-axis, its value range is [-1,1], or other applicable normalized intervals.

[0081] S4: Unit circle characterization

[0082] like Figure 2 As shown in Figure 1, unit circle characterization is a character encoding method that maps the normalized arc of a vector to the unit circle. By dividing the unit circle into multiple concentric layers and evenly dividing equiangular areas in each layer, the position of each point on the unit circle can be uniquely identified and accurately located using ASCII character-based numbers.

[0083] S41: Divide the unit circle into layers and equal parts

[0084] Determine the total number of layers (N):

[0085] According to application requirements, the total number of layers N into which the unit circle is divided is set. The more layers N, the higher the encoding accuracy.

[0086] Divide each layer into equal parts (M n ):

[0087] For the nth layer, calculate the number of equal parts M n , the formula is as follows:

[0088] M n =2 n+1

[0089] Where n is the current layer number, 1≤n≤N. For example:

[0090] Layer 1 (n=1): M1=2^(1+1)=4 equiangular regions

[0091] Second layer (n=2): M2=2^(2+1)=8 equiangular regions

[0092] Layer 3 (n=3): M3=2^(3+1)=16 equiangular regions

[0093] And so on, as the number of layers increases, the number of equal portions doubles.

[0094] S42: Generate character encoding

[0095] Each layer corresponds to a base character, starting with the ASCII lowercase letter 'a'. As the number of layers increases, the ASCII code of the base character increases. For example:

[0096] Layer 1: 'a'

[0097] Layer 2: 'b'

[0098] Layer 3: 'c'

[0099] And so on.

[0100] The equiangular regions in each layer are numbered counterclockwise from the positive x-axis, ranging from 0 to

[0101] M n -1. For example:

[0102] The 4 areas on the first layer are numbered a0, a1, a2, and a3

[0103] The 8 regions in the second layer are numbered b0, b1, ..., b7

[0104] The 16 regions in the third layer are numbered c0, c1, ..., c15

[0105] Coding example:

[0106] Layer 1, Area 0: a0

[0107] Layer 2, Area 7: b7

[0108] Layer 3, Area 15: c15

[0109] Decoding example:

[0110] Given an encoded string "c15".

[0111] Layer calculation:

[0112] n=ASCII(′c′)-ASCII(′a′)+1=3

[0113] The number of equal portions of the third layer M3=16.

[0114] The recovery area is the 3rd layer, the 15th isotropic area.

[0115] The rules for layering and area division of the unit circle:

[0116] Layer Number:

[0117] According to the normalized radius r of the offset point, determine the circle layer number where it is located. For example:

[0118] If r is within the range of level 1, the number is a.

[0119] If r is within the range of the second layer, the number is b.

[0120] The radius range of the circle layer is determined by the division rules and can be set according to the accuracy requirements.

[0121] n=1 (first level): The unit circle is divided into four equal-angle regions, numbered a0, a1, a2, and a3 in clockwise direction starting from the positive x-axis.

[0122] n=2 (second layer): the circle is expanded to the second layer and divided into 8 equiangular regions, which are numbered b0, b1, ..., b7 in sequence.

[0123] n=3 (third layer): further expanded to the third layer, divided into 16 equiangular regions, numbered c0, c1, ..., c15.

[0124] Expand to the nth layer: The layering and area division can be expanded infinitely according to actual needs, and the area numbering rules remain consistent.

[0125] 5: Vector character representation

[0126] Vector character representation is a method of converting two-dimensional vector information into character encoding. By combining the unit circle characterization technology, the length and direction of the vector are encoded separately, achieving concise representation and efficient retrieval of vector data.

[0127] S51: Vector length normalization and layer number determination:

[0128] Normalize the length L of the vector to the unit interval [0,1]. The normalization formula is as follows:

[0129]

[0130] Among them, L_max is the maximum value of the vector length, which is determined according to the data set.

[0131] The unit interval [0,1] is evenly divided according to the total number of layers N to obtain an array of segmentation points.

[0132] The insertion position n of the normalized length L_normalized is determined by binary search as the number of layers corresponding to the vector length.

[0133] For example, if N=3 and the split points are {0, 1 / 3, 2 / 3, 1}, then:

[0134] 0≤L_normalized<1 / 3, n=1

[0135] 1 / 3≤L_normalized<2 / 3, n=2

[0136] 2 / 3≤L_normalized≤1, n=3

[0137] S52: Vector direction encoding:

[0138] Calculate the angle θ between the vector and the positive direction of the x-axis, in the range [0°, 360°), counterclockwise.

[0139] Divide the angle interval and determine the direction area number (μ)

[0140] According to the number of equal parts M of the nth layer n , divide 360 ​​degrees into M n An equiangular area.

[0141] Use the binary search method to determine the angle region number μ where θ is located. The calculation formula is as follows:

[0142]

[0143] S53: Generate vector character encoding

[0144] The layer number character C corresponding to the length layer n Combined with the direction area number μ, it forms the character code C of the vector n μ. For example:

[0145] Layer 2, Area 1: b1

[0146] Layer 3, Area 3: c3

[0147] This process effectively combines vector information with the character encoding of the unit circle, converting vector data into a concise character encoding form. Such a conversion not only facilitates the subsequent application of the data, but also greatly improves the efficiency of rapid retrieval. Figure 3 shown.

[0148] The position of the offset point in the unit circle is determined by calculating the normalized radius r and the angle θ.

[0149] The arrows in the figure indicate that the offset of P1 falls into a certain layer and area, such as c3.

[0150] Area Code:

[0151] According to the angle θ of the offset point, determine the angle zone number where it is located. For example:

[0152] If θ is within an angular range in the first quadrant, the number is 0.

[0153] And so on, the area number where θ is located is added to the layer number to form a complete character code.

[0154] Trajectory sequence generation:

[0155] The initial point P0 stores its complete latitude and longitude coordinates, and subsequent track points are stored according to the character encoding of their offsets. For example:

[0156] Assuming that the character codes corresponding to the trajectory offset are a0, b1, c2..., the generated trajectory sequence is:

[0157] P0:(121.5,30.3), trajectory encoding: {a0,b1,c2,...}

[0158] S6: Feature extraction

[0159] Through the above process, it is assumed that the spatiotemporal trajectory data of a voyage is converted into a character sequence F. Next, the N-gram method is used to segment these character sequences, and a specific window length W and step length S are selected to cut each character sequence, thereby generating a series of minimum fine-grained trajectory feature information tokens. This process will generate a new character sequence Y, where Y = (y1, y2, ..., y l ). Where N = (lW) / S, N is an integer, l is the length y of the sequence F i =(f i ,f i+1 ,...,f i+W ), f is the i-th element of the character sequence F. This process effectively completes the feature extraction of trajectory data and retains the upper and lower continuous trajectory information, providing a basis for subsequent analysis and processing. For the feature extraction process, we can Figure 4Repeat the above steps to complete the extraction of spatiotemporal data features for all voyages.

[0160] From the perspective of data, representation needs to extract effective information from the data and reflect the true distribution of the data. From the perspective of the algorithm, the representation needs to meet the application format of the algorithm, facilitate the learning of the algorithm, and improve the accuracy of prediction. Trajectory representation aims to represent different trajectories as vectors in the same dimensional space to facilitate downstream task applications.

[0161] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solution of the present invention can be modified or replaced by equivalents without departing from the purpose and scope of the technical solution, which should be included in the scope of the claims of the present invention.

Claims

1. A method for characterizing and extracting features of spatiotemporal trajectory data, characterized in that: Includes the following step: S1: Obtain the spatiotemporal trajectory dataset to be processed; S2: Preprocessing the spatiotemporal trajectory dataset, including: S21: sampling the spatiotemporal trajectory dataset at a preset time interval; S22: interpolating the sampled spatiotemporal trajectory data to fill in the missing data; S3: Perform standard vectorization processing on the preprocessed spatiotemporal trajectory data, including: S31: normalizing the longitude and latitude information in the spatiotemporal trajectory data so that it falls within a preset range; S32: Calculate the vector between adjacent trajectory points and express the vector in a two-dimensional coordinate form; S33: Calculate the normalized radian between the vector and the reference axis; S4: performing unit circle characterization on the normalized radians, including: S41: Divide the unit circle into a preset number of layers, and divide each layer into a preset number of equal parts; S42: determining the position of the vector on the unit circle according to the normalized radian, and expressing the position using character code; S5: combining the character code with the length information of the vector to obtain a character representation of the vector; S6: Extracting features from the character representation, including: S61: Segment the character representation into sequences of preset length and step size; S62: Using the sequence as feature information of the spatiotemporal trajectory data.

2. The method for characterizing and extracting features of spatiotemporal trajectory data according to claim 1, characterized in that: The preset time interval is T seconds.

3. The method for characterizing and extracting features of spatiotemporal trajectory data according to claim 1, characterized in that: The interpolation process adopts linear interpolation method, Lagrange interpolation method or spline interpolation method.

4. The method for characterizing and extracting features of spatiotemporal trajectory data according to claim 1, characterized in that: The preset interval is (-1, 1).

5. The method for characterizing and extracting features of spatiotemporal trajectory data according to claim 1, characterized in that: The reference axis is any one of the x-axis, the y-axis and the z-axis.

6. The method for characterizing and extracting features of spatiotemporal trajectory data according to claim 1, characterized in that: The character code starts with the lowercase letter 'a' in ASCII code.

7. The method for characterizing and extracting features of spatiotemporal trajectory data according to claim 1, characterized in that: In the feature extraction, the preset length is L and the preset step size is S.

8. A spatiotemporal trajectory data characterization and feature extraction system, characterized by: include: A data acquisition module, used to acquire the spatiotemporal trajectory data set to be processed; The preprocessing module is used to preprocess the spatiotemporal trajectory dataset, including: A sampling module, used to sample the spatiotemporal trajectory dataset at a preset time interval; An interpolation module, used for performing interpolation processing on the sampled spatiotemporal trajectory data to fill in missing data; A vectorization module is used to perform standard vectorization processing on the preprocessed spatiotemporal trajectory data, including: A standardization module, used for standardizing the latitude and longitude information in the spatiotemporal trajectory data; A vector calculation module, used for calculating the vectors between adjacent trajectory points and expressing the vectors in a two-dimensional coordinate form; A radian calculation module, used for calculating the normalized radian between the vector and the reference axis; A characterization module, used for performing unit circle characterization on the normalized radian, comprises: A unit circle division module, used for dividing the unit circle into a preset number of layers, and dividing each layer into a preset number of equal parts; A position determination module, used to determine the position of the vector on the unit circle according to the normalized radian; An encoding module, used for representing the position using a character encoding; A combining module, used for combining the character code with the length information of the vector to obtain a character representation of the vector; A feature extraction module, used for extracting features from the character representation, comprises: A segmentation module, used for segmenting the characterized representation into sequences of preset length and step size; The sequence generation module is used to use the sequence as feature information of the spatiotemporal trajectory data.

9. The spatiotemporal trajectory data characterization and feature extraction system according to claim 8, characterized in that: The data acquisition module is connected to an external data source and is used to receive a spatiotemporal trajectory data set provided by the external data source.

10. The spatiotemporal trajectory data characterization and feature extraction system according to claim 8, characterized in that: The feature extraction module is connected to the subsequent analysis and processing module and is used to transmit the feature information to the subsequent analysis and processing module.

Images