Moving target sparse trajectory classification method and system based on real-time coordinates
By using a sparse trajectory classification method with real-time coordinates, combined with multi-dimensional feature calculation and resampling algorithms, the problem of low positioning accuracy for large-scale, long-distance moving targets is solved, achieving low-cost, all-weather, 360-degree multi-target trajectory monitoring, suitable for harsh environments.
Patent Information
- Application Number
- CN202411404881.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-10-10
- Publication Date
- 2024-11-08
- Estimated Expiration
- 2044-10-10
AI Technical Summary
Existing technologies have low accuracy in locating moving targets over large areas and long distances, suffer from large errors when relying on Bluetooth beacons, cannot effectively classify trajectories in multi-target scenarios, and have high sensor costs, making it impossible to provide 360-degree all-weather coverage in harsh environments.
A sparse trajectory classification method for moving targets based on real-time coordinates is adopted. Feature values are calculated through distance, angle, speed and type dimensions. Combined with a resampling algorithm, high-precision classification of sparse trajectories and accurate tracking of multiple targets are achieved. Low-power sensors are used for area monitoring.
It achieves high-precision trajectory classification within a range of 100m to 1km, making it suitable for low-cost monitoring in special situations such as plateaus, uninhabited areas, and jungles. It has all-weather, 360-degree coverage capabilities and strong applicability.
Smart Images

Figure CN118918393A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of regional monitoring, and in particular to a method and system for classifying sparse trajectories of moving targets based on real-time coordinates. Background Art
[0002] Regional monitoring technology is based on multi-dimensional sensing wireless monitoring technology. By collecting real-time sensing signals from various types of sensors such as infrared, sound and vibration, it collaboratively perceives the location information of the target and combines it with three-point positioning technology. It can locate the coordinate points of targets in the area in all-weather, multi-dimensional and unattended situations, track and obtain the real-time location information of moving targets, and restore their movement trajectories. It can meet the needs of intelligence collection and situational awareness in the fields of regional monitoring, intrusion detection, and dangerous target trajectory analysis.
[0003] The patent document with application number 202111218698.0 records "a method, device, computer equipment and storage medium for determining motion trajectory". The method includes the following steps: 1. Obtain the first coordinate point of the first beacon whose distance from the target object is less than or equal to the preset threshold, and construct a polar coordinate system with the first coordinate point as the origin; 2. When the target object moves to the Nth position, the distance between the Nth position and the first beacon is greater than the preset threshold, and there is no second beacon whose distance from the Nth position is less than or equal to the preset distance threshold, determine the N polar coordinate points in the polar coordinate system when the target object moves to the Nth position; 3. Fuse the first coordinate point with each polar coordinate point respectively to obtain N fused coordinate points; 4. Determine the motion trajectory of the target object based on the first coordinate point and the N fused coordinate points.
[0004] This method has the following disadvantages: 1. This solution is based on Bluetooth beacons to detect moving targets and transmit signals. The positioning accuracy of Bluetooth beacons is high within a short distance and a small range, but as the distance increases, its transmission distance and positioning accuracy will be greatly reduced. Therefore, it is not suitable for positioning and trajectory detection of moving targets in large and long-distance areas.
[0005] 2. This solution relies on Bluetooth beacons to collect the real-time coordinate points of the moving target to achieve trajectory tracking. Although the number of Bluetooth beacons is reduced, the trajectory of the moving target in this solution is very dependent on the density of the target coordinate points. The more abundant the coordinate points are, the more accurate the target motion trajectory will be. When the coordinate points are sparse, the target trajectory will have a large error.
[0006] 3. This solution only analyzes the trajectory of a single moving target scenario, and does not classify and track the trajectory points corresponding to multiple trajectories of different targets in a multi-target scenario, which has great limitations in application scenarios.
[0007] 4. This solution can only track the trajectory of moving targets. Since the positioning data is collected only by Bluetooth beacon detection, the acquisition method is single and it is impossible to further identify the category of the detected target based on the inherent characteristic information of the target.
[0008] The patent document with application number 202410298624.X records "a method, device, equipment and storage medium for detecting motion trajectories by multiple cameras". The method includes the following steps: 1. The method includes obtaining multiple videos to be identified taken by multiple cameras, and selecting a target video from the multiple videos to be identified based on the target feature information of the target object; 2. For each of the target videos, the motion trajectory points of the target object in the monitoring area are determined based on the target video and the pixel coordinate grid corresponding to the target video, and a time tag is marked for each of the motion trajectory points; 3. The target motion trajectory of the target object in the monitoring area is generated based on each of the motion trajectory points and the time tag corresponding to each of the motion trajectory points.
[0009] This method has the following disadvantages: 1. This type of solution is based on pure visual technology. It obtains the two-dimensional image data of the target object and analyzes it to obtain the target's motion trajectory. Therefore, this type of solution is very dependent on the clarity of the image of the target object, and only supports monitoring at specific angles. It cannot completely achieve 360-degree coverage without blind spots. If there are obstructions to the target or the camera monitoring angle changes, the target will be lost. Therefore, this solution is greatly affected by the external environment and human factors, and cannot achieve all-weather, 360-degree coverage.
[0010] 2. The procurement and subsequent system operation and maintenance costs of this type of solution are relatively high, and it is only suitable for special areas with high-value monitoring fields, such as banks, schools, military camps, etc. However, it is not cost-effective for monitoring scenes in areas with harsh natural environments such as plateaus, deserts, jungles, and uninhabited islands. Summary of the invention
[0011] In view of the above problems, the present invention provides a method and system for classifying sparse trajectories of moving targets based on real-time coordinates for overcoming the above problems or at least partially solving the above problems.
[0012] The present invention provides the following scheme: A method for classifying sparse trajectories of moving targets based on real-time coordinates, comprising: Acquire the attribute information of all the locations to be classified in the area at the current time, wherein the attribute information of each location to be classified includes at least the current time, longitude, latitude and corresponding target type; Based on the regional range boundary, determining whether the target corresponding to each of the position points to be classified is a new track target or an existing track target; Determine the target as a new track, and assign a new track ID number in an incremental manner based on the existing track ID number; Determine that it is an existing track target, and calculate and determine the existing track ID number to which the position point to be classified belongs through the track classification method of real-time coordinates; The real-time coordinate trajectory classification method includes calculating and obtaining four-dimensional feature values between the position point to be classified and each existing trajectory from the distance dimension, angle dimension, speed dimension and type dimension; Determine the four-dimensional feature thresholds corresponding to each of the existing trajectories; Determine the magnitude relationship between each dimension feature value and each dimension feature threshold value one by one; After determining that the dimensional feature value is less than the dimensional feature threshold, determining that the position point to be classified belongs to the existing track corresponding to the dimensional feature threshold, and using the existing track as a track to be selected; All the determined trajectories to be selected are stored in a trajectory list, the number of occurrences of each trajectory is counted, and the trajectory with the largest number of occurrences is determined as the trajectory to which the position point to be classified belongs.
[0013] Preferably: the calculation of the distance dimension includes using The distance formula realizes the calculation of the distance between each of the position points to be classified and each of the existing trajectories; The calculation of the angle dimension includes using a vector dot product formula to calculate the relative angle between each of the position points to be classified and each of the existing trajectories; The calculation of the speed dimension includes calculating the difference between the target ideal travel distance of each of the position points to be classified and the trajectory theoretical travel distance of each existing trajectory.
[0014] Preferably: the use of The distance formula for calculating the distance between each of the to-be-classified positions and each of the existing trajectories includes: Assume The longitude of the position point to be classified is , No. The last coordinate point of the existing trajectory is ; Calculate the The distance between the position point to be classified and the last coordinate point of each existing trajectory is obtained distance values; By taking The maximum and minimum values of the distance values are used to calculate the relative value of the absolute distance. The specific calculation formula is as follows: Where: represents the radius of the Earth, Indicates The latitude of the location point to be classified, Indicates The latitude of the last coordinate point of an existing track, Indicates The longitude of the location to be classified, Indicates The longitude of the last coordinate point of an existing track, Indicates The position points to be classified and The absolute distance between the last coordinate point of an existing track. Indicates The position points to be classified and The relative distance between the last coordinate points of an existing track.
[0015] Preferably, the method of using the vector dot product formula to calculate the relative angle between each of the position points to be classified and each of the existing trajectories includes: Take the first Points to be classified With The last coordinate point of an existing track As the target vector ; Take the last two points of each trajectory and As a reference vector , each position point to be classified is calculated to obtain Angle value , the specific calculation formula is as follows: In the formula, Indicates The position points to be classified and The target vector between the last coordinate points of the existing trajectories, Indicates The last coordinate point of the existing trajectory is The reference vector between the coordinate points, Indicates The position points to be classified and The angle between the last coordinate points of an existing track.
[0016] Preferably, the step of calculating the difference between the target ideal travel distance of each position point to be classified and the trajectory theoretical travel distance of each existing trajectory comprises: Calculate the target ideal travel distance based on the typical speed and sampling time of the location point to be classified; Take the last two points of each existing trajectory to calculate the theoretical travel distance of the trajectory, and calculate the difference between the target ideal travel distance of each position point to be classified and the theoretical travel distance of each existing trajectory. The difference in travel distance; Pick The maximum and minimum values of the travel distance differences are used to calculate the relative value of the absolute distance difference. The specific calculation formula is as follows: Where: Indicates The ideal travel distance of a target with an existing trajectory, Indicates The theoretical travel distance of the existing tracks, Indicates The position points to be classified and The ideal and theoretical distance difference between the existing trajectories, Indicates The position points to be classified and The ideal and theoretical relative distance difference between the existing trajectories.
[0017] Preferably, the step of calculating the difference between the target ideal speed of each position point to be classified and the trajectory theoretical travel speed of each existing trajectory comprises: Take the last two points of each trajectory to calculate the theoretical travel distance of the trajectory and the travel time between the two points; Calculate the theoretical speed of each trajectory; use the relative value method to calculate the relative speed difference. The specific calculation formula is as follows: In the formula, Indicates The target ideal speed of each trajectory, Indicates The theoretical travel speed of the track, Represents the relative velocity difference between two trajectories.
[0018] Preferably: combining the new trajectory ID number and the trajectory ID number to which the to-be-classified position point belongs, the trajectory ID numbers of all the to-be-classified position points at the current moment are returned.
[0019] A moving target sparse trajectory classification system based on real-time coordinates, the system comprising: Input data module, used to clean and integrate raw data; Data flow center module, used to manage, update and store data; A classification module, used to classify the trajectory of the target in the input data based on the above-mentioned sparse trajectory classification method of the moving target based on the real-time coordinates; The visualization module is used for two-dimensional image presentation of targets and trajectories, so as to realize at least input data visualization, classification trajectory visualization and comprehensive data visualization.
[0020] Preferably: the implementation of data management, updating and storage includes: The data is stored in the form of a hash table and uniform processing of the data is achieved based on a resampling algorithm.
[0021] Preferably: the resampling algorithm comprises: Calculate the uniform sampling interval based on the target speed and sampling time ; Starting from the first point of the trajectory, follow the uniform intervals described , interpolate and select new points on the trajectory in turn, and define the new points as , continue with new interpolation; Make sure the current point is not within the two control points and calculate the remaining distance , minus the remaining distance and taking the next control point as the starting point, continue sampling. The specific calculation formula is as follows: In the formula, is a uniform sampling interval, is the current remaining distance, Indicates the trajectory Point and The distance between the points, is the new point obtained by current interpolation, For the next control point, is the current starting point.
[0022] According to the specific embodiments provided by the present invention, the present invention discloses the following technical effects: The embodiment of the present invention provides a method and system for classifying sparse trajectories of moving targets based on real-time coordinates. The method belongs to the field of regional monitoring and can achieve effective trajectory classification prediction for multiple targets, multiple trajectories and sparse coordinate points, and achieve high-precision classification of the target's trajectory through four types of rule features. At the same time, it can achieve regional monitoring within a range of 100m to 1km, and adopts low-power sensors with long working time and low cost. It is suitable for large-scale deployment in the monitoring area without recovery, and the operation and maintenance cost is quite low. Therefore, it is particularly suitable for regional monitoring in some special occasions such as plateaus, uninhabited areas, jungles and other scenes where people are inconvenient to move.
[0023] Of course, any product implementing the present invention does not necessarily need to achieve all of the advantages described above at the same time. BRIEF DESCRIPTION OF THE DRAWINGS
[0024] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention, and for ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.
[0025] Figure 1 Is a flowchart of a method for classifying sparse trajectories of moving targets based on real-time coordinates according to an embodiment of the present invention; Figure 2 is a flow chart of calculation of four latitude categories provided by an embodiment of the present invention; Figure 3 is a schematic diagram of a moving target sparse trajectory classification system based on real-time coordinates provided by an embodiment of the present invention; Figure 4 is an unclassified target coordinate point distribution map provided by an embodiment of the present invention; Figure 5 is a classified target trajectory distribution map provided by an embodiment of the present invention; Figure 6 is a target trajectory distribution diagram after resampling provided by an embodiment of the present invention; Figure 7 It is a schematic diagram of a case where some trajectory predictions fail according to an embodiment of the present invention. DETAILED DESCRIPTION
[0026] The technical scheme in the embodiment of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiment of the present invention. Obviously, the described embodiment is only a part of the embodiment of the present invention, not all of the embodiments. Based on the embodiment of the present invention, all other embodiments obtained by ordinary technicians in this field belong to the scope of protection of the present invention.
[0027] See also Figure 1 , is a moving target sparse trajectory classification method based on real-time coordinates provided by an embodiment of the present invention, such as Figure 1 As shown, the method may include: Acquire the attribute information of all the locations to be classified in the area at the current time, wherein the attribute information of each location to be classified includes at least the current time, longitude, latitude and corresponding target type; Based on the regional range boundary, determining whether the target corresponding to each of the position points to be classified is a new track target or an existing track target; Determine the target as a new track, and assign a new track ID number in an incremental manner based on the existing track ID number; Determine that it is an existing track target, and calculate and determine the existing track ID number to which the position point to be classified belongs through the track classification method of real-time coordinates; Among them, Figure 2 As shown, the trajectory classification method of the real-time coordinates includes calculating and obtaining four-dimensional feature values between the position point to be classified and each existing trajectory from the distance dimension, angle dimension, speed dimension and type dimension; In a specific implementation, the present invention can provide that the distance dimension is calculated by using The distance formula realizes the calculation of the distance between each of the position points to be classified and each of the existing trajectories; The calculation of the angle dimension includes using a vector dot product formula to calculate the relative angle between each of the position points to be classified and each of the existing trajectories; The calculation of the speed dimension includes calculating the difference between the target ideal travel distance of each of the position points to be classified and the trajectory theoretical travel distance of each existing trajectory; The calculation of the type dimension includes calculating the difference between the target ideal speed of each position point to be classified and the trajectory theoretical travel speed of each existing trajectory.
[0028] The adoption The distance formula for calculating the distance between each of the to-be-classified positions and each of the existing trajectories includes: Assume The longitude of the position point to be classified is , No. The last coordinate point of the existing trajectory is ; Calculate the The distance between the position point to be classified and the last coordinate point of each existing trajectory is obtained distance values; By taking The maximum and minimum values of the distance values are used to calculate the relative value of the absolute distance. The specific calculation formula is as follows: Where: represents the radius of the Earth, Indicates The latitude of the location point to be classified, Indicates The latitude of the last coordinate point of an existing track, Indicates The longitude of the location to be classified, Indicates The longitude of the last coordinate point of an existing track, Indicates The position points to be classified and The absolute distance between the last coordinate point of an existing track. Indicates The position points to be classified and The relative distance between the last coordinate points of an existing track.
[0029] The use of the vector dot product formula to calculate the relative angle between each of the position points to be classified and each of the existing trajectories includes: Take the first Points to be classified With The last coordinate point of an existing track As the target vector ; Take the last two points of each trajectory and As a reference vector , each position point to be classified is calculated to obtain Angle value , the specific calculation formula is as follows: In the formula, Indicates The position points to be classified and The target vector between the last coordinate points of the existing trajectories, Indicates The last coordinate point of the existing trajectory is The reference vector between the coordinate points, Indicates The position points to be classified and The angle between the last coordinate points of the existing trajectories is calculated using the vector dot product formula.
[0030] The calculation of the difference between the target ideal travel distance of each position point to be classified and the trajectory theoretical travel distance of each existing trajectory includes: Calculate the target ideal travel distance based on the typical speed and sampling time of the location point to be classified; Take the last two points of each existing trajectory to calculate the theoretical travel distance of the trajectory, and calculate the difference between the target ideal travel distance of each position point to be classified and the theoretical travel distance of each existing trajectory. The difference in travel distance; Pick The maximum and minimum values of the travel distance differences are used to calculate the relative value of the absolute distance difference. The specific calculation formula is as follows: Where: Indicates The ideal travel distance of a target with an existing trajectory, Indicates The theoretical travel distance of the existing tracks, Indicates The position points to be classified and The ideal and theoretical distance difference between the existing trajectories, Indicates The position points to be classified and The ideal and theoretical relative distance difference between the existing trajectories.
[0031] The calculating the difference between the target ideal speed of each position point to be classified and the theoretical travel speed of each existing trajectory comprises: taking the last two points of each trajectory to calculate the theoretical travel distance of the trajectory and the travel time between the two points; Calculate the theoretical speed of each trajectory; use the relative value method to calculate the relative speed difference. The specific calculation formula is as follows: In the formula, Indicates The target ideal speed of each trajectory, Indicates The theoretical travel speed of the track, Represents the relative velocity difference between two trajectories.
[0032] Determine the four-dimensional feature thresholds corresponding to each of the existing trajectories; Determine the magnitude relationship between each dimension feature value and each dimension feature threshold value one by one; After determining that the dimensional feature value is less than the dimensional feature threshold, determining that the position point to be classified belongs to the existing track corresponding to the dimensional feature threshold, and using the existing track as a track to be selected; All the determined to-be-selected trajectories are stored in a trajectory list, the number of occurrences of each trajectory is counted, and the trajectory with the largest number of occurrences is determined as the trajectory to which the position point to be classified belongs.
[0033] In order to facilitate viewing of the classification results, the embodiment of the present invention may further provide a method of combining the new trajectory ID number and the trajectory ID number to which the to-be-classified location point belongs, and returning the trajectory ID numbers of all the to-be-classified location points at the current moment.
[0034] In some application scenarios with a large number of targets, the method provided in the prior art can accurately classify each target by obtaining target-related information, and generate a corresponding motion trajectory for each target. However, as the target moves, the collected points after movement may be relatively sparse. The purpose of the method provided in an embodiment of the present invention is to accurately classify these sparse trajectory points.
[0035] It is understandable that the method provided in the embodiment of the present invention focuses on classifying the existing trajectories to which the location point to be classified belongs after obtaining the attribute information such as the current time, longitude, latitude and corresponding target type of the location point to be classified. The attribute information of the location point to be classified can be obtained using the target positioning and recognition method in the prior art. The existing trajectory can be generated using the target trajectory determination method in the prior art. The target type can be a person or a vehicle. Before implementing the method provided by the present invention, the motion trajectories of all existing targets within the regional range have been established.
[0036] Similarly, when judging whether the target corresponding to each position point to be classified is a new target, the target recognition method in the prior art can also be used for judgment. For example, the target type can judge whether the target is a person or a vehicle, and then judge whether the target is already an existing target within the area based on the area boundary and the features of the extracted target. If it is determined that the target type is a person, the facial features of the face can be extracted, and then compared with the facial features of all existing people within the area to determine whether the target is a new target or an existing track target.
[0037] The method for classifying sparse trajectories of moving targets based on real-time coordinates provided by the embodiment of the present invention can be used in regional monitoring scenarios to solve the problem of low accuracy in classifying the trajectories of moving targets with multiple targets, multiple types, and multiple trajectories, as well as the problem that only sparse coordinate points can be collected for fast-moving targets such as cars due to sensor performance, monitoring area, and coordinate reporting frequency, and the trajectory of the moving targets cannot be accurately tracked. Through experimental verification, the experimental results verify that the proposed scheme has high accuracy and good applicability in classifying different motion trajectories of multiple moving targets and tracking sparse trajectories of moving targets in the region.
[0038] The method provided by the embodiment of the present invention is described in detail below.
[0039] The method provided by the embodiment of the present invention can classify the trajectories of the moving target coordinate points input at different times through a coordinate point classification method, thereby realizing the continuous trajectory recognition of the moving target with sparse coordinate points in the area.
[0040] 1. Input all target information in the area at the current time (attribute information of the location points to be classified). The information of a single target includes the current time, longitude, latitude, target type, etc.; 2. Based on the regional boundary, determine whether the target corresponding to the position point to be classified is a new trajectory target or an existing trajectory target; 3. For new trajectory targets, new trajectory ID numbers are assigned in an incremental manner based on the existing trajectory ID numbers; 4. For existing trajectory targets, calculate the trajectory ID number of the target with the highest probability through the trajectory classification method of real-time coordinates; 5. Combining the trajectory ID numbers of all targets in steps 3 and 4 above, return the trajectory ID numbers of all target information at the current moment.
[0041] In step S4, the trajectory classification method of the real-time coordinates realizes category calculation based on the rules of distance dimension, angle dimension, speed dimension, type dimension, etc., for each existing trajectory target (position point to be classified) , where N is the number of existing targets at the current moment, and is calculated with the last coordinate point of the existing trajectory. , where M is the number of existing trajectories.
[0042] A method for implementing category calculation based on rules such as distance dimension, angle dimension, speed dimension, and type dimension, and setting a feature threshold for each dimension. When the feature value of the dimension is less than the feature threshold, the target can be considered to belong to this track, otherwise, it does not belong to the track. According to the calculation based on the four-dimensional rules, four groups of track lists to be selected can be obtained, which contain optional tracks that meet the feature threshold of the dimension. By counting the number of occurrences of each track, the track with the most occurrences in the four groups of track lists to be selected is calculated, and the track is identified as the target track, and the track ID number of the track is returned. Therefore, when the coordinates of the target point are sparse, the target track can still be classified with high accuracy by relying on dimensions such as distance, angle, speed, and type.
[0043] The distance dimension rule calculation mentioned above proposes a Haversine distance calculation method, in which the distance between each existing trajectory target and the last coordinate point of each existing trajectory is calculated respectively, and the relative distance is calculated in a normalized manner to eliminate the influence of the absolute distance value; In the specific implementation, the Haversine distance formula is used to calculate the distance between a set of longitudes and latitudes in the distance dimension rule calculation. The distance between each existing trajectory target and the last coordinate point of each existing trajectory is calculated separately, that is, each trajectory target calculates the distance M times. At the same time, in order to eliminate the influence of the absolute distance value, the relative distance is calculated in a normalized manner. The specific method is to calculate the relative value of the absolute distance by taking the maximum and minimum values of the M values. The specific calculation formula of the above content is as follows: Where: represents the radius of the Earth, Indicates The latitude of the location point to be classified, Indicates The latitude of the last coordinate point of an existing track, Indicates The longitude of the location to be classified, Indicates The longitude of the last coordinate point of an existing track, Indicates The position points to be classified and The absolute distance between the last coordinate point of an existing track. Indicates The position points to be classified and The relative distance between the last coordinate points of an existing track.
[0044] The angle dimension rule calculation proposes a relative angle calculation based on the vector dot product formula, and calculates the last coordinate point of each existing trajectory target and each existing trajectory as the target vector angle value θ.
[0045] In the specific implementation, the angle dimension rule calculation is performed in the above-mentioned method, and the vector dot product formula is used to realize the relative angle calculation. The last coordinate point of each existing trajectory target and each existing trajectory is taken as the target vector , take the last two points of each trajectory as the reference vector , that is, each trajectory target is calculated Angle value , the specific calculation formula is as follows: In the formula, Indicates The position points to be classified and The target vector between the last coordinate points of the existing trajectories, Indicates The last coordinate point of the existing trajectory is The reference vector between the coordinate points, Indicates The position points to be classified and The angle between the last coordinate points of the existing trajectories is calculated using the vector dot product formula.
[0046] The speed dimension rule calculation proposed a method to calculate the ideal travel distance of the target based on the typical speed and sampling time of the existing trajectory target, that is, the ideal-theoretical distance difference between each existing trajectory target and each trajectory target. At the same time, the relative distance difference is calculated in a normalized way to eliminate the dimensional influence of the absolute distance difference.
[0047] In the specific implementation, the speed dimension rule is calculated, and the ideal travel distance of the target is calculated according to the typical speed and sampling time of the existing trajectory target, where the existing trajectory target types include people and vehicles. Different typical speeds can be set according to the specific scene; the last two points of each trajectory are taken to calculate the theoretical travel distance of the trajectory, and the difference between the target ideal travel distance of each existing trajectory target and the trajectory theoretical travel distance of each existing trajectory is calculated, that is, each trajectory target calculates the ideal-theoretical distance difference M times. At the same time, in order to eliminate the dimensional influence of the absolute distance difference, the relative distance difference is calculated in a normalized manner. The specific method is to take The maximum and minimum values among the values are used to calculate the relative value of the absolute distance difference. The specific calculation formula for the above content is as follows: Where: Indicates The ideal travel distance of a target with an existing trajectory, Indicates The theoretical travel distance of the existing tracks, Indicates The position points to be classified and The ideal and theoretical distance difference between the existing tracks, in km, Indicates The position points to be classified and The ideal and theoretical relative distance difference between the existing trajectories, unitless.
[0048] The type dimension rule calculation proposes a method of calculating the theoretical travel speed of each trajectory based on the typical speed of the existing trajectory target as the target ideal speed, and calculating the relative speed difference using a relative value method.
[0049] In the specific implementation, in the type dimension rule calculation, the typical speed of the existing trajectory target is recorded as the target ideal speed, where the existing trajectory target types include people and vehicles. Different typical speeds can be set according to specific scenarios; the last two points of each trajectory are taken to calculate the theoretical travel distance of the trajectory and the travel time between the two points, and the theoretical travel speed of each trajectory is calculated; at the same time, the relative speed difference is calculated by the relative value method. The specific calculation formula is as follows: In the formula, Indicates The target ideal speed of each trajectory, Indicates The theoretical travel speed of the track, Represents the relative velocity difference between two trajectories.
[0050] In summary, for the existing trajectory target in step 4, the different dimensional feature values between each existing trajectory can be calculated, and each dimensional feature threshold can be set. When the dimensional feature value is less than the feature threshold, it can be considered that the target belongs to this trajectory, otherwise, it does not belong to this trajectory. Therefore, under the above four dimensional rules, four groups of trajectory lists to be selected can be obtained, which contain optional trajectories that meet the dimensional feature threshold; by counting the number of occurrences of each trajectory, the trajectory with the most occurrences in the four groups of trajectory lists to be selected is calculated, and the trajectory is identified as the trajectory to which the target belongs, and the trajectory ID number of the trajectory is returned.
[0051] See also Figure 3, the embodiment of the present invention can also provide a moving target sparse trajectory classification system based on real-time coordinates, such as Figure 3 As shown, the system may include: Input data module, used to clean and integrate raw data; The data flow center module is used to manage, update and store data. When implemented, the embodiment of the present invention can store data in the form of a hash table and achieve uniform processing of data based on a resampling algorithm.
[0052] In the data stream center module, the resampling algorithm is designed to solve the problems that the original real-time input data acquisition is easily interfered by network factors, hardware factors, etc., resulting in packet loss, delay, etc., which leads to unstable input data frequency update, thus affecting subsequent algorithm judgment, etc. The resampling algorithm includes: Calculate the uniform sampling interval based on the target speed and sampling time ; Starting from the first point of the trajectory, follow the uniform intervals described , interpolate and select new points on the trajectory in turn, and define the new points as , continue with new interpolation; Make sure the current point is not within the two control points and calculate the remaining distance , minus the remaining distance and taking the next control point as the starting point, continue sampling. The specific calculation formula is as follows: In the formula, is a uniform sampling interval, is the current remaining distance, Indicates the trajectory Point and The distance between the points, is the new point obtained by current interpolation, For the next control point, is the current starting point.
[0053] A classification module, used to classify the trajectory of the target in the input data based on the above-mentioned sparse trajectory classification method of the moving target based on the real-time coordinates; The visualization module is used for two-dimensional image presentation of targets and trajectories, so as to realize at least input data visualization, classification trajectory visualization and comprehensive data visualization.
[0054] The method provided by the embodiment of the present invention can achieve effective classification prediction for sparse trajectories. In a smaller area, such as 100m to 1km, if the target is fast, such as a car, only 3 to 10 position points are detected, so the connection between points is poor, which is a sparse trajectory; while for targets with slower speeds, such as pedestrians, more position points are detected, up to more than 100. Therefore, this method achieves high-precision classification of targets through four types of rule features, especially maintaining good applicability for sparse targets.
[0055] In order to verify the purpose that can be achieved by the present invention, a sparse trajectory classification method for moving targets based on real-time coordinates is proposed. By identifying targets within a range of 2 kilometers, the targets include people and vehicles. The typical speeds are set to 5km / h and 50km / h respectively. The perception data includes reporting time, longitude, latitude and target type. The data reporting interval is 10s.
[0056] The present invention uses the above method to effectively classify the trajectories of moving targets and realize visual display.
[0057] Figure 4 It is represented as a distribution diagram of classified target coordinate points. Since multiple target trajectories are prone to cross-section, the human eye cannot effectively distinguish the trajectory of each target.
[0058] Figure 5 For data classified by the algorithm, different target trajectories are represented by lines of different colors, and key points are used to represent each sampling point. Due to adverse factors such as hardware and network, the sampling points cannot be evenly sampled, so they are processed by the resampling algorithm.
[0059] like Figure 6 As shown in the figure, the distance between the sampling points of each trajectory remains consistent at this time, which also provides a good foundation for future trajectory prediction based on data.
[0060] After testing on 10 comprehensive data sets, the algorithm's overall accuracy can reach 92.02%. Figure 7 Shows a failure scenario where the prediction of subsequent points on a trajectory is based on the previous prediction and is prone to continuous errors, but a successful prediction is re-achieved at the intersection point.
[0061] In summary, the sparse trajectory classification method for moving targets based on real-time coordinates provided by the present invention can achieve effective classification prediction for multiple targets, multiple trajectories and sparse coordinate points. In a smaller area, such as 100m to 1km, if the target is fast, such as a car, only 3 to 10 position points are detected, so the connection between points is poor, which is a sparse trajectory; while for targets with slower speeds, such as pedestrians, more position points are detected, up to more than 100. Therefore, this method achieves high-precision classification of the target's trajectory through four types of rule features, especially for sparse position point targets, it still maintains good applicability.
[0062] At the same time, it can realize regional monitoring within the range of 100m to 1km, and uses low-power sensors, with the characteristics of long working time and low cost. It is suitable for large-scale scattering deployment in the monitoring area without the need for recovery, and the operation and maintenance cost is quite low. Therefore, it is particularly suitable for regional monitoring in some special occasions such as plateaus, uninhabited areas, jungles and other scenes where people have difficulty in moving.
[0063] In addition, infrared, sound, vibration and other sensors can be used to collect characteristic signals of moving targets. This type of signal can sense the target in all weather conditions and 360 degrees, and can further identify the type of moving target based on the target object's infrared characteristics, voiceprint characteristics, etc., which will provide richer and more accurate intelligence information for regional monitoring scenarios.
[0064] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "comprise a ..." do not exclude the existence of other identical elements in the process, method, article or device including the elements.
[0065] It can be seen from the above description of the implementation mode that those skilled in the art can clearly understand that the present invention can be implemented by means of software plus a necessary general hardware platform. Based on such an understanding, the technical solution of the present invention can be essentially or partly contributed to the prior art in the form of a software product, which can be stored in a storage medium such as ROM / RAM, a magnetic disk, an optical disk, etc., and includes several instructions for enabling a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in the various embodiments of the present invention or certain parts of the embodiments.
[0066] Each embodiment in this specification is described in a progressive manner, and the same or similar parts between the embodiments can refer to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the system or system embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can refer to the partial description of the method embodiment. The system and system embodiments described above are merely schematic, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without creative work.
[0067] The above description is only a preferred embodiment of the present invention and is not intended to limit the protection scope of the present invention. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention are included in the protection scope of the present invention.
Claims
1. A method for classifying sparse trajectories of moving targets based on real-time coordinates, characterized in that: include: Acquire the attribute information of all the locations to be classified in the area at the current time, wherein the attribute information of each location to be classified includes at least the current time, longitude, latitude and corresponding target type; Based on the regional range boundary, determining whether the target corresponding to each of the position points to be classified is a new track target or an existing track target; Determine the target as a new track, and assign a new track ID number in an incremental manner based on the existing track ID number; Determine that it is an existing track target, and calculate and determine the existing track ID number to which the position point to be classified belongs through the track classification method of real-time coordinates; The real-time coordinate trajectory classification method includes calculating and obtaining four-dimensional feature values between the position point to be classified and each existing trajectory from the distance dimension, angle dimension, speed dimension and type dimension; Determine the four-dimensional feature thresholds corresponding to each of the existing trajectories; Determine the magnitude relationship between each dimensional feature value and each dimensional feature threshold value one by one; After determining that the dimensional feature value is less than the dimensional feature threshold, determining that the position point to be classified belongs to the existing track corresponding to the dimensional feature threshold, and using the existing track as a track to be selected; All the determined trajectories to be selected are stored in a trajectory list, the number of occurrences of each trajectory is counted, and the trajectory with the largest number of occurrences is determined as the trajectory to which the position point to be classified belongs.
2. The method for classifying sparse trajectories of moving targets based on real-time coordinates according to claim 1, characterized in that: The distance dimension is calculated by using The distance formula realizes the calculation of the distance between each of the position points to be classified and each of the existing trajectories; The calculation of the angle dimension includes using a vector dot product formula to calculate the relative angle between each of the position points to be classified and each of the existing trajectories; The calculation of the speed dimension includes calculating the difference between the target ideal travel distance of each of the position points to be classified and the trajectory theoretical travel distance of each existing trajectory.
3. The method for classifying sparse trajectories of moving targets based on real-time coordinates according to claim 2, characterized in that: The adoption The distance formula for calculating the distance between each of the to-be-classified positions and each of the existing trajectories includes: Assume The longitude of the position point to be classified is , No. The last coordinate point of the existing trajectory is ; Calculate the The distance between the position point to be classified and the last coordinate point of each existing trajectory is obtained distance values; By taking The maximum and minimum values of the distance values are used to calculate the relative value of the absolute distance. The specific calculation formula is as follows: Where: represents the radius of the Earth, Indicates The latitude of the location point to be classified, Indicates The latitude of the last coordinate point of an existing track, Indicates The longitude of the location point to be classified, Indicates The longitude of the last coordinate point of an existing track, Indicates The position points to be classified and The absolute distance between the last coordinate point of an existing track. Indicates The position points to be classified and The relative distance between the last coordinate points of an existing track.
4. The method for classifying sparse trajectories of moving targets based on real-time coordinates according to claim 2, characterized in that: The use of the vector dot product formula to calculate the relative angle between each of the position points to be classified and each of the existing trajectories includes: Take the first Points to be classified With The last coordinate point of an existing track As the target vector ; Take the last two points of each trajectory and As a reference vector , each position point to be classified is calculated to obtain Angle value , the specific calculation formula is as follows: In the formula, Indicates The position points to be classified and The target vector between the last coordinate points of the existing trajectories, Indicates The last coordinate point of the existing trajectory is The reference vector between the coordinate points, Indicates The position points to be classified and The angle between the last coordinate points of an existing track.
5. The method for classifying sparse trajectories of moving targets based on real-time coordinates according to claim 2, characterized in that: The calculating of the difference between the target ideal travel distance of each position point to be classified and the trajectory theoretical travel distance of each existing trajectory comprises: Calculate the target ideal travel distance according to the typical speed and sampling time of the position point to be classified; Take the last two points of each existing trajectory to calculate the trajectory theoretical travel distance, and calculate the difference between the target ideal travel distance of each position point to be classified and the trajectory theoretical travel distance of each existing trajectory. The difference in travel distance; Pick The maximum and minimum values of the travel distance differences are used to calculate the relative value of the absolute distance difference. The specific calculation formula is as follows: Where: Indicates The ideal travel distance of a target with an existing trajectory, Indicates The theoretical travel distance of the existing tracks, Indicates The position points to be classified and The ideal and theoretical distance difference between the existing trajectories, Indicates The position points to be classified and The ideal and theoretical relative distance difference between the existing trajectories.
6. The method for classifying sparse trajectories of moving targets based on real-time coordinates according to claim 2, characterized in that: The step of calculating the difference between the target ideal speed of each position point to be classified and the theoretical speed of each existing trajectory includes: Take the last two points of each trajectory to calculate the theoretical travel distance of the trajectory and the travel time between the two points; Calculate the theoretical speed of each trajectory; use the relative value method to calculate the relative speed difference. The specific calculation formula is as follows: In the formula, Indicates The target ideal speed of each trajectory, Indicates The theoretical travel speed of the track, Represents the relative velocity difference between two trajectories.
7. The method for classifying sparse trajectories of moving targets based on real-time coordinates according to claim 1, characterized in that: Combine the new track ID number and the track ID number to which the to-be-classified position point belongs, and return the track ID numbers of all the to-be-classified position points at the current moment.
8. A moving target sparse trajectory classification system based on real-time coordinates, characterized in that: The system comprises: Input data module, used to clean and integrate raw data; Data flow center module, used to manage, update and store data; A classification module, used for classifying the trajectory of the target in the input data based on the sparse trajectory classification method of the moving target based on the real-time coordinates according to any one of claims 1 to 7; The visualization module is used for two-dimensional image presentation of targets and trajectories, so as to realize at least input data visualization, classification trajectory visualization and comprehensive data visualization.
9. The moving target sparse trajectory classification system based on real-time coordinates according to claim 8, characterized in that: The implementation of managing, updating and storing data includes: The data is stored in the form of a hash table and uniform data processing is achieved based on a resampling algorithm.
10. The moving target sparse trajectory classification system based on real-time coordinates according to claim 9, characterized in that: The resampling algorithm includes: Calculate the uniform sampling interval based on the target speed and sampling time ; Starting from the first point of the trajectory, follow the uniform intervals described , interpolate and select new points on the trajectory in turn, and define the new points as , continue with new interpolation; Make sure the current point is not within the two control points and calculate the remaining distance , minus the remaining distance and taking the next control point as the starting point, continue sampling. The specific calculation formula is as follows: In the formula, is a uniform sampling interval, is the current remaining distance, Indicates the trajectory Point and The distance between the points, is the new point obtained by current interpolation, For the next control point, is the current starting point.
Citation Information
Patent Citations
Motion track determination method and device, computer equipment and storage medium
CN113660724A
Multi-camera motion trail detection method, device and equipment and storage medium
CN117896626A
Method for controlling mobile robot, device, and mobile robot system
CN110678822A
Robot self-following method and unmanned environmental sanitation robot
CN114510021A
Local path planning control method and device, robot and storage medium
CN114995464A
