Space-time three-dimensional traffic state dynamic sensing method and system based on air-ground cooperation

Abstract

The invention discloses a space-time three-dimensional traffic state dynamic sensing method and system based on air-ground cooperation, and the method comprises the following steps: obtaining road traffic video data at different height visual angles, and constructing an air-based traffic information data set; constructing a deep learning neural network, and training the deep learning neural network based on the air-based traffic data set and the ground-based traffic data set to obtain a traffic target detection model; constructing an air-ground traffic information fusion model, and training the air-ground traffic information fusion model based on the air-based traffic data set and the ground-based traffic data set to obtain a dynamic perception model; acquiring real-time air-based traffic data and real-time foundation traffic data, and acquiring traffic target track information by using the traffic target detection model; and obtaining lane-level traffic parameters and vehicle-level traffic parameters based on the traffic target trajectory information by using the dynamic perception model.

Description

Technical Field

[0001] The present invention belongs to the interdisciplinary technical field of unmanned system technology and computer vision technology, and specifically relates to a method and system for dynamic perception of spatiotemporal traffic status based on air-ground collaboration. Background Art

[0002] With the development of information and intelligent technologies, the concept of smart cities has permeated every aspect of production and life. As a key component of smart cities, smart transportation is facing an increasing demand for traffic status assessment and perception. Traditional traffic status perception methods primarily rely on devices such as gantries, sensor coils, and cameras. This approach suffers from issues such as poor information fusion and difficulty in representing traffic status.

[0003] Compared to traditional traffic status perception methods, drones offer significant advantages in terms of high flexibility, wide viewing angle, low cost, and three-dimensional information collection. Therefore, they have the potential to better perceive road traffic status information. However, compared to traditional roadside camera state perception, drones also suffer from issues such as limited endurance. Furthermore, with the implementation of intelligent transportation systems and the application of data-driven modeling methods, high-resolution traffic status data, such as vehicle-level and lane-level micro-traffic parameters, has attracted increasing attention. This has also provided new insights into three-dimensional traffic perception. Given this technical background, a method for dynamic spatiotemporal traffic status perception based on air-ground collaboration would be desirable. By integrating drones and road cameras as front-end information acquisition devices, feeding video streams into a deep learning model using convolutional neural networks, and achieving a two-dimensional spatial and temporal assessment model of road traffic conditions through air-ground information perception, thus enabling dynamic three-dimensional perception of the road network. This would effectively improve the current state of traffic status perception, which relies on fragmented traffic status information collected by devices such as traffic gantries, and enhance the credibility and real-time performance of road traffic status perception, thereby achieving dynamic three-dimensional traffic status perception. Summary of the Invention

[0004] The present invention aims to solve the deficiencies of the prior art and provides the following solutions:

[0005] A method for dynamic perception of spatiotemporal traffic status based on air-ground collaboration includes the following steps:

[0006] Acquire road traffic video data from different height perspectives and construct an airborne traffic information dataset;

[0007] Constructing a deep learning neural network, and training the deep learning neural network based on the air-based traffic dataset and the ground-based traffic dataset to obtain a traffic target detection model;

[0008] Constructing an air-ground traffic information fusion model, and training the air-ground traffic information fusion model based on the air-based traffic dataset and the ground-based traffic dataset to obtain a dynamic perception model;

[0009] Acquire real-time air-based traffic data and real-time ground-based traffic data, and obtain traffic target trajectory information using the traffic target detection model;

[0010] The dynamic perception model is used to obtain lane-level traffic parameters and vehicle-level traffic parameters based on the traffic target trajectory information.

[0011] Preferably, the method for constructing the air-based traffic dataset includes:

[0012] Using drones to collect air-based traffic video data;

[0013] performing data cleaning on the air-based traffic video data to obtain cleaned video data;

[0014] Intercepting the cleaned video data frame by frame to create an image data set;

[0015] Manually annotate the image dataset to obtain the air-based traffic dataset.

[0016] Preferably, the method for obtaining the traffic target trajectory information includes:

[0017] Use drones to obtain the original trajectory of the vehicle;

[0018] Using wavelet analysis to identify outliers in the original trajectory of the vehicle and perform corrections;

[0019] The corrected trajectory is subjected to noise reduction using a wavelet filter to obtain the traffic target trajectory information.

[0020] Preferably, the lane-level traffic parameters include: lane time average speed, lane space average speed, lane traffic flow and lane traffic density;

[0021] The lane time average speed is obtained by the average longitudinal speed of vehicles traveling on the lane during the time interval;

[0022] The lane space average speed is:

[0023]

[0024] in, represents the lane space average speed, represents the lane time average speed, represents the variance of the lane’s time-averaged speed, and r represents the rth lane;

[0025] The traffic volume of the lane is:

[0026]

[0027] Among them, q r represents the lane traffic volume, m r It represents the number of vehicles passing a certain section on the lane, and ΔT represents the time interval;

[0028] The lane traffic density is:

[0029]

[0030] Among them, k r (t) represents the lane traffic density, represents the average number of vehicles passing through a certain section, l r Indicates the length of the local lane;

[0031] The vehicle-level traffic parameters include: vehicle instantaneous speed, headway and headway time;

[0032] The instantaneous speed of the vehicle is:

[0033]

[0034] in, represents the instantaneous speed of the vehicle, represents the traffic target trajectory information, τ represents the vehicle-level parameter calculation time interval, and i represents the i-th vehicle;

[0035] The headway is:

[0036]

[0037] in, Indicates the headway between vehicles. The longitudinal position coordinates representing the traffic trajectory information;

[0038] The headway time is:

[0039]

[0040] in, Indicates the headway time, Represents the longitudinal component of the vehicle's instantaneous velocity.

[0041] The present invention also provides a spatiotemporal traffic state dynamic perception system based on air-ground collaboration, wherein the perception system applies any of the above-mentioned perception methods and comprises: a data set construction module, a first model construction module, a second model construction module, a trajectory acquisition module, and a parameter perception module;

[0042] The data set construction module is used to obtain road traffic video data at different heights and construct an air-based traffic information data set;

[0043] The first model building module is used to build a deep learning neural network, and train the deep learning neural network based on the air-based traffic dataset and the ground-based traffic dataset to obtain a traffic target detection model;

[0044] The second model building module is used to build an air-ground traffic information fusion model, and train the air-ground traffic information fusion model based on the air-based traffic dataset and the ground-based traffic dataset to obtain a dynamic perception model;

[0045] The trajectory acquisition module is used to acquire real-time air-based traffic data and real-time ground-based traffic data, and to acquire traffic target trajectory information using the traffic target detection model;

[0046] The parameter perception module utilizes the dynamic perception model to obtain lane-level traffic parameters and vehicle-level traffic parameters based on the traffic target trajectory information.

[0047] Preferably, the data set construction module includes: a video data acquisition unit, a data processing unit, an image data acquisition unit and a labeling unit;

[0048] The video data acquisition unit collects air-based traffic video data using a drone;

[0049] The data processing unit is used to clean the air-based traffic video data to obtain cleaned video data;

[0050] The image data acquisition unit is used to intercept the cleaned video data frame by frame to produce an image data set;

[0051] The labeling unit is used to manually label the image dataset to obtain the air-based traffic dataset.

[0052] Preferably, the trajectory acquisition module includes: an original trajectory acquisition unit, a correction unit and a noise reduction unit;

[0053] The original trajectory acquisition unit acquires the original trajectory of the vehicle using a drone;

[0054] The correction unit uses wavelet analysis to identify abnormal values ​​of the original trajectory of the vehicle and perform corrections;

[0055] The noise reduction unit uses a wavelet filter to reduce noise on the corrected trajectory to obtain the traffic target trajectory information.

[0056] Preferably, the lane-level traffic parameters include: lane time average speed, lane space average speed, lane traffic flow and lane traffic density;

[0057] The lane time average speed is obtained by the average longitudinal speed of vehicles traveling on the lane during the time interval;

[0058] The lane space average speed is:

[0059]

[0060] in, represents the lane space average speed, represents the lane time average speed, represents the variance of the lane’s time-averaged speed, and r represents the rth lane;

[0061] The traffic volume of the lane is:

[0062]

[0063] Among them, q r represents the lane traffic volume, m r It represents the number of vehicles passing a certain section on the lane, and ΔT represents the time interval;

[0064] The lane traffic density is:

[0065]

[0066] Among them, k r (t) represents the lane traffic density, represents the average number of vehicles passing through a certain section, l r Indicates the length of the local lane;

[0067] The vehicle-level traffic parameters include: vehicle instantaneous speed, headway and headway time;

[0068] The instantaneous speed of the vehicle is:

[0069]

[0070] in, represents the instantaneous speed of the vehicle, represents the traffic target trajectory information, τ represents the vehicle-level parameter calculation time interval, and i represents the i-th vehicle;

[0071] The headway is:

[0072]

[0073] in, Indicates the headway between vehicles. The longitudinal position coordinates representing the traffic trajectory information;

[0074] The headway time is:

[0075]

[0076] in, Indicates the headway time, Represents the longitudinal component of the vehicle's instantaneous velocity.

[0077] Compared with the prior art, the present invention has the following beneficial effects:

[0078] The present invention integrates traffic video information from real-time video stream acquisition devices from air-based and ground-based sources, combines target detection and target tracking in computer vision technology, and realizes trajectory and status assessment of traffic targets, achieving dynamic and three-dimensional road traffic status perception, providing a technical basis for abnormal traffic status warning, and reducing the incidence of road accidents. BRIEF DESCRIPTION OF THE DRAWINGS

[0079] In order to more clearly illustrate the technical solution of the present invention, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0080] Figure 1 Schematic diagram of a method flow in an embodiment of the present invention;

[0081] Figure 2 A method block diagram of an embodiment of the present invention. DETAILED DESCRIPTION

[0082] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0083] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0084] Example 1

[0085] In this embodiment, if Figure 1 、 Figure 2 As shown, a method for dynamic perception of spatiotemporal traffic status based on air-ground collaboration includes the following steps:

[0086] S1. Obtain road traffic video data from different height perspectives and construct an airborne traffic information dataset.

[0087] The method for constructing an air-based traffic dataset includes: using a drone to collect air-based traffic video data; performing data cleaning on the air-based traffic video data to obtain cleaned video data; intercepting the cleaned video data frame by frame to produce an image dataset; and manually annotating the image dataset to obtain an air-based traffic dataset.

[0088] In this embodiment, an air-based traffic information dataset is constructed through air-based information. The specific implementation plan is to use a drone to cruise at a fixed orbit and altitude at a fixed time to collect traffic video streams on designated roads as the main data body. The data should include common traffic behavior subjects such as vehicles and pedestrians. The detailed process is as follows: The first step is to collect more than 500 video clips with a duration of 15s to 25s containing targets such as vehicles and pedestrians through the video data collected by the drone. The second step is to capture the collected video clips frame by frame to create a dataset containing 18,000 images. The resolution of the collected original image samples is 4096×2160. The Labelmg image labeling tool is then used to manually annotate the dataset. After labelmg has labeled each sample and saved it, it will generate a text in XML format. The generated XML files will be used in the next step of model training. The labeling method is as follows: First, click the Generate Label Box button in the menu bar on the right side of the page. After selecting the area where the vehicle is located in the video frame, a label selection box will pop up. Enter the corresponding label name and click OK. After completing the labeling of all targets in the sample, click Save to complete the labeling of an image sample.

[0089] S2. Build a deep learning neural network and train it based on the airborne and ground-based traffic datasets to obtain a traffic target detection model.

[0090] In this embodiment, the ground-based traffic data is data obtained by the camera, the training environment is the Linux Ubuntu 20.04 operating system, the graphics card used is the GeForce GTX 3080, the division ratio of the training set and the test set is 7:3, the training parameter batch_size is 32, and the initial learning rate is set to 0.0001.

[0091] S3. Construct an air-ground traffic information fusion model and train it based on the air-based traffic dataset and the ground-based traffic dataset to obtain a dynamic perception model.

[0092] In this embodiment, when using a drone for air-based traffic status perception, the drone needs to hover directly above the detection area and shoot the road conditions vertically downward. The drone needs to be equipped with a communication module. Let the world coordinate of the drone be P U (X U , Y U , Z U ), its projection point on the road plane is (X U , Y U ) is the world coordinate origin of the detection area, then the detection area for:

[0093]

[0094] in, Indicates the estimated length of the drone camera's shooting area. Indicates the estimated width of the drone camera shooting area, n x Indicates the number of pixels in the X direction of the image captured by the camera, n y Indicates the number of pixels in the Y direction of the image captured by the camera, λ x Indicates the actual physical size of the pixel in the X direction, λ y Indicates the actual physical size of the pixel in the Y direction, r indicates the scaling ratio, F indicates the focal length of the camera, and Z G Indicates the road surface height in the detection area;

[0095] According to the detection area, the ground coordinates P of all key road points in the area can be obtained from the ground information system. W :

[0096]

[0097] Among them, W represents the world coordinate system, d represents the number of key points, Indicates the X-direction coordinate of the d-th key point in the world coordinate system, Represents the Y-direction coordinate of the d-th key point in the world coordinate system. On the other hand, based on the image data of the drone camera, the detection algorithm can be used to obtain the air-based pixel coordinate P C :

[0098]

[0099] Where C represents the pixel coordinate system, Represents the X-direction coordinate of the d-th key point in the pixel coordinate system, Indicates the Y coordinate of the dth key point in the pixel coordinate system. W and space-based information P C , you can apply key points and the vertical projection point of the drone (XU , Y U ) is matched with the relative position vector ΔP of the drone. Affected by the self-motion of the drone, the air-based information P C With foundation information P W There is usually a certain deviation E, namely:

[0100]

[0101] The three-dimensional traffic perception ground-air information fusion model can achieve the optimal fusion effect by minimizing the ground-air information deviation E.

[0102] S4. Acquire real-time air-based traffic data and real-time ground-based traffic data, and use a traffic target detection model to obtain traffic target trajectory information.

[0103] The method for obtaining traffic target trajectory information includes: using a drone to obtain the original vehicle trajectory; using wavelet analysis to identify outliers in the original vehicle trajectory and correct it; using a wavelet filter to reduce noise on the corrected trajectory to obtain traffic target trajectory information.

[0104] In this embodiment, the original trajectory of the vehicle is obtained by using a drone, and the original trajectory of the vehicle obtained by the drone state perception algorithm is preprocessed: first, wavelet analysis is used to identify outliers and correct them, and then a wavelet filter is used to reduce noise to obtain vehicle trajectory information.

[0105]

[0106] in, is the estimated trajectory of vehicle i at time t, obtained through a deep learning convolutional neural network model; and is the position of the vehicle on the x-axis and y-axis.

[0107] S5. Use the dynamic perception model to obtain lane-level traffic parameters and vehicle-level traffic parameters based on traffic target trajectory information.

[0108] The road section monitored by drones typically ranges from 100 to 500 meters, providing a spatial basis for obtaining lane-level traffic state parameters. These can be inferred from vehicle-level microscopic traffic parameters within a time interval ΔT. In practical traffic flow management, ΔT typically ranges from 10 to 120 seconds.

[0109] Lane-level traffic parameters include: lane time-average speed, lane spatial-average speed, lane traffic volume, and lane traffic density. Lane time-average speed is calculated by the average longitudinal speed of vehicles traveling on the lane at a certain time interval. Lane spatial-average speed is:

[0110]

[0111] in, represents the lane space average speed, represents the lane time average speed, represents the variance of the lane’s time-averaged speed, r represents the rth lane; the lane traffic flow is:

[0112]

[0113] Among them, q r represents the lane traffic volume, m r represents the number of vehicles passing a certain section on the lane, ΔT represents the time interval; the lane traffic density is:

[0114]

[0115] Among them, k r (t) represents the lane traffic density, represents the average number of vehicles passing through a certain section, l r Indicates the length of the local lane;

[0116] Vehicle-level traffic parameters include: vehicle instantaneous speed, headway, and headway time; vehicle instantaneous speed is:

[0117]

[0118] in, represents the instantaneous speed of the vehicle, represents the traffic target trajectory information, τ represents the vehicle-level parameter calculation time interval, i represents the i-th vehicle; the headway is:

[0119]

[0120] in, Indicates the headway between vehicles. Represents the longitudinal position coordinates of traffic trajectory information; the headway is:

[0121]

[0122] in, Indicates the headway time, Represents the longitudinal component of the vehicle's instantaneous velocity.

[0123] Example 2

[0124] In this embodiment, a spatiotemporal traffic status dynamic perception system based on air-ground collaboration includes: a data set construction module, a first model construction module, a second model construction module, a trajectory acquisition module and a parameter perception module.

[0125] The dataset construction module is used to obtain road traffic video data from different height perspectives and construct an air-based traffic information dataset.

[0126] The dataset construction module includes: a video data acquisition unit, a data processing unit, an image data acquisition unit and an annotation unit; the video data acquisition unit uses a drone to collect air-based traffic video data; the data processing unit is used to clean the air-based traffic video data to obtain cleaned video data; the image data acquisition unit is used to capture the cleaned video data frame by frame to produce an image dataset; the annotation unit is used to manually annotate the image dataset to obtain an air-based traffic dataset.

[0127] The first model building module is used to build a deep learning neural network, train the deep learning neural network based on the air-based traffic dataset and the ground-based traffic dataset, and obtain a traffic target detection model.

[0128] The second model construction module is used to construct an air-ground traffic information fusion model, and train the air-ground traffic information fusion model based on the air-based traffic dataset and the ground-based traffic dataset to obtain a dynamic perception model.

[0129] The trajectory acquisition module is used to obtain real-time air-based traffic data and real-time ground-based traffic data, and obtain traffic target trajectory information using the traffic target detection model.

[0130] The trajectory acquisition module includes: an original trajectory acquisition unit, a correction unit and a noise reduction unit; the original trajectory acquisition unit uses a drone to obtain the vehicle's original trajectory; the correction unit uses wavelet analysis to identify outliers in the vehicle's original trajectory and correct it; the noise reduction unit uses a wavelet filter to reduce the noise of the corrected trajectory to obtain traffic target trajectory information.

[0131] The parameter perception module uses a dynamic perception model to obtain lane-level traffic parameters and vehicle-level traffic parameters based on traffic target trajectory information.

[0132] Lane-level traffic parameters include: lane time-average speed, lane spatial-average speed, lane traffic volume, and lane traffic density. Lane time-average speed is calculated by the average longitudinal speed of vehicles traveling on the lane at a certain time interval. Lane spatial-average speed is:

[0133]

[0134] in, represents the lane space average speed, represents the lane time average speed, represents the variance of the lane’s time-averaged speed, r represents the rth lane; the lane traffic flow is:

[0135]

[0136] Among them, q r represents the lane traffic volume, m r represents the number of vehicles passing a certain section on the lane, ΔT represents the time interval; the lane traffic density is:

[0137]

[0138] Among them, k r (t) represents the lane traffic density, represents the average number of vehicles passing through a certain section, l r Indicates the length of the local lane;

[0139] Vehicle-level traffic parameters include: vehicle instantaneous speed, headway, and headway time; vehicle instantaneous speed is:

[0140]

[0141] in, represents the instantaneous speed of the vehicle, represents the traffic target trajectory information, τ represents the vehicle-level parameter calculation time interval, i represents the i-th vehicle; the headway is:

[0142]

[0143] in, Indicates the headway between vehicles. Represents the longitudinal position coordinates of traffic trajectory information; the headway is:

[0144]

[0145] in, Indicates the headway time, Represents the longitudinal component of the vehicle's instantaneous velocity.

[0146] The embodiments described above are merely descriptions of preferred embodiments of the present invention and are not intended to limit the scope of the present invention. Without departing from the spirit of the present invention, various modifications and improvements made to the technical solutions of the present invention by persons skilled in the art should fall within the scope of protection defined by the claims of the present invention.

Claims

1. A method for dynamic perception of spatiotemporal traffic status based on air-ground collaboration, characterized by: The following steps are involved: Acquire road traffic video data from different height perspectives and construct an airborne traffic information dataset; Constructing a deep learning neural network, and training the deep learning neural network based on the air-based traffic dataset and the ground-based traffic dataset to obtain a traffic target detection model; Constructing an air-ground traffic information fusion model, and training the air-ground traffic information fusion model based on the air-based traffic dataset and the ground-based traffic dataset to obtain a dynamic perception model; Acquire real-time air-based traffic data and real-time ground-based traffic data, and obtain traffic target trajectory information using the traffic target detection model; The dynamic perception model is used to obtain lane-level traffic parameters and vehicle-level traffic parameters based on the traffic target trajectory information.

2. The method for dynamic perception of spatiotemporal traffic status based on air-ground collaboration according to claim 1 is characterized in that: The method for constructing the air-based traffic dataset includes: Using drones to collect air-based traffic video data; performing data cleaning on the air-based traffic video data to obtain cleaned video data; Intercepting the cleaned video data frame by frame to create an image data set; Manually annotate the image dataset to obtain the air-based traffic dataset.

3. The method for dynamic perception of spatiotemporal traffic status based on air-ground collaboration according to claim 1 is characterized in that: The method for obtaining the traffic target trajectory information includes: Use drones to obtain the original trajectory of the vehicle; Using wavelet analysis to identify outliers in the original trajectory of the vehicle and perform corrections; The corrected trajectory is subjected to noise reduction using a wavelet filter to obtain the traffic target trajectory information.

4. The method for dynamic perception of spatiotemporal traffic status based on air-ground collaboration according to claim 1 is characterized in that: The lane-level traffic parameters include: lane time average speed, lane space average speed, lane traffic flow and lane traffic density; The lane time average speed is obtained by the average longitudinal speed of vehicles traveling on the lane during the time interval; The lane space average speed is: in, represents the lane space average speed, represents the lane time average speed, represents the variance of the lane’s time-averaged speed, and r represents the rth lane; The traffic volume of the lane is: Among them, q r represents the lane traffic volume, m r It represents the number of vehicles passing a certain section on the lane, and ΔT represents the time interval; The lane traffic density is: Among them, k r (t) represents the lane traffic density, represents the average number of vehicles passing through a certain section, l r Indicates the length of the local lane; The vehicle-level traffic parameters include: vehicle instantaneous speed, headway and headway time; The instantaneous speed of the vehicle is: in, represents the instantaneous speed of the vehicle, represents the traffic target trajectory information, τ represents the vehicle-level parameter calculation time interval, and i represents the i-th vehicle; The headway is: in, Indicates the headway between vehicles. The longitudinal position coordinates representing the traffic trajectory information; The headway time is: in, Indicates the headway time, Represents the longitudinal component of the vehicle's instantaneous velocity.

5. A spatial-temporal dynamic traffic state perception system based on air-ground collaboration, wherein the perception system applies the perception method according to any one of claims 1 to 4, characterized in that: include: A data set construction module, a first model construction module, a second model construction module, a trajectory acquisition module, and a parameter perception module; The data set construction module is used to obtain road traffic video data at different heights and construct an air-based traffic information data set; The first model building module is used to build a deep learning neural network, and train the deep learning neural network based on the air-based traffic dataset and the ground-based traffic dataset to obtain a traffic target detection model; The second model building module is used to build an air-ground traffic information fusion model, and train the air-ground traffic information fusion model based on the air-based traffic dataset and the ground-based traffic dataset to obtain a dynamic perception model; The trajectory acquisition module is used to acquire real-time air-based traffic data and real-time ground-based traffic data, and to acquire traffic target trajectory information using the traffic target detection model; The parameter perception module utilizes the dynamic perception model to obtain lane-level traffic parameters and vehicle-level traffic parameters based on the traffic target trajectory information.

6. The space-time three-dimensional traffic status dynamic perception system based on air-ground collaboration according to claim 5 is characterized by: The data set construction module includes: a video data acquisition unit, a data processing unit, an image data acquisition unit and a labeling unit; The video data acquisition unit collects air-based traffic video data using a drone; The data processing unit is used to clean the air-based traffic video data to obtain cleaned video data; The image data acquisition unit is used to intercept the cleaned video data frame by frame to produce an image data set; The labeling unit is used to manually label the image dataset to obtain the air-based traffic dataset.

7. The space-time three-dimensional traffic status dynamic perception system based on air-ground collaboration according to claim 5 is characterized by: The trajectory acquisition module includes: an original trajectory acquisition unit, a correction unit and a noise reduction unit; The original trajectory acquisition unit acquires the original trajectory of the vehicle using a drone; The correction unit uses wavelet analysis to identify abnormal values ​​of the original trajectory of the vehicle and perform corrections; The noise reduction unit uses a wavelet filter to reduce noise on the corrected trajectory to obtain the traffic target trajectory information.

8. The space-time three-dimensional traffic status dynamic perception system based on air-ground collaboration according to claim 5 is characterized by: The lane-level traffic parameters include: lane time average speed, lane space average speed, lane traffic flow and lane traffic density; The lane time average speed is obtained by the average longitudinal speed of vehicles traveling on the lane during the time interval; The lane space average speed is: in, represents the lane space average speed, represents the lane time average speed, represents the variance of the lane’s time-averaged speed, and r represents the rth lane; The traffic volume of the lane is: Among them, q r represents the lane traffic volume, m r It represents the number of vehicles passing a certain section on the lane, and ΔT represents the time interval; The lane traffic density is: Among them, k r (t) represents the lane traffic density, represents the average number of vehicles passing through a certain section, l r Indicates the length of the local lane; The vehicle-level traffic parameters include: vehicle instantaneous speed, headway and headway time; The instantaneous speed of the vehicle is: in, represents the instantaneous speed of the vehicle, represents the traffic target trajectory information, τ represents the vehicle-level parameter calculation time interval, and i represents the i-th vehicle; The headway is: in, Indicates the headway between vehicles. Represents the longitudinal position coordinate of the traffic trajectory information; the headway time is: in, Indicates the headway time, Represents the longitudinal component of the vehicle's instantaneous velocity.

Images