WO2023201904A1 - 车辆异常行驶检测方法、电子设备、存储介质 - Google Patents

车辆异常行驶检测方法、电子设备、存储介质 Download PDF

Info

Publication number
WO2023201904A1
WO2023201904A1 PCT/CN2022/104661 CN2022104661W WO2023201904A1 WO 2023201904 A1 WO2023201904 A1 WO 2023201904A1 CN 2022104661 W CN2022104661 W CN 2022104661W WO 2023201904 A1 WO2023201904 A1 WO 2023201904A1
Authority
WO
WIPO (PCT)
Prior art keywords
target vehicle
driving
road
duration
vehicle
Prior art date
Application number
PCT/CN2022/104661
Other languages
English (en)
French (fr)
Inventor
刘成沛
孙全俊
薛艺淳
王前选
Original Assignee
五邑大学
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 五邑大学 filed Critical 五邑大学
Publication of WO2023201904A1 publication Critical patent/WO2023201904A1/zh

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/24—Classification techniques
    • G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
    • G06F18/2415—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on parametric or probabilistic models, e.g. based on likelihood ratio or false acceptance rate versus a false rejection rate
    • G06F18/24155—Bayesian classification
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/23—Clustering techniques
    • G06F18/232—Non-hierarchical techniques
    • G06F18/2321—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions
    • G06F18/23213—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions with fixed number of clusters, e.g. K-means clustering
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/045—Combinations of networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G06N3/084—Backpropagation, e.g. using gradient descent
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/20—Analysis of motion
    • G06T7/246—Analysis of motion using feature-based methods, e.g. the tracking of corners or segments

Definitions

  • the invention relates to the field of road vehicle driving detection, and in particular to a vehicle abnormal driving detection method, electronic equipment, and storage media.
  • the intelligent traffic detection system determines whether the target vehicle is driving abnormally by analyzing the detected driving information of the target vehicle, and then sends the judgment results to the traffic management department. , so that the traffic management department can control vehicle driving based on the judgment results.
  • the current traffic detection system uses the Kalman filter algorithm for target tracking. However, in complex traffic environments, the prediction accuracy of the Kalman filter algorithm is low, which leads to poor tracking results and low abnormality judgment accuracy of the intelligent traffic detection system.
  • the present invention aims to solve at least one of the technical problems existing in the prior art. To this end, the present invention proposes a vehicle abnormal driving detection method, electronic device, and storage medium, which are beneficial to improving the detection accuracy of vehicle abnormal driving detection.
  • a first embodiment of the present invention provides a method for detecting abnormal driving of a vehicle.
  • the method includes:
  • the target vehicle to be tracked is tracked through an improved Kalman filter algorithm to obtain the corresponding driving trajectory, wherein the covariance matrix of the improved Kalman filter algorithm is optimized through a preset BP neural network model;
  • the driving trajectory it is determined whether the target vehicle is driving abnormally.
  • the embodiment of the first aspect of the present invention provides a vehicle abnormal driving detection method, which has at least the following beneficial effects: the covariance matrix of the Kalman filter algorithm represents the impact of noise on the target vehicle position estimate.
  • BP neural network is used After the network algorithm converts the nonlinear non-Gaussian noise in the covariance matrix into a linear state, it updates the covariance matrix. Compared with before the improvement, the updated covariance matrix takes into account the influence of nonlinear non-Gaussian noise, so it can Obtain a more accurate position estimate of the target vehicle, thereby obtaining the vehicle's driving trajectory more quickly and accurately, which will help improve the detection accuracy of abnormal vehicle driving detection.
  • determining the target vehicle to be tracked based on the road segment video includes:
  • the road driving background image is identified through a road area target vehicle detection model to determine the target vehicle to be tracked.
  • the method before identifying the road driving background image through the road area target vehicle detection model and determining the target vehicle to be tracked, the method further includes:
  • the target vehicle detection model is a convolutional neural network model.
  • determining whether the target vehicle is driving abnormally based on the driving trajectory includes:
  • the method before calculating the duration for which the target vehicle maintains the same driving trajectory, the method includes:
  • the storage trigger condition is that the detected target vehicle has speed and/or direction.
  • calculating the duration for which the target vehicle's driving trajectory continues to be abnormal includes:
  • determining that the target vehicle is driving abnormally includes:
  • calculating the duration for which the target vehicle's driving trajectory continues to be abnormal includes:
  • determining that the target vehicle is driving abnormally includes:
  • the target vehicle is drunk driving or drug driving.
  • calculating the duration for which the target vehicle's driving trajectory continues to be abnormal includes:
  • determining that the target vehicle is driving abnormally includes:
  • the target vehicle is determined to be overspeeding.
  • a second embodiment of the present invention provides an electronic device, including:
  • the program is stored in the memory, and the processor executes at least one of the programs to implement the vehicle abnormal driving detection method according to any embodiment of the first aspect of the present invention.
  • a third embodiment of the present invention provides a computer-readable storage medium.
  • the computer-readable storage medium stores computer-executable signals.
  • the computer-executable signals are used to execute any implementation of the first aspect of the present invention.
  • Figure 1 is a flow chart of a vehicle abnormal driving detection method provided by some embodiments of the present application.
  • Figure 2 is a flow chart of a vehicle abnormal driving detection method provided by other embodiments of the present application.
  • the Kalman filter algorithm can only remove linear Gaussian noise. However, when using the Kalman filter algorithm for target tracking, it is affected by many environmental noises, such as: nonlinear noise generated when multiple target vehicles are running on the road at the same time, and different Nonlinear noise generated by weather conditions will affect the tracking accuracy of the Kalman filter algorithm, which in turn affects the detection accuracy of abnormal vehicle driving detection. Based on this, this application proposes a vehicle abnormal driving detection method, electronic device, and storage medium, which are beneficial to improving the detection accuracy of vehicle abnormal driving detection.
  • BP neural network is a back propagation algorithm. It uses the results of the output layer to infer the weights of the connections between different network layers. It does not need to pre-calculate the weights between layers. The best value is derived through the data of the output layer. Optimal weight values, thereby helping the Kalman filter algorithm iterate out a suitable covariance matrix.
  • a first embodiment of the present invention provides a vehicle abnormal driving detection method, including:
  • Step 110 Obtain the road segment video, and determine the target vehicle to be tracked based on the road segment video.
  • Step 120 Track the target vehicle to be tracked through the improved Kalman filter algorithm to obtain the corresponding driving trajectory, in which the covariance matrix of the improved Kalman filter algorithm is optimized through the preset BP neural network model.
  • Step 130 Based on the driving trajectory, determine whether the target vehicle is driving abnormally.
  • the covariance matrix of the Kalman filter algorithm represents the impact of noise on the target vehicle position estimate.
  • the BP neural network algorithm is used to convert the nonlinear non-Gaussian noise in the covariance matrix into a linear state. After that, the covariance matrix is updated. Compared with before the improvement, the updated covariance matrix takes into account the influence of nonlinear non-Gaussian noise, so it can obtain a more accurate position estimate of the target vehicle, and then obtain it more quickly and accurately. Vehicle driving trajectory, which in turn helps improve the detection accuracy of abnormal vehicle driving detection.
  • step 110 is to obtain the road segment video and determine the target vehicle to be tracked based on the road segment video, including: classifying the noise information in the road segment video through a Bayesian algorithm model; and removing the road segment video through a Gaussian model.
  • Noise is obtained to obtain the road driving background image; the road driving background image is identified through the road area target vehicle detection model to determine the target vehicle to be tracked.
  • the Bayesian algorithm can be used to classify different types of noise.
  • the classified noise is beneficial to the Gaussian model to provide appropriate construction.
  • Model weights, and then convolution operations can be performed based on appropriate modeling weights to achieve better denoising effects.
  • the road driving background image when constructing the road driving background image, it also includes marking the road lane lines in the road segment video and specifying the traveling direction of the road.
  • the method before identifying the road driving background image through the road area target vehicle detection model and determining the target vehicle to be tracked, the method also includes: using a training data set with a vehicle identification frame on the road to identify the target vehicle detection model. Carry out training; verify whether the training results meet the preset standards by shooting road driving videos. If not, continue training until the preset standards are met; where the target vehicle detection model is a convolutional neural network model.
  • the target vehicle detection model is a convolutional neural network model. Specifically, it can be a YOLOV5 model, which includes 20 convolutional layers and 1 fully connected layer. Each convolutional layer includes A set of normalized networks and 1 Sigmod activation network, each pooling layer is responsible for maximizing the convolutional layer, and the fully connected layer is responsible for outputting the detected target vehicle.
  • the weight parameters of each layer of the established neural network model are updated through the training data set with vehicle recognition frames on the road, and the appropriate connection weights between layers are found to improve the accuracy of detecting the target vehicle. Liang Mindu.
  • road driving videos are taken to verify whether the training results meet the detection standards. For example, if the detection accuracy reaches 98%, the trained neural network can meet the requirements.
  • step 130 determines whether the target vehicle is driving abnormally, including: calculating the duration for which the target vehicle's driving trajectory continues to be abnormal; when the duration exceeds the triggering duration of the preset timer, determining whether the target vehicle is driving abnormally abnormal.
  • the target vehicle before calculating the duration for which the target vehicle maintains the same driving trajectory, it includes: storing the driving trajectory of the target vehicle that satisfies the preset storage trigger condition in a space-time sequence in frame units.
  • the storage trigger condition is that the detected target vehicle has speed information, that is, the target vehicle has speed and direction in the frame.
  • speed information that is, the target vehicle has speed and direction in the frame.
  • Spatial information saves resources, and on the other hand, reduces the impact of noise factors, which is beneficial to improving the detection accuracy of abnormal vehicle driving detection.
  • multiple timers can be set on the spatio-temporal sequence, such as: speed timer, blocked lane line timer, etc., and the triggers corresponding to different timers
  • the duration can be the same or different.
  • calculating the duration that the target vehicle's driving trajectory continues to be abnormal includes: calculating the number of frames in which the target vehicle continues to block the lane line to obtain the first duration; correspondingly, determining that the target vehicle is driving abnormally includes: determining that the target vehicle is driving abnormally Occupying multiple lanes for abnormal lane changes or violations.
  • the target vehicle when the target vehicle continues to block the lane line for longer than the triggering time of the blocked lane line timer, the target vehicle is determined to have abnormally changed lanes or illegally occupied multiple lanes.
  • calculating the duration during which the target vehicle's driving trajectory continues to be abnormal includes: calculating the number of frames in which the target vehicle's trajectory is different from the trajectories of other driving vehicles to obtain the second duration; correspondingly, determining that the target vehicle's driving trajectory is abnormal, Including: determining whether the target vehicle is drunk or drug-driving.
  • the trajectory of the target vehicle when the trajectory of the target vehicle is different from the trajectories of other driving vehicles for longer than the triggering duration of the timer, it is determined that the target vehicle is driving under the influence of alcohol or drugs. Specifically, the trajectory of the target vehicle is different from the trajectories of other driving vehicles. For example, the trajectory of the target vehicle is curved, while the trajectories of other vehicles are straight.
  • calculating the duration that the target vehicle's driving trajectory continues to be abnormal includes: calculating the number of frames in which the target vehicle's driving speed continues to exceed the speed extreme value, and obtaining the third duration; correspondingly, determining that the target vehicle's driving trajectory is abnormal includes: : Based on the K-Mean algorithm, cluster the driving speeds of corresponding consecutive multi-frame images within the third duration to obtain the average speed; when the average speed exceeds the extreme speed value, the target vehicle is determined to be speeding.
  • the target vehicle when the average speed of the target vehicle in consecutive frames exceeds the speed extreme value, the target vehicle is determined to be speeding.
  • the clustering process of driving speeds in consecutive multiple frames of images based on the K-Mean algorithm is: obtaining the driving speed of the target vehicle in each frame of images; among these driving speeds , randomly select K objects as the initial clustering centers, then calculate the distance between the remaining objects and each clustering center, and then assign each remaining object to the clustering center closest to it, and assign the clustering center to them objects form clusters.
  • the termination condition can be that no (or a minimum number) objects are reassigned to different clusters, and no (or a minimum number) cluster centers change. At this time, the sum of squared errors is locally minimized. Therefore, the obtained speed mean can be more Accurately reflects the operating status of the target vehicle, which is helpful to improve the accuracy of abnormal vehicle driving detection.
  • the average speed required is directional and is a vector. In some embodiments, the speed along the direction of travel of the road is specified as positive, and vice versa as negative.
  • calculating the duration during which the target vehicle's driving trajectory continues to be abnormal also includes: calculating the number of frames in which the target vehicle's driving speed continues to be negative to obtain the fourth duration; correspondingly, judging The target vehicle is driving abnormally, including: it is determined that the target vehicle is traveling in the wrong direction.
  • the process of obtaining the driving speed of the target vehicle in each frame of image is: obtaining the positional relationship between the image coordinate system and the world coordinate system according to the position information of the camera;
  • the detection model obtains the image coordinates of the target vehicle in each frame of the image in the area to be detected; according to the positional relationship between the image coordinate system and the world coordinate system, the image coordinates in each frame of the image are converted into corresponding world coordinates; according to The world coordinate uses the speed-time displacement formula to calculate the driving speed in each frame of image in the area to be detected.
  • the image coordinate system is a two-dimensional coordinate system and the world coordinate system is a three-dimensional coordinate system.
  • the two-dimensional coordinates are obtained from the video, but the speed of the target vehicle needs to be calculated based on the three-dimensional coordinates. Therefore, the obtained image needs to be
  • the coordinate system is converted into world coordinates.
  • the positional relationship between the image coordinate system and the world coordinate system is obtained according to the deflection angle and pitch angle of the camera. Specifically, the following example is used to illustrate the mapping relationship formula between the image coordinate system and the world coordinate system.
  • ⁇ is the pitch angle of the camera
  • ⁇ is the deflection angle of the camera
  • a 1 is the image coordinate
  • B 2 is the world coordinate
  • Q is the translation vector of the camera on the Z axis
  • H 1 is the rotation matrix
  • the rotation matrix H 1 is obtained by rotating the Z axis of the world coordinate system by an angle ⁇ and then rotating it around the X axis by an angle ⁇ .
  • T is the internal parameter of the camera.
  • S220 Use the improved Kalman filter algorithm to track the target vehicle and obtain the driving trajectory of the target vehicle in each frame of image.
  • S240 Based on the average speed, driving trajectory, pre-marked lane lines, pre-set driving direction and other information, combined with the abnormal driving logic judgment algorithm, determine whether the target vehicle is driving abnormally.
  • the second embodiment of the present application provides an electronic device, including:
  • the program is stored in the memory, and the processor executes at least one program to implement the vehicle abnormal driving detection method according to any embodiment of the first aspect of the present application.
  • the processor and memory may be connected via a bus or other means.
  • memory can be used to store non-transitory software instructions and non-transitory instructions.
  • the memory may include high-speed random access memory and may also include non-transitory memory, such as at least one magnetic disk storage device, flash memory device, or other non-transitory solid-state storage device.
  • the memory may optionally include memories remotely located relative to the processor, and these remote memories may be connected to the processor through a network. Examples of the above-mentioned networks include but are not limited to the Internet, intranets, local area networks, mobile communication networks, and its combination.
  • the processor executes various functional applications and data processing by running non-transient software instructions, instructions and signals stored in the memory, that is, the vehicle abnormal driving detection method of the first embodiment.
  • the non-transient software instructions and instructions required to implement the abnormal vehicle driving detection method of the embodiment of the first aspect are stored in the memory.
  • the above-mentioned vehicle abnormal driving detection method is executed.
  • the above-described FIG. 1 is executed.
  • the method steps S110 to S130 in FIG. 2 are executed, and the method steps S210 to S240 in FIG. 2 described above are performed.
  • the electronic device of the second aspect executes the vehicle abnormal driving detection method of the embodiment of the first aspect of the application, it has all the beneficial effects of the embodiment of the first aspect of the application.
  • the third embodiment of the present application provides a computer-readable storage medium.
  • the computer-readable storage medium stores computer-executable signals.
  • the computer-executable signals are used to perform vehicle anomalies according to any embodiment of the first aspect of the present application. Driving detection method.
  • the computer storage medium of the third aspect can execute the vehicle abnormal driving detection method of the embodiment of the first aspect of the present application, it has all the beneficial effects of the first aspect of the present application.
  • the device embodiments described above are only illustrative.
  • the units described as separate components may or may not be physically separated.
  • the components shown as units may or may not be physical units, that is, they may be located in one place. , or it can be distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.
  • Computer storage media includes, but is not limited to, RAM, ROM, EEPROM, flash memory or other memory technology, CD-ROM, Digital Versatile Disk (DVD) or other optical disk storage, magnetic cassettes, tapes, disk storage or other magnetic storage devices, or may Any other medium used to store the desired information and that can be accessed by a computer.
  • communication media typically embodies a computer-readable signal, data structure, instruction module, or other data in a modulated data signal such as a carrier wave or other transport mechanism, and may include any information delivery media .

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • General Physics & Mathematics (AREA)
  • Evolutionary Computation (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Artificial Intelligence (AREA)
  • General Engineering & Computer Science (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Computational Linguistics (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Probability & Statistics with Applications (AREA)
  • Health & Medical Sciences (AREA)
  • Biomedical Technology (AREA)
  • Biophysics (AREA)
  • Evolutionary Biology (AREA)
  • General Health & Medical Sciences (AREA)
  • Molecular Biology (AREA)
  • Computing Systems (AREA)
  • Mathematical Physics (AREA)
  • Software Systems (AREA)
  • Multimedia (AREA)
  • Traffic Control Systems (AREA)

Abstract

本发明公开了一种车辆异常行驶检测方法、电子设备、存储介质，涉及道路车辆行驶检测领域，所述方法包括获取路段视频，并根据所述路段视频，确定待跟踪的目标车辆；通过改进的卡尔曼滤波算法对所述待跟踪的目标车辆进行跟踪，得到对应的行驶轨迹，其中，所述改进的卡尔曼滤波算法的协方差矩阵通过预设的BP神经网络模型进行优化；根据所述行驶轨迹，判断所述目标车辆是否行驶异常。本发明有利于提高车辆异常行驶检测的精度。

Description

车辆异常行驶检测方法、电子设备、存储介质 技术领域
本发明涉及道路车辆行驶检测领域，尤其是涉及一种车辆异常行驶检测方法、电子设备、存储介质。
背景技术
现在的交通管理部门大多通过智能交通检测系统进行车辆异常行驶行为的检测，智能交通检测系统通过分析检测得到的目标车辆的行驶信息来判断目标车辆行驶是否异常，然后将判断结果发送给交通管理部门，以使交通管理部门根据判断结果对车辆行驶进行管控。目前的交通检测系统使用卡尔曼滤波算法进行目标跟踪，但是在复杂交通环境中，卡尔曼滤波算法预测精度低，进而导致跟踪效果差，使得智能交通检测系统的异常判断精度偏低。
发明内容
本发明旨在至少解决现有技术中存在的技术问题之一。为此，本发明提出一种车辆异常行驶检测方法、电子设备、存储介质，有利于提高车辆异常行驶检测的检测精度。
本发明第一方面实施例提供了一种车辆异常行驶检测方法，所述方法包括：
获取路段视频，并根据所述路段视频，确定待跟踪的目标车辆；
通过改进的卡尔曼滤波算法对所述待跟踪的目标车辆进行跟踪，得到对应的行驶轨迹，其中，所述改进的卡尔曼滤波算法的协方差矩阵通过预设的BP神经网络模型进行优化；
根据所述行驶轨迹，判断所述目标车辆是否行驶异常。
本发明第一方面实施例提供了一种车辆异常行驶检测方法，至少具有如下有益效果：卡尔曼滤波算法的协方差矩阵表征噪声对目标车辆位置估计值的影响，在本申请中，利用BP神经网络算法将协方差矩阵中的非线性非高斯的噪声转化为线性状态后，更新协方差矩阵，相较于未改进前，更新后的协方差矩阵考虑到了非线性非高斯噪声的影响，因此能够获取更准确地目标车辆的位置估计值，进而更快速、准确的获取车辆行驶轨迹，进而有利于提高车辆异常行驶检测的检测精度。
根据本发明第一方面的一些实施例，所述根据所述路段视频，确定待跟踪的目标车辆，包括：
通过贝叶斯算法模型对所述路段视频中的噪声信息进行分类；
通过高斯模型对所述路段视频进行除噪，得到道路行驶背景图像；
通过道路区域目标车辆检测模型对所述道路行驶背景图像进行识别，确定待跟踪的目标车辆。
根据本发明第一方面的一些实施例，在所述通过道路区域目标车辆检测模型对所述道路行驶背景图像进行识别，确定待跟踪的目标车辆之前，所述方法还包括：
通过道路上带有车辆识别框的训练数据集对所述目标车辆检测模型进行训练；
通过拍摄道路行车视频来验证训练结果是否满足预设标准，如不满足，继续训练，直至满足所述预设标准；
其中，所述目标车辆检测模型为卷积神经网络模型。
根据本发明第一方面的一些实施例，所述根据所述行驶轨迹，判断所述目标车辆是否行驶异常，包括：
计算所述目标车辆的行驶轨迹持续异常的持续时长；
当所述持续时长超过预设的定时器的触发时长，判断所述目标车辆行驶异常。
根据本发明第一方面的一些实施例，在计算所述目标车辆维持相同行驶轨迹的持续时长之前，包括：
以帧为单位将满足预设的存储触发条件的所述目标车辆的行驶轨迹存储至时空序列；
其中，所述存储触发条件为检测到的目标车辆具有速度大小和/或方向。
根据本发明第一方面的一些实施例，所述计算所述目标车辆的行驶轨迹持续异常的持续时长，包括：
计算所述目标车辆持续遮挡车道线的帧数，得到第一持续时长；
对应地，所述判断所述目标车辆行驶异常，包括：
判定所述目标车辆为异常变道或者违规占用多车道。
根据本发明第一方面的一些实施例，所述计算所述目标车辆的行驶轨迹持续异常的持续时长，包括：
计算所述目标车辆的轨迹和其他行驶车辆的轨迹不同的帧数，得到第二持续时长；
对应地，所述判断所述目标车辆行驶异常，包括：
判定所述目标车辆为酒驾或者毒驾。
根据本发明第一方面的一些实施例，所述计算所述目标车辆的行驶轨迹持续异常的持续时长，包括：
计算所述目标车辆的行驶速度持续超过速度极值的帧数，得到第三持续时长；
对应地，所述判断所述目标车辆行驶异常，包括：
基于K-Mean算法，对所述第三持续时长内对应的连续多帧图像的行驶速度进行聚类处理，得到速度均值；
当所述速度均值超过所述速度极值时，判定所述目标车辆超速。
本发明第二方面实施例提供了一种电子设备，包括：
至少一个存储器；
至少一个处理器；
至少一个程序；
所述程序被存储在所述存储器中，所述处理器执行至少一个所述程序以实现如本发明第一方面任一项实施例所述的车辆异常行驶检测方法。
本发明第三方面实施例提供了一种计算机可读存储介质，所述计算机可读存储介质存储有计算机可执行信号，所述计算机可执行信号用于执行如本发明第一方面任一项实施例所述的车辆异常行驶检测方法。
附图说明
本申请的附加方面和优点结合下面附图对实施例的描述中将变得明显和容易理解，其中：
图1为本申请一些实施例提供的车辆异常行驶检测方法的流程图；
图2为本申请另一些实施例提供的车辆异常行驶检测方法的流程图。
具体实施方式
为了使本申请的目的、技术方案及优点更加清楚明白，以下结合附图及实施例，对本申请进行进一步详细说明。应当理解，此处所描述的具体实施例仅用以解释本申请，并不用于限定本申请。
需要说明的是，虽然在流程图中示出了逻辑顺序，但是在某些情况下，可以以不同于流程图中的顺序执行所示出或描述的步骤。说明书和权利要求书及上述附图中的术语等是用于区别类似的对象，而不必用于描述特定的顺序或先后次序。
在本申请的描述中，如果有描述到第一、第二只是用于区分技术特征为目的，而不 能理解为指示或暗示相对重要性或者隐含指明所指示的技术特征的数量或者隐含指明所指示的技术特征的先后关系。
本申请的描述中，除非另有明确的限定，设置、安装、连接等词语应做广义理解，所属技术领域技术人员可以结合技术方案的具体内容合理确定上述词语在本申请中的具体含义。
附图中所示的流程图仅是示例性说明，不是必须包括所有的内容和操作/步骤，也不是必须按所描述的顺序执行。例如，有的操作/步骤还可以分解，而有的操作/步骤可以合并或部分合并，因此实际执行的顺序有可能根据实际情况改变。
卡尔曼滤波算法仅能除去线性高斯噪声，然而，在使用卡尔曼滤波算法进行目标跟踪时，受到诸多环境噪声影响，比如：多个目标车辆同时在道路运行时产生的非线性噪声，同时不同的天气状况产生的非线性噪声，均会影响卡尔曼滤波算法的跟踪精度，进而影响车辆异常行驶检测的检测精度。基于此，本申请提出一种车辆异常行驶检测方法、电子设备、存储介质，有利于提高车辆异常行驶检测的检测精度。
下面是对本申请实施例中应用的术语的解释：
BP神经网络是一种反向传播算法，通过输出层的结果来推断不同网络层之间的连接的权值，不需要预先计算层与层之间的权值，通过输出层的数据来推导最优权值，从而帮助卡尔曼滤波算法迭代出合适的协方差矩阵。
下面结合附图，对本申请实施例作进一步阐述。
参照图1，本发明第一方面实施例提供了一种车辆异常行驶检测方法，包括：
步骤110、获取路段视频，并根据路段视频，确定待跟踪的目标车辆。
步骤120、通过改进的卡尔曼滤波算法对待跟踪的目标车辆进行跟踪，得到对应的行驶轨迹，其中，改进的卡尔曼滤波算法的协方差矩阵通过预设的BP神经网络模型进行优化。
步骤130、根据行驶轨迹，判断目标车辆是否行驶异常。
需要说明的是，卡尔曼滤波算法的协方差矩阵表征噪声对目标车辆位置估计值的影响，在本申请中，利用BP神经网络算法将协方差矩阵中的非线性非高斯的噪声转化为线性状态后，更新协方差矩阵，相较于未改进前，更新后的协方差矩阵考虑到了非线性非高斯噪声的影响，因此能够获取更准确地目标车辆的位置估计值，进而更快速、准确的获取车辆行驶轨迹，进而有利于提高车辆异常行驶检测的检测精度。
可以理解的是，步骤110、获取路段视频，并根据路段视频，确定待跟踪的目标车辆，包括：通过贝叶斯算法模型对路段视频中的噪声信息进行分类；通过高斯模型对路段视频进行除噪，得到道路行驶背景图像；通过道路区域目标车辆检测模型对道路行驶背景图像进行识别，确定待跟踪的目标车辆。
需要说明的是，贝叶斯算法可以用于对不同的噪声种类进行分类，在一些实施例中，利用高斯模型构建道路行驶背景图像时，分类后的噪声有利于高斯模型给出合适的的建模权值，进而可以根据合适的的建模权值进行卷积运算，以达到更好的除噪效果。
需要说明的是，在一些实施例中，在构建道路行驶背景图像时，还包括标记路段视频中的道路车道线、指定道路的行进方向。
可以理解的是，在通过道路区域目标车辆检测模型对道路行驶背景图像进行识别，确定待跟踪的目标车辆之前，方法还包括：通过道路上带有车辆识别框的训练数据集对目标车辆检测模型进行训练；通过拍摄道路行车视频来验证训练结果是否满足预设标准，如不满足，继续训练，直至满足预设标准；其中，目标车辆检测模型为卷积神经网络模型。
需要说明的是，在一些实施例中，目标车辆检测模型为卷积神经网络模型，具体的，可以为YOLOV5模型，其中包括20个卷积层和1个全连接层，每个卷积层包含一组归一化网络和1个Sigmod激活网络，每个池化层负责最大化卷积层，全连接层则负责输出检测到的目标车辆。在训练阶段，通过道路上带有车辆识别框的训练数据集对建立的神经网络模型每一层的权值参数更新，找到层与层之间合适的连接权值，以提高检测到目标车辆的梁敏度。在验证阶段，通过拍摄道路行车视频来验证训练结果是否符合检测标准，例如：若检测精度达到98％，则训练的神经网路可以符合要求。
可以理解的是，步骤130、根据行驶轨迹，判断目标车辆是否行驶异常，包括：计算目标车辆的行驶轨迹持续异常的持续时长；当持续时长超过预设的定时器的触发时长，判断目标车辆行驶异常。
可以理解的是，在计算目标车辆维持相同行驶轨迹的持续时长之前，包括：以帧为单位将满足预设的存储触发条件的目标车辆的行驶轨迹存储至时空序列。
需要说明的是，在一些实施例中，存储触发条件为检测到的目标车辆具有速度信息，即目标车辆在所在帧有速度大小和方向，针对未动的车辆，不需要存储行驶轨迹等时间。空间信息，一方面，节省资源，另一方面，减少噪声因素的影响，有利于提高车辆异常 行驶检测的检测精度。
需要说明的是，在一些实施例中，根据目标车辆维持的轨迹类型，可以在时空序列上设置多个计时器，如：速度计时器、遮挡车道线计时器等，不同计时器所对应的触发时长可以相同也可以不同。
可以理解的是，计算目标车辆的行驶轨迹持续异常的持续时长，包括：计算目标车辆持续遮挡车道线的帧数，得到第一持续时长；对应地，判断目标车辆行驶异常，包括：判定目标车辆为异常变道或者违规占用多车道。
需要说明的是，在一些实施例中，当目标车辆持续遮挡车道线的时长大于遮挡车道线计时器的触发时长时，则判定目标车辆为异常变道或者违规占用多车道。
可以理解的是，计算目标车辆的行驶轨迹持续异常的持续时长，包括：计算目标车辆的轨迹和其他行驶车辆的轨迹不同的帧数，得到第二持续时长；对应地，判断目标车辆行驶异常，包括：判定目标车辆为酒驾或者毒驾。
需要说明的是，在一些实施例中，当目标车辆的轨迹和其他行驶车辆的轨迹不同的时长大于计时器的触发时长时，则判定目标车辆为酒驾或者毒驾。具体地，目标车辆的轨迹和其他行驶车辆的轨迹不同表现例如：目标车辆的轨迹呈曲线，其他车辆的轨迹呈直线。
可以理解的是，计算目标车辆的行驶轨迹持续异常的持续时长，包括：计算目标车辆的行驶速度持续超过速度极值的帧数，得到第三持续时长；对应的，判断目标车辆行驶异常，包括：基于K-Mean算法，对第三持续时长内对应的连续多帧图像的行驶速度进行聚类处理，得到速度均值；当速度均值超过速度极值时，判定目标车辆超速。
需要说明的是，在一些实施例中，当目标车辆在连续帧内的速度均值超过速度极值时，则判定目标车辆为超速。
需要说明的是，在一些实施例中，基于K-Mean算法对连续多帧图像中的行驶速度进行聚类处理过程为：获取目标车辆在每一帧图像中的行驶速度；在这些行驶速度中，随机选取K个对象作为初始的聚类中心，然后计算剩余对象与各个聚类中心之间的距离，随后把每个剩余对象分配给距离它最近的聚类中心，聚类中心以及分配给它们的对象组成聚类。每分配一个对象，聚类的聚类中心会根据聚类中现有的对象被重新计算。这个过程将不断重复直到满足某个终止条件。终止条件可以是没有(或最小数目)对象被重新分配给不同的聚类，没有(或最小数目)聚类中心再发生变化，此时误差平方和 局部最小，因此，得到的速度均值，可以更精确的体现出目标车辆的运行状态，有利于提高车辆异常行驶检测的精度。
需要说明的是，所求的速度均值具有方向性，是矢量。在一些实施例中，规定沿着道路的行进方向的速度为正，反之为负。
需要说明的是，在一些实施例中，计算目标车辆的行驶轨迹持续异常的持续时长，还包括：计算目标车辆的行驶速度持续为负值的帧数，得到第四持续时长；对应的，判断目标车辆行驶异常，包括：判定目标车辆为逆行。
需要说明的是，在一些实施例中，获取目标车辆在每一帧图像中的行驶速度的过程为：根据相机的位置信息，获取图像坐标系和世界坐标系之间的位置关系；通过目标车辆检测模型获取目标车辆在待检测区域内每一帧图像中的图像坐标；根据图像坐标系和世界坐标系之间的位置关系，将每一帧图像中的图像坐标转换为对应的世界坐标；根据世界坐标利用速度时间位移公式计算待检测区域内每一帧图像中的行驶速度。
需要说明的是，图像坐标系为二维坐标系，世界坐标系为三维坐标系，从视频中获取的是二维坐标，但是需要根据三维坐标计算目标车辆的速度，因此，需要将获取的图像坐标系转化世界坐标。在一些实施例中，根据相机的偏转角与俯仰角获取图像坐标系和世界坐标系之间的位置关系，具体的，以下面例子进行说明，图像坐标系和世界坐标系之间的映射关系公式(1)所示，其中，φ为相机的俯仰角，θ为相机的偏转角，A 1为图像坐标，B 2为世界坐标，Q为相机在Z轴的平移向量，H 1为旋转矩阵，旋转矩阵H 1是世界坐标系Z轴通过旋转θ角度后再绕X轴旋转φ的角度得到的，T为相机的内部参数。
A 1＝B 2*Q*H 1*T     (1)
参照图2，以一个具体的实施例来描述整个车辆异常行驶检测方法的实施过程：
S210、从拍摄的路段视频中获得道路行驶背景图像，确定待跟踪的目标车辆，并对道路行驶背景进行车道线标注，确定道路的行车方向。
S220、使用改进的卡尔曼滤波算法对目标车辆进行跟踪，获得目标车辆在每一帧图像中的行驶轨迹。
S230、获取目标车辆在每一帧图像中的行驶速度，基于K-Mean算法，对连续多帧图像中的行驶速度进行聚类处理，得到目标车辆在连续多帧图像中的速度均值。
S240、根据速度均值、行驶轨迹、预先标注的车道线以及预先设置的行车方向等信息，结合异常行驶逻辑判断算法，判断目标车辆是否行驶异常。
本申请第二方面实施例提供了一种电子设备，包括：
至少一个存储器；
至少一个处理器；
至少一个程序；
程序被存储在存储器中，处理器执行至少一个程序以实现如本申请第一方面任一项实施例的车辆异常行驶检测方法。
处理器和存储器可以通过总线或者其他方式连接。
存储器作为一种非暂态可读存储介质，可用于存储非暂态软件指令以及非暂态性可指令。此外，存储器可以包括高速随机存取存储器，还可以包括非暂态存储器，例如至少一个磁盘存储器件、闪存器件、或其他非暂态固态存储器件。可以理解的是，存储器可选包括相对于处理器远程设置的存储器，这些远程存储器可以通过网络连接至该处理器，上述网络的实例包括但不限于互联网、企业内部网、局域网、移动通信网及其组合。
处理器通过运行存储在存储器中的非暂态软件指令、指令以及信号，从而执行各种功能应用以及数据处理，即第一方面实施例的车辆异常行驶检测方法。
实现第一方面实施例的车辆异常行驶检测方法所需的非暂态软件指令以及指令存储在存储器中，当被处理器执行时，执行上述车辆异常行驶检测方法，例如，执行以上描述的图1中的方法步骤S110至步骤S130，执行以上描述的图2中的方法步骤S210至步骤S240。
由于第二方面的电子设备执行本申请第一方面实施例的车辆异常行驶检测方法，因此具有本申请第一方面实施例的所有有益效果。
本申请第三方面实施例提供了一种计算机可读存储介质，计算机可读存储介质存储有计算机可执行信号，计算机可执行信号用于执行如本申请第一方面任一项实施例的车辆异常行驶检测方法。
执行以上描述的图1中的方法步骤S110至步骤S130，执行以上描述的图2中的方法步骤S210至步骤S240。
由于第三方面的计算机存储介质可执行本申请第一方面实施例的车辆异常行驶检测方法，因此具有本申请第一方面的所有有益效果。
以上所描述的装置实施例仅仅是示意性的，其中作为分离部件说明的单元可以是或者也可以不是物理上分开的，作为单元显示的部件可以是或者也可以不是物理单元，即可以位于一个地方，或者也可以分布到多个网络单元上。可以根据实际的需要选择其中的部分或者全部单元来实现本实施例方案的目的。
通过以上的实施方式的描述，本领域普通技术人员可以理解，上文中所公开方法中的全部或某些步骤、系统可以被实施为软件、固件、硬件及其适当的组合。某些物理组件或所有物理组件可以被实施为由处理器，如中央处理器、数字信号处理器或微处理器执行的软件，或者被实施为硬件，或者被实施为集成电路，如专用集成电路。这样的软件可以分布在可读介质上，可读介质可以包括计算机存储介质(或非暂时性介质)和通信介质(或暂时性介质)。如本领域普通技术人员公知的，术语计算机存储介质包括在用于存储信息(诸如计算机可读信号、数据结构、指令模块或其他数据)的任何方法或技术中实施的易失性和非易失性、可移除和不可移除介质。计算机存储介质包括但不限于RAM、ROM、EEPROM、闪存或其他存储器技术、CD-ROM、数字多功能盘(DVD)或其他光盘存储、磁盒、磁带、磁盘存储或其他磁存储装置、或者可以用于存储期望的信息并且可以被计算机访问的任何其他的介质。此外，本领域普通技术人员公知的是，通信介质通常包含计算机可读信号、数据结构、指令模块或者诸如载波或其他传输机制之类的调制数据信号中的其他数据，并且可包括任何信息递送介质。
上面结合附图对本申请实施例作了详细说明，但是本申请不限于上述实施例，在所属技术领域普通技术人员所具备的知识范围内，还可以在不脱离本申请宗旨的前提下，做出各种变化。

Claims (10)

  1. 一种车辆异常行驶检测方法，其特征在于，包括：
    获取路段视频，并根据所述路段视频，确定待跟踪的目标车辆；
    通过改进的卡尔曼滤波算法对所述待跟踪的目标车辆进行跟踪，得到对应的行驶轨迹，其中，所述改进的卡尔曼滤波算法的协方差矩阵通过预设的BP神经网络模型进行优化；
    根据所述行驶轨迹，判断所述目标车辆是否行驶异常。
  2. 根据权利要求1所述的方法，其特征在于，所述根据所述路段视频，确定待跟踪的目标车辆，包括：
    通过贝叶斯算法模型对所述路段视频中的噪声信息进行分类；
    通过高斯模型对所述路段视频进行除噪，得到道路行驶背景图像；
    通过道路区域目标车辆检测模型对所述道路行驶背景图像进行识别，确定待跟踪的目标车辆。
  3. 根据权利要求2所述的方法，其特征在于，在所述通过道路区域目标车辆检测模型对所述道路行驶背景图像进行识别，确定待跟踪的目标车辆之前，所述方法还包括：
    通过道路上带有车辆识别框的训练数据集对所述目标车辆检测模型进行训练；
    通过拍摄道路行车视频来验证训练结果是否满足预设标准，如不满足，继续训练，直至满足所述预设标准；
    其中，所述目标车辆检测模型为卷积神经网络模型。
  4. 根据权利要求1所述的方法，其特征在于，所述根据所述行驶轨迹，判断所述目标车辆是否行驶异常，包括：
    计算所述目标车辆的行驶轨迹持续异常的持续时长；
    当所述持续时长超过预设的定时器的触发时长，判断所述目标车辆行驶异常。
  5. 根据权利要求4所述的方法，其特征在于，在计算所述目标车辆维持相同行驶轨迹的持续时长之前，包括：
    以帧为单位将满足预设的存储触发条件的所述目标车辆的行驶轨迹存储至时空序列；
    其中，所述存储触发条件为检测到的目标车辆具有的速度大小和/或方向。
  6. 根据权利要求5所述的方法，其特征在于，所述计算所述目标车辆的行驶轨迹持续异常的持续时长，包括：
    计算所述目标车辆持续遮挡车道线的帧数，得到第一持续时长；
    对应地，所述判断所述目标车辆行驶异常，包括：
    判定所述目标车辆为异常变道或者违规占用多车道。
  7. 根据权利要求5所述的方法，其特征在于，所述计算所述目标车辆的行驶轨迹持续异常的持续时长，包括：
    计算所述目标车辆的轨迹和其他行驶车辆的轨迹不同的帧数，得到第二持续时长；
    对应地，所述判断所述目标车辆行驶异常，包括：
    判定所述目标车辆为酒驾或者毒驾。
  8. 根据权利要求5所述的方法，其特征在于，所述计算所述目标车辆的行驶轨迹持续异常的持续时长，包括：
    计算所述目标车辆的行驶速度持续超过速度极值的帧数，得到第三持续时长；
    对应地，所述判断所述目标车辆行驶异常，包括：
    基于K-Mean算法，对所述第三持续时长内对应的连续多帧图像的行驶速度进行聚类处理，得到速度均值；
    当所述速度均值超过所述速度极值时，判定所述目标车辆超速。
  9. 一种电子设备，其特征在于，包括：
    至少一个存储器；
    至少一个处理器；
    至少一个程序；
    所述程序被存储在所述存储器中，所述处理器执行至少一个所述程序以实现如权利要求1至8任一项所述的车辆异常行驶检测方法。
  10. 一种计算机可读存储介质，其特征在于，所述计算机可读存储介质存储有计算机可执行信号，所述计算机可执行信号用于执行如权利要求1至8任一项所述的车辆异常行驶检测方法。
PCT/CN2022/104661 2022-04-19 2022-07-08 车辆异常行驶检测方法、电子设备、存储介质 WO2023201904A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202210408112.5A CN114925747A (zh) 2022-04-19 2022-04-19 车辆异常行驶检测方法、电子设备、存储介质
CN202210408112.5 2022-04-19

Publications (1)

Publication Number Publication Date
WO2023201904A1 true WO2023201904A1 (zh) 2023-10-26

Family

ID=82806126

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2022/104661 WO2023201904A1 (zh) 2022-04-19 2022-07-08 车辆异常行驶检测方法、电子设备、存储介质

Country Status (2)

Country Link
CN (1) CN114925747A (zh)
WO (1) WO2023201904A1 (zh)

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117150388A (zh) * 2023-11-01 2023-12-01 江西现代职业技术学院 汽车底盘的异常状态检测方法及系统
CN117237926A (zh) * 2023-11-14 2023-12-15 广州斯沃德科技有限公司 一种车辆行驶状态确定方法、系统、设备及可读存储介质
CN117315602A (zh) * 2023-11-29 2023-12-29 西南交通大学烟台新一代信息技术研究院 基于图像识别的高速公路异常识别方法及系统
CN117576926A (zh) * 2024-01-09 2024-02-20 厦门星纵物联科技有限公司 车辆违规检测方法、设备及存储介质
CN117649165A (zh) * 2024-01-30 2024-03-05 南昌理工学院 一种运输安全管控方法及系统

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2012003549A (ja) * 2010-06-17 2012-01-05 Toshiba Teli Corp 異常走行車両検出システムおよび道路監視プログラム
CN110178167A (zh) * 2018-06-27 2019-08-27 潍坊学院 基于摄像机协同接力的路口违章视频识别方法
CN112630659A (zh) * 2020-06-01 2021-04-09 青岛大学 一种基于改进bp-ekf算法的锂电池soc估算方法
CN114201890A (zh) * 2022-02-16 2022-03-18 苏州博宇鑫交通科技有限公司 一种高速公路异常事件场景库构建方法

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2012003549A (ja) * 2010-06-17 2012-01-05 Toshiba Teli Corp 異常走行車両検出システムおよび道路監視プログラム
CN110178167A (zh) * 2018-06-27 2019-08-27 潍坊学院 基于摄像机协同接力的路口违章视频识别方法
CN112630659A (zh) * 2020-06-01 2021-04-09 青岛大学 一种基于改进bp-ekf算法的锂电池soc估算方法
CN114201890A (zh) * 2022-02-16 2022-03-18 苏州博宇鑫交通科技有限公司 一种高速公路异常事件场景库构建方法

Cited By (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117150388A (zh) * 2023-11-01 2023-12-01 江西现代职业技术学院 汽车底盘的异常状态检测方法及系统
CN117150388B (zh) * 2023-11-01 2024-01-26 江西现代职业技术学院 汽车底盘的异常状态检测方法及系统
CN117237926A (zh) * 2023-11-14 2023-12-15 广州斯沃德科技有限公司 一种车辆行驶状态确定方法、系统、设备及可读存储介质
CN117237926B (zh) * 2023-11-14 2024-02-06 广州斯沃德科技有限公司 一种车辆行驶状态确定方法、系统、设备及可读存储介质
CN117315602A (zh) * 2023-11-29 2023-12-29 西南交通大学烟台新一代信息技术研究院 基于图像识别的高速公路异常识别方法及系统
CN117315602B (zh) * 2023-11-29 2024-02-06 西南交通大学烟台新一代信息技术研究院 基于图像识别的高速公路异常识别方法及系统
CN117576926A (zh) * 2024-01-09 2024-02-20 厦门星纵物联科技有限公司 车辆违规检测方法、设备及存储介质
CN117576926B (zh) * 2024-01-09 2024-03-12 厦门星纵物联科技有限公司 车辆违规检测方法、设备及存储介质
CN117649165A (zh) * 2024-01-30 2024-03-05 南昌理工学院 一种运输安全管控方法及系统

Also Published As

Publication number Publication date
CN114925747A (zh) 2022-08-19

Similar Documents

Publication Publication Date Title
WO2023201904A1 (zh) 车辆异常行驶检测方法、电子设备、存储介质
US11069072B2 (en) Dynamic multi-camera tracking of moving objects in motion streams
US9292750B2 (en) Method and apparatus for detecting traffic monitoring video
US10969788B2 (en) ECU, autonomous vehicle including ECU, and method of determining driving lane for the same
WO2021155685A1 (zh) 一种更新地图的方法、装置和设备
CN111209361B (zh) 跟车目标选择方法、装置、电子设备及可读存储介质
CN105513349A (zh) 基于双视角学习的山区高速公路车辆事件检测方法
CN105979143B (zh) 一种调整球机拍摄参数的方法及装置
US20220237919A1 (en) Method, Apparatus, and Computing Device for Lane Recognition
Liu et al. Vision-based long-distance lane perception and front vehicle location for full autonomous vehicles on highway roads
WO2019065970A1 (ja) 車両用外界認識装置
CN112447060A (zh) 识别车道的方法、装置及计算设备
CN111507126B (zh) 一种驾驶辅助系统的报警方法、装置及电子设备
CN113029185B (zh) 众包式高精度地图更新中道路标线变化检测方法及系统
CN114220040A (zh) 一种泊车方法、终端及计算机可读存储介质
CN114627409A (zh) 一种车辆异常变道的检测方法及装置
CN113111707A (zh) 一种基于卷积神经网络的前车检测与测距方法
JP2022537557A (ja) 運転可能な領域情報を決定するための方法及び装置
CN116990768A (zh) 预测航迹处理方法、装置、电子设备及可读介质
CN115222779B (zh) 一种车辆切入检测方法、装置及存储介质
CN112906428A (zh) 影像侦测区域取得方法及空间使用情况的判定方法
JP6962726B2 (ja) 走路認識装置
CN115526053A (zh) 一种道路建模方法、装置以及设备
TWI730509B (zh) 影像偵測區域取得方法及空間使用情況的判定方法
CN113128264B (zh) 一种车辆区域确定方法、装置及电子设备