Road condition safety analysis method based on vehicle-road cooperation and related device

Abstract

The invention discloses a road condition safety analysis method based on vehicle-road cooperation and a related device, and relates to the technical field of intelligent traffic, and the method comprises the steps: carrying out the speed analysis of a current vehicle based on an echo signal; performing face and hand positioning and definition optimization on the plurality of frames of driver images to obtain a plurality of frames of driver face images and driver hand images; analyzing the position relation of key points of each frame of driver face image and hand image so as to recognize abnormal driving behaviors in combination with an abnormal behavior discrimination model based on a fusion attention mechanism; performing environmental road condition analysis based on the plurality of frames of road section images in combination with visibility analysis; performing road section congestion analysis based on the plurality of frames of road section images; and determining road condition safety information based on the driving speed, the driving abnormal behavior, the environment road condition information and the road section congestion information so as to carry out road condition early warning on a vehicle terminal. According to the invention, sufficient comprehensive and accurate road condition information can be provided, so that corresponding early warning can be timely and accurately sent to the vehicle terminal.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent transportation, and particularly relates to a road condition safety analysis method and related device based on vehicle-road cooperation. Background Art

[0002] With the continuous increase in the number of automobiles, the road conditions have become more complex, resulting in an increasing number of traffic accidents. In response to this, vehicle-road cooperation technology has been gradually introduced to analyze road conditions. The vehicle-road cooperation technology can analyze road condition information in real time, realize the sharing and interaction of road condition information between vehicles and road facilities, so that drivers can understand the current road condition safety, thereby reducing the occurrence of traffic accidents. The analysis of drivers' abnormal driving behaviors is an important part of road condition analysis. Currently, the analysis of abnormal driving behaviors is usually carried out only by training models with full-map data, but this method consumes a large amount of computing resources, and the missed detection rate of abnormal behaviors is also relatively high, seriously affecting the reliability of the analysis of abnormal driving behaviors. At the same time, most current road condition analyses lack the analysis of road section visibility, resulting in the final obtained road condition information being incomplete and unable to provide drivers with sufficient specific road condition information, thus failing to effectively improve the reliability of road condition safety analysis. Summary of the Invention

[0003] The purpose of the present invention is to overcome the deficiencies of the prior art. The present invention provides a road condition safety analysis method and related device based on vehicle-road cooperation, which can provide comprehensive and accurate road condition information, so as to issue corresponding warnings to vehicle terminals in a timely and accurate manner.

[0004] To solve the above technical problems, the present invention provides a road condition safety analysis method based on vehicle-road cooperation, which is applied to a plurality of information collection devices and information release devices, and the information collection devices are communicatively connected to the information release devices; the method includes: Performing speed analysis of the current vehicle based on the echo signals collected by the information collection devices to obtain the corresponding driving speed; Performing face positioning, hand positioning and clarity optimization on a plurality of frames of driver images of the current vehicle collected by a plurality of information collection devices to obtain corresponding a plurality of frames of driver face images and driver hand images; Analyzing the key point position relationships of each frame of driver face images and driver hand images, and identifying abnormal driving behaviors based on the key point position relationships in combination with an abnormal behavior discrimination model based on a fusion attention mechanism; Performing environmental road condition analysis based on a plurality of frames of road section images collected by a plurality of information collection devices in combination with visibility analysis to obtain environmental road condition information; Performing road section congestion analysis based on a plurality of frames of road section images to obtain road section congestion information; Based on the driving speed, driving abnormal behaviors, environmental road conditions information, and road section congestion information, corresponding road condition safety information is determined, and the information publishing device issues a road condition warning to the vehicle terminal based on the road condition safety information.

[0005] Optionally, the speed analysis of the current vehicle based on the echo signal collected by the information collection device to obtain the corresponding driving speed includes: Mix the echo signal with the local oscillator signal to obtain an intermediate frequency signal, and determine the corresponding speed spectrum and spectrum position based on the peak spectral line of the intermediate frequency signal; Based on the speed spectrum and spectrum position, combined with the radial speed of the radar in the information collection device, determine the corresponding driving speed of the current vehicle.

[0006] Optionally, the facial positioning, hand positioning, and clarity optimization of several frames of driver images of the current vehicle collected by several information collection devices to obtain the corresponding several frames of driver facial images and driver hand images include: Input each frame of the driver image into the target positioning model for facial positioning and hand positioning to obtain the corresponding facial positioning image and hand positioning image for each frame; Determine the first clarity corresponding to the facial positioning image and hand positioning image for each frame based on the contrast and noise value of each pixel point in the facial positioning image and hand positioning image for each frame; Filter the facial positioning image and hand positioning image for each frame with filters of different filter coefficients to obtain the filtered facial positioning image and filtered hand positioning image corresponding to different filter coefficients; Conduct horizontal clarity analysis and vertical clarity analysis on the intermediate frequency images of the filtered facial positioning image and filtered hand positioning image for each frame to obtain horizontal clarity analysis data and vertical clarity analysis data, and determine the second clarity based on the horizontal clarity analysis data and vertical clarity analysis data; Determine the target clarity of the facial positioning image and hand positioning image for each frame based on the first clarity and the second clarity, and optimize the clarity of the facial positioning image and hand positioning image for each frame based on the comparison result between the target clarity and the preset clarity threshold to obtain the corresponding several frames of driver facial images and driver hand images.

[0007] Optionally, analyzing the key point position relationship of each frame of the driver facial image and driver hand image, and identifying driving abnormal behaviors based on the key point position relationship combined with the abnormal behavior discrimination model based on the fusion attention mechanism includes: Analyze the eye key points, ear key points, and lip key points of the driver's facial images in each frame, as well as the hand key points of the driver's hand images in each frame, and determine the first key point position distance between the ear key points and the hand key points and the second key point position distance between the lip key points and the hand key points; Analyze the first distracted driving behavior of the driver based on the first key point position distance and the second key point position distance, and analyze the second distracted driving behavior of the driver based on the eye key points of the driver's facial images in each frame; Based on the driver's facial images in each frame and the eye key points, use an abnormal behavior discrimination model based on a fusion attention mechanism to analyze the driver's fatigue driving behavior.

[0008] Optionally, the environmental road condition analysis based on several frames of road section images collected by several information collection devices combined with visibility analysis to obtain environmental road condition information includes: Perform grayscale processing on each frame of road section image to obtain each frame of grayscale road section image, and extract texture features, pixel brightness features, and color features based on each frame of grayscale road section image to obtain texture feature information, pixel brightness feature information, and color feature information; Based on the texture feature information, pixel brightness feature information, and color feature information, identify the road section icing area to obtain road section icing area information; Based on each frame of road section image, use a fog detection model to determine the fog concentration of the road section, and based on the fog concentration of the road section, determine the visibility of the road section. Generate environmental road condition information based on the road section icing area information and the road section visibility.

[0009] Optionally, the road section congestion analysis based on several frames of road section images to obtain road section congestion information includes: Calculate the total vehicle area based on several frames of road section images, and determine the congested road section based on the ratio of the total vehicle area to the road section area; Divide the congested road section into several grid lane road sections, and count the vehicle density of each grid lane road section; Based on the vehicle density of each grid lane road section combined with the statistics of lane-changing vehicles, determine the road section congestion information.

[0010] Optionally, determining the corresponding road condition safety information based on the driving speed, driving abnormal behavior, environmental road condition information, and road section congestion information, and the information publishing device issues a road condition warning to the vehicle terminal based on the road condition safety information, including: Compare the driving speed with the preset speed limit value, and determine the vehicle speeding information based on the comparison result; Based on the vehicle speeding information, driving abnormal behavior, environmental road condition information, and road section congestion information, determine the corresponding road condition safety information, and match the corresponding warning level based on the road condition safety information; The information publishing device gives a road condition warning to the vehicle terminal based on the road condition safety information and the corresponding warning level.

[0011] In addition, the present invention also provides a road condition safety analysis device based on vehicle-road collaboration, which is applied to a plurality of information collection devices and information publishing devices, and the information collection devices are communicatively connected to the information publishing devices; the device includes: A speed analysis module: configured to analyze the speed of the current vehicle based on the echo signals collected by the information collection devices to obtain the corresponding driving speed; A face and hand positioning module: configured to perform face positioning, hand positioning, and clarity optimization on a plurality of frames of driver images of the current vehicle collected by a plurality of information collection devices to obtain the corresponding plurality of frames of driver face images and driver hand images; An abnormal behavior analysis module: configured to analyze the key point position relationships of each frame of driver face images and driver hand images, and identify driving abnormal behaviors based on the key point position relationships in combination with an abnormal behavior discrimination model based on a fusion attention mechanism; An environmental road condition analysis module: configured to perform environmental road condition analysis based on a plurality of frames of road section images collected by a plurality of information collection devices in combination with visibility analysis to obtain environmental road condition information; A congestion analysis module: configured to perform road section congestion analysis based on a plurality of frames of road section images to obtain road section congestion information; A safety warning module: configured to determine the corresponding road condition safety information based on the driving speed, driving abnormal behaviors, environmental road condition information, and road section congestion information, and the information publishing device gives a road condition warning to the vehicle terminal based on the road condition safety information.

[0012] In addition, the present invention also provides a road condition safety analysis system based on vehicle-road collaboration, the system includes a plurality of information collection devices and information publishing devices, the information collection devices are communicatively connected to the information publishing devices, and the system is configured to execute the above-mentioned road condition safety analysis method based on vehicle-road collaboration.

[0013] In addition, the present invention also provides a computer-readable storage medium, the computer-readable storage medium stores computer instructions, and when the computer instructions run on an electronic device, the electronic device is caused to execute the above-mentioned road condition safety analysis method based on vehicle-road collaboration.

[0014] In the embodiments of the present invention, each frame of driver image is input into a target positioning model to obtain corresponding facial positioning images and hand positioning images for each frame. By analyzing the target clarity of each frame of facial positioning image and hand positioning image to optimize their clarity, clearer driver facial images and hand images can be obtained, improving the accuracy of driving abnormal behavior analysis. Analyzing the key point position relationships of each frame of driver facial image and driver hand image, and identifying driving abnormal behaviors based on the key point position relationships in combination with an abnormal behavior discrimination model based on a fusion attention mechanism can avoid consuming excessive computing resources, greatly reducing the missed detection rate of abnormal behaviors and effectively improving the reliability of driving abnormal behavior recognition. Based on the analysis of the environmental road conditions by combining several frames of road section images collected by a number of information collection devices with visibility analysis, the environmental road condition analysis becomes more comprehensive. Conducting road section congestion analysis based on several frames of road section images, and determining corresponding road condition safety information based on driving speed, driving abnormal behaviors, environmental road condition information, and road section congestion information, so as to obtain sufficiently comprehensive and accurate road condition information, more accurately analyze the current road safety situation, and thus be able to issue corresponding warnings to the vehicle terminal in a timely and accurate manner, reducing the occurrence of accidents. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the following described drawings are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings.

[0016] Figure 1 is a flowchart of a road condition safety analysis method based on vehicle-road collaboration in the embodiments of the present invention; Figure 2 is a flowchart of a road condition safety analysis method based on vehicle-road collaboration in another embodiment of the present invention; Figure 3 is a schematic structural composition diagram of a road condition safety analysis system based on vehicle-road collaboration in the embodiments of the present invention; Figure 4 is a schematic structural composition diagram of a road condition safety analysis device based on vehicle-road collaboration in the embodiments of the present invention; Figure 5 is an application diagram of an information collection device and an information publishing device in the embodiments of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0017] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without making creative efforts belong to the scope of protection of the present invention.

[0018] Embodiment 1

[0019] Please refer to Figure 1 , Figure 1 , which is a schematic flowchart of a road condition safety analysis method based on vehicle-road collaboration in the embodiments of the present invention. The method is applied to a plurality of information collection devices and information publishing devices, and the information collection devices are communicatively connected to the information publishing devices; the method includes: S11: Analyze the speed of the current vehicle based on the echo signal collected by the information collection device to obtain the corresponding driving speed; In the specific implementation process of the present invention, the analyzing the speed of the current vehicle based on the echo signal collected by the information collection device to obtain the corresponding driving speed includes: mixing the echo signal with the local oscillator signal to obtain an intermediate frequency signal, and determining the corresponding speed spectrum and spectrum position based on the peak spectral line of the intermediate frequency signal; determining the corresponding driving speed of the current vehicle based on the speed spectrum and spectrum position in combination with the radial speed of the radar in the information collection device.

[0020] Specifically, the radar built into the information collection device emits pulse signals. When the pulse signals encounter a moving vehicle, echo signals are generated. The device collects the echo signals, and the edge calculation and processing device of the information collection device processes the echo signals. The echo signals are mixed with the local oscillator signals. The local oscillator signals refer to signals generated by the radar of the information collection device itself that have the same frequency as the transmitted signals at a certain moment. The echo signals and the local oscillator signals are mixed through a mixer. The mixer is a multiplier for signals, that is, a multiplication operation is performed on the echo signals and the local oscillator signals to obtain intermediate-frequency signals. Based on the peak spectral lines of the intermediate-frequency signals, the corresponding velocity spectrum and spectral position are determined. Fourier spectral analysis is performed on the intermediate-frequency signals to obtain the distance spectrum of the intermediate-frequency signals, that is, the time-domain information of the intermediate-frequency signals is observed and analyzed through frequency-domain characteristics, and the distance spectrum is obtained by using Fourier transform. The number of peak spectral lines whose peaks exceed the preset distance amplitude threshold in the distance spectrum is obtained, and the positions of the peak spectral lines in the distance spectrum are obtained. Fourier spectral analysis is performed on each peak spectral line to obtain the velocity spectrum, the peak spectral lines that exceed the preset velocity amplitude threshold in the velocity spectrum are determined, and the position of the peak spectral line in the intermediate-frequency signal is recorded, which is the spectral position. Based on the velocity spectrum and the spectral position, combined with the radial velocity of the radar in the information collection device, the corresponding driving speed of the current vehicle is determined. The radial distance from the radar to the monitoring point is determined according to the spectral position of the peak spectral lines that exceed the velocity amplitude threshold in the velocity spectrum. A monitoring point of a radar can be preset for each lane, and the monitoring point is set in the edge area of the radar monitoring range. The radial velocity of the radar is determined according to the frequency corresponding to each spectral position that exceeds the preset velocity amplitude threshold. The radial velocity can be the radial velocity of the current vehicle relative to the radar. According to the radial velocity and the radial distance, combined with the positional relationship between the monitoring point and the radar, the corresponding driving speed of the current vehicle is determined.

[0021] S12: Perform face localization, hand localization, and clarity optimization on several frames of driver images of the current vehicle collected by several information collection devices to obtain corresponding several frames of driver face images and driver hand images; In the specific implementation process of the present invention, the face localization, hand localization, and clarity optimization of several frames of driver images of the current vehicle collected by several information collection devices are performed to obtain corresponding several frames of driver face images and driver hand images, including: inputting each frame of driver image into a target localization model for face localization and hand localization to obtain corresponding face localization images and hand localization images for each frame; determining the first clarity corresponding to each frame of face localization image and hand localization image based on the contrast and noise value of each pixel point in each frame of face localization image and hand localization image; filtering each frame of face localization image and hand localization image with filters having different filtering coefficients to obtain filtered face localization images and filtered hand localization images corresponding to different filtering coefficients; performing horizontal clarity analysis and vertical clarity analysis on the intermediate frequency images of each frame of filtered face localization image and filtered hand localization image to obtain horizontal clarity analysis data and vertical clarity analysis data, and determining the second clarity based on the horizontal clarity analysis data and the vertical clarity analysis data; determining the target clarity of each frame of face localization image and hand localization image based on the first clarity and the second clarity, and performing clarity optimization on each frame of face localization image and hand localization image based on the comparison result between the target clarity and a preset clarity threshold to obtain corresponding several frames of driver face images and driver hand images.

[0022] Specifically, a number of frames of driver images of the current vehicle are continuously collected by a camera in the information collection device. Each frame of the driver image is input into the target positioning model for face positioning and hand positioning. The target positioning model uses a deep neural network model, and an image data set for face annotation and hand annotation is used as the sample data set to train the deep neural network model. The trained deep neural network model is used as the target positioning model to obtain the corresponding face positioning images and hand positioning images for each frame. The located face positioning images and hand positioning images for each frame are in the same frame of the driver image. Based on the contrast and noise values of each pixel point in the face positioning images and hand positioning images for each frame, the first sharpness corresponding to the face positioning images and hand positioning images for each frame is determined. The contrast of each pixel point can be used to determine the gray value of each pixel point through a gray histogram, and the contrast of each pixel point is determined according to the gray value. The noise value of each pixel point can be determined through the gray level value of each pixel point. The gray level value represents the brightness level relationship between the darkest black and the brightest white of the pixel point. The noise value corresponding to each gray level value is matched through a database. The contrast and noise value of each pixel point in the image are factors affecting the image sharpness. The first sharpness corresponding to the face positioning images and hand positioning images for each frame is determined through the contrast and noise value of each pixel point and their corresponding weights. Filters with different filter coefficients are used to filter the face positioning images and hand positioning images for each frame. The parameters of each filter are set according to different filter coefficients. The filter coefficients can include a first filter coefficient and a second filter coefficient, and the filter coefficients are set by experience to filter the face positioning images and hand positioning images for each frame, and the filtered face positioning images and filtered hand positioning images corresponding to different filter coefficients are obtained.Perform horizontal sharpness analysis and vertical sharpness analysis on the intermediate-frequency images of each frame of filtered facial localization images and filtered hand localization images. Calculate the difference images between the filtered facial localization images corresponding to each filtering coefficient, that is, perform pixel-by-pixel subtraction between the filtered facial localization images with different filtering coefficients to obtain the difference images. These difference images are images containing intermediate-frequency information, so these difference images are the intermediate-frequency images. The determination method of the intermediate-frequency images of each frame of filtered hand localization images is the same as that of the filtered facial localization images and will not be elaborated here. Each frame of image is processed by difference calculation according to the above steps to obtain the corresponding intermediate-frequency images. Traverse each row of pixel points of the intermediate-frequency images of each frame of filtered facial localization images and filtered hand localization images, and determine the pixel point with the largest corresponding value as the first pixel point. Calculate the average value of the values corresponding to all the determined first pixel points to obtain the corresponding first average value, which is the horizontal sharpness analysis data. Traverse each column of pixel points of the intermediate-frequency images of each frame of filtered facial localization images and filtered hand localization images, and determine the pixel point with the largest value in each column of pixel points as the second pixel point. Calculate the average value of the values corresponding to all the determined second pixel points to obtain the corresponding second average value, which is the vertical sharpness analysis data. Determine the second sharpness based on the horizontal sharpness analysis data and the vertical sharpness analysis data. The second sharpness can be determined by calculating the average value or selecting the maximum value of the horizontal sharpness analysis data and the vertical sharpness analysis data. Determine the target sharpness of each frame of facial localization images and hand localization images based on the first sharpness and the second sharpness. Determine the target sharpness of each frame of facial localization images and hand localization images through the first sharpness, the second sharpness, and their weights. And perform sharpness optimization on each frame of facial localization images and hand localization images based on the comparison result between the target sharpness and the preset sharpness threshold. If the target sharpness is greater than or equal to the preset sharpness threshold, directly use each frame of facial localization images and hand localization images as the corresponding several frames of driver facial images and driver hand images. If the target sharpness threshold is less than the preset sharpness threshold, perform sharpness optimization on each frame of facial localization images and hand localization images. The sharpness optimization uses image enhancement processing, and the image sharpness can be enhanced by histogram equalization. Histogram equalization is to change the gray histogram of the original image from a relatively concentrated gray interval to a uniform distribution within the entire gray range, thereby increasing the image clarity to obtain the corresponding several frames of driver facial images and driver hand images. Judging whether to optimize the image sharpness through sharpness can improve the sharpness of images with insufficient sharpness, and at the same time avoid excessive image enhancement for images with sufficient sharpness.

[0023] S13: Analyze the key point position relationship between the driver's facial images and hand images of each frame, and identify abnormal driving behaviors based on the key point position relationship in combination with the abnormal behavior discrimination model based on the fusion attention mechanism; In the specific implementation process of the present invention, the analysis of the key point position relationship between the driver's facial images and hand images of each frame, and the identification of abnormal driving behaviors based on the key point position relationship in combination with the abnormal behavior discrimination model based on the fusion attention mechanism include: analyzing the eye key points, ear key points, and lip key points of the driver's facial images of each frame, and the hand key points of the driver's hand images of each frame, and determining the first key point position distance between the ear key points and the hand key points and the second key point position distance between the lip key points and the hand key points; analyzing the first distracted driving behavior of the driver based on the first key point position distance and the second key point position distance, and analyzing the second distracted driving behavior of the driver based on the eye key points of the driver's facial images of each frame; analyzing the fatigue driving behavior of the driver based on the driver's facial images of each frame and the eye key points using the abnormal behavior discrimination model based on the fusion attention mechanism, based on the driver's facial images of each frame and the eye key points.

[0024] Specifically, the eye key points, ear key points, and lip key points of each frame of the driver's facial image and the hand key points of each frame of the driver's hand image are analyzed through a key point recognition model. The key point recognition model is a converged model obtained by training a sample data set input into a deep neural network. The eye key points include the left eye key point and the right eye key point, the ear key points include the right ear key point and the left ear key point, and the hand key points include the left hand key point and the right hand key point. The key point information may include the key point information marked as needed, such as hand bone points and eye contour points and their position coordinates, etc., and determine the first key point position distance between the ear key points and the hand key points and the second key point position distance between the lip key points and the hand key points, that is, determine the key point distance between the ear and the hand and the key point distance between the lip and the hand in each frame of the image. Calculate the position distance between the two based on the position information of the ear key points and the position information of the hand key points, which is the first key point position distance, such as the distance from the right hand to the right ear. Calculate the position distance between the two based on the position information of the lip key points and the position information of the hand key points, which is the second key point position distance, such as the distance from the left hand to the lip. Analyze the first distracted driving behavior of the driver based on the first key point position distance and the second key point position distance. The first distracted driving behavior is that the driver is distracted while holding an item. If the first key point position distance of the corresponding frame image is less than the preset first distance threshold, target item recognition is performed on the driver's facial image and the driver's hand image corresponding to the first key point position distance less than the first distance threshold. If a target item is recognized, the target item includes but is not limited to a mobile phone, etc., then the driver has a distracted driving behavior. If the second key point position distance is less than the preset second distance threshold, target item recognition is performed on the driver's facial image and the driver's hand image corresponding to the second key point position distance less than the second distance threshold. The target item includes but is not limited to a cigarette, etc. If a target item is recognized, then the driver has a distracted driving behavior. The two are comprehensively analyzed to determine whether the driver has the first distracted driving behavior. Through the analysis of the key point position relationship, it is possible to avoid false detection caused by the driver's normal operations such as raising the hand or stroking the hair. Analyze the second distracted driving behavior of the driver based on the eye key points of each frame of the driver's facial image. There is also a distracted driving behavior where the user is not holding an item, but their attention is not in the front driving area. The second distracted driving behavior is the distracted driving behavior where the user's attention direction deviates. Analyze the line-of-sight direction based on the eye key points to obtain the corresponding line-of-sight direction, and analyze the head rotation angle of the driver based on each frame of the driver's facial image. Analyze the driver's fixation area based on the head rotation angle and the line-of-sight direction. If the fixation area is other areas outside the vehicle's front windshield area, then the driver has the second distracted driving behavior.Based on each frame of the driver's facial image and eye key points, the fatigue driving behavior of the driver is analyzed using an abnormal behavior discrimination model based on a fusion attention mechanism. The abnormal behavior discrimination model based on a fusion attention mechanism includes a sequence input layer, a long short-term memory network layer, an attention layer, and an output layer. The long short-term memory network layer has better performance in processing sequence data. It includes an input gate, an output gate, and a forget gate. Introducing the forget gate can prevent gradient explosion or gradient disappearance. The learning mechanism in the attention layer is implemented through a fully connected layer. Due to the relatively complex temporal correlation in each frame of the driver's facial image, the abnormal behavior discrimination model based on a fusion attention mechanism can effectively improve the accuracy of fatigue driving behavior analysis. The driving abnormal behaviors of the driver consist of fatigue driving behavior, the first distracted driving behavior, and the second distracted driving behavior.

[0025] S14: Based on several frames of road section images collected by several information collection devices and combined with visibility analysis, perform environmental road condition analysis to obtain environmental road condition information; In the specific implementation process of the present invention, the step of performing environmental road condition analysis based on several frames of road section images collected by several information collection devices and combined with visibility analysis to obtain environmental road condition information includes: performing grayscale processing on each frame of road section image to obtain each frame of grayscale road section image, and extracting texture features, pixel brightness features, and color features based on each frame of grayscale road section image to obtain texture feature information, pixel brightness feature information, and color feature information; identifying the road section icing area based on the texture feature information, pixel brightness feature information, and color feature information to obtain road section icing area information; determining the road section fog concentration using a group fog detection model based on each frame of road section image, and determining the road section visibility based on the road section fog concentration, and generating environmental road condition information based on the road section icing area information and the road section visibility.

[0026] Specifically, the camera in the information collection device captures images of the current driving section, which are section images. Each frame of the section images is grayscale processed. The grayscale weighting coefficients of each frame of the section images are analyzed through the seagull algorithm. The seagull algorithm is an iterative calculation process. In each iteration process, the optimal seagull position in that iteration is recorded. Then, through the entire iteration process, the grayscale weighting coefficients of the images are determined. The weighting weights of each channel in each frame of the section images are adjusted through the grayscale weighting coefficients to achieve the grayscale processing of the images, and each frame of grayscale section images is obtained, making the overall information of the grayscale image closer to that of the original image. Based on each frame of grayscale section images, texture features, pixel brightness features, and color features are extracted. Each frame of grayscale section images is segmented to obtain several segmented images corresponding to each frame of grayscale section images. The segmented images are used to extract texture features, pixel brightness features, and color features through corresponding feature extraction models. The texture feature extraction can adopt the local binary pattern feature extraction method, the pixel brightness feature extraction can adopt an image brightness extraction tool, and the color feature extraction can adopt an image color extraction tool to obtain texture feature information, pixel brightness feature information, and color feature information. Based on the texture feature information, pixel brightness feature information, and color feature information, the icing area of the section is identified. The similarity between the texture feature information of each segmented image and the standard ice surface texture feature is calculated, and the segmented images with similarity greater than or equal to the preset similarity threshold are retained. The retained segmented images are used as the first image set. The average pixel brightness is calculated based on the pixel brightness feature information of each segmented image in the first image set. The segmented images with average pixel brightness greater than or equal to the preset brightness threshold are retained to form the second image set. The road surface temperature is matched according to the color feature information of each segmented image in the second image set, that is, the corresponding road surface temperature is matched according to the color features of each pixel point. The linear relationship between the color feature information and the road surface temperature has been stored in the database. The pixel points with road surface temperature greater than the preset temperature threshold in each segmented image are removed, and then the segmented images after removing the pixel points are merged to obtain the icing area of the section. At the same time, the position of the icing area is determined and the area of the icing area is calculated, that is, the icing area information of the section is obtained, which can improve the accuracy of icing area recognition. Each frame of the section images is processed according to the above to identify the icing area, and the icing area information of each frame of the section images is obtained, so that the change of the icing area of the section can be known. Based on each frame of the section images, the fog concentration of the section is determined using the fog detection model. The fog detection model is obtained by learning the mapping relationship between the section images and the fog concentration, and the visibility of the section is determined based on the fog concentration of the section. The corresponding visibility of the section is matched in the database according to the fog concentration of the section. The environmental road condition information is generated based on the icing area information of the section and the visibility of the section. The environmental road condition information obtained in this way is more comprehensive.

[0027] S15: Analyze the congestion of a road section based on a number of road section images to obtain road section congestion information; In the specific implementation process of the present invention, the analyzing the congestion of a road section based on a number of road section images to obtain road section congestion information includes: calculating the total vehicle area based on a number of road section images, and determining the congested road section based on the ratio of the total vehicle area to the road section area; dividing the congested road section into a number of grid lane road sections, and counting the vehicle density of each grid lane road section; determining the road section congestion information based on the vehicle density of each grid lane road section in combination with the statistics of lane-changing vehicles.

[0028] Specifically, calculate the total vehicle area based on a number of road section images, identify the vehicle features in each frame of road section images, calculate the total vehicle area according to the identified vehicle features, and determine the congested road section based on the ratio of the total vehicle area to the road section area. Take the ratio of the total vehicle area to the road section area as the suspected congestion value. If the suspected congestion value is greater than or equal to the preset congestion threshold, then take the corresponding road section as the congested road section. Divide the congested road section into a number of grid lane road sections, that is, longitudinally divide the congested road section into a number of lane road sections according to the position of each lane in the road section, and make a horizontal division in each lane road section according to a preset distance to obtain a number of grid lane road sections. Count the vehicle density of each grid lane road section, count the number of vehicles in each grid lane road section, and calculate the vehicle density according to the number of vehicles and the length of each grid lane road section. Determine the road section congestion information based on the vehicle density of each grid lane road section in combination with the statistics of lane-changing vehicles. Compose the vehicle density of each lane according to the vehicle density of each grid lane road section, identify the vehicles that change lanes according to each frame of road section images, count the new vehicle density of each lane after the vehicles change lanes, analyze the congestion value of each lane according to the new vehicle density of each lane, and form the road section congestion information from the congestion values of each lane.

[0029] S16: Determine the corresponding road condition safety information based on the driving speed, driving abnormal behavior, environmental road condition information and road section congestion information, and the information publishing device gives a road condition warning to the vehicle terminal based on the road condition safety information.

[0030] In the specific implementation process of the present invention, the determining the corresponding road condition safety information based on the driving speed, driving abnormal behavior, environmental road condition information and road section congestion information, and the information publishing device gives a road condition warning to the vehicle terminal based on the road condition safety information includes: comparing the driving speed with the preset speed limit value, and determining the vehicle speeding information based on the comparison result; determining the corresponding road condition safety information based on the vehicle speeding information, driving abnormal behavior, environmental road condition information and road section congestion information, and matching the corresponding warning level based on the road condition safety information; the information publishing device gives a road condition warning to the vehicle terminal based on the road condition safety information and the corresponding warning level.

[0031] Specifically, compare the driving speed with the preset speed limit value, and determine the vehicle speeding information based on the comparison result, that is, determine whether the driving speed exceeds the preset speed limit value. If it exceeds the preset speed limit value, determine how much the exceeded speed value is. Determine the corresponding road condition safety information based on the vehicle speeding information, driving abnormal behaviors, environmental road condition information, and road section congestion information. Comprehensively analyze the current road section safety level based on the vehicle speeding information, driving abnormal behaviors of the driver, information on the road surface icing area, road section visibility, and road section congestion value, which is the road condition safety information. Then, match the corresponding warning level based on the road condition safety information. Different road condition safety levels correspond to different warning levels, and for different warning levels, the font display of the information release device is different. The information release device gives a road condition warning to the vehicle terminal based on the road condition safety information and the corresponding warning level. Transmit the road condition safety information and the warning level to the information release device. The information release device displays the corresponding road condition safety information. The information release device can also display the vehicle speeding information, driving abnormal behaviors, environmental road condition information, and road section congestion information. At the same time, the information release device transmits the road condition safety information, warning level, vehicle speeding information, driving abnormal behaviors, environmental road condition information, and road section congestion information to the vehicle terminal to provide a safety warning to the vehicle terminal, such as reminding the driver that the vehicle is speeding, the driver has driving abnormal behaviors, or reminding the driver that there is an icing area in the front road section, etc., to achieve the road condition safety warning of vehicle-road cooperation.

[0032] In the embodiment of the present invention, input each frame of the driver image into the target positioning model to obtain the corresponding face positioning image and hand positioning image of each frame. Analyze the target clarity of each frame of the face positioning image and hand positioning image to optimize its clarity, and clearer driver face images and hand images can be obtained, improving the accuracy of analyzing driving abnormal behaviors. Analyze the key point position relationship of each frame of the driver face image and the driver hand image, and identify the driving abnormal behaviors based on the key point position relationship in combination with the abnormal behavior discrimination model based on the fusion attention mechanism, which can avoid consuming too much computing resources, greatly reduce the missed detection rate of abnormal behaviors, and effectively improve the reliability of identifying driving abnormal behaviors. Conduct environmental road condition analysis based on several frames of road section images collected by several information collection devices in combination with visibility analysis, making the environmental road condition analysis more comprehensive. Conduct road section congestion analysis based on several frames of road section images. Determine the corresponding road condition safety information based on the driving speed, driving abnormal behaviors, environmental road condition information, and road section congestion information. In this way, sufficient comprehensive and accurate road condition information can be obtained, and the current road safety situation can be analyzed more accurately, so as to issue corresponding warnings to the vehicle terminal in a timely and accurate manner, reducing the occurrence of accidents.

[0033] Embodiment 2

[0034] Please refer to Figure 2 , Figure 2It is a schematic flowchart of a road condition safety analysis method based on vehicle-road cooperation in another embodiment of the present invention. The method is applied to a plurality of information collection devices and information release devices, and the information collection devices are communicatively connected to the information release devices. The method includes: S201: Analyze the speed of the current vehicle based on the echo signal collected by the information collection device to obtain the corresponding driving speed; S202: Input each frame of the driver image into the target positioning model for face positioning and hand positioning to obtain the corresponding face positioning image and hand positioning image for each frame; S203: Determine the first clarity corresponding to the face positioning image and hand positioning image for each frame based on the contrast and noise value of each pixel point in the face positioning image and hand positioning image for each frame; S204: Filter the face positioning image and hand positioning image for each frame with filters of different filter coefficients to obtain the filtered face positioning image and filtered hand positioning image corresponding to different filter coefficients; S205: Perform horizontal clarity analysis and vertical clarity analysis on the intermediate frequency images of the filtered face positioning image and filtered hand positioning image for each frame to obtain horizontal clarity analysis data and vertical clarity analysis data, and determine the second clarity based on the horizontal clarity analysis data and vertical clarity analysis data. Determine the target clarity of the face positioning image and hand positioning image for each frame based on the first clarity and the second clarity; S206: Determine whether the target clarity is greater than or equal to the preset clarity threshold; S207: Use the face positioning image and hand positioning image for each frame as the corresponding several frames of driver face images and driver hand images; S208: Optimize the clarity of the face positioning image and hand positioning image for each frame to obtain the corresponding several frames of driver face images and driver hand images; S209: Analyze the key point position relationship of each frame of driver face image and driver hand image, and identify driving abnormal behaviors based on the key point position relationship in combination with the abnormal behavior discrimination model based on the fusion attention mechanism; S210: Analyze the environmental road conditions based on the visibility analysis in combination with several frames of road section images collected by a plurality of information collection devices to obtain environmental road condition information; S211: Analyze the road section congestion based on several frames of road section images to obtain road section congestion information; S212: Determine the corresponding road condition safety information based on the driving speed, driving abnormal behaviors, environmental road condition information, and road section congestion information. The information release device issues a road condition warning to the vehicle terminal based on the road condition safety information.

[0035] In an embodiment of the present invention, each frame of driver image is input into a target positioning model to obtain corresponding facial positioning images and hand positioning images for each frame. By analyzing the target clarity of each frame of facial positioning image and hand positioning image to optimize their clarity, clearer driver facial images and hand images can be obtained, improving the accuracy of analyzing abnormal driving behaviors. Analyzing the key point position relationships of each frame of driver facial image and driver hand image, and identifying abnormal driving behaviors based on the key point position relationships in combination with an abnormal behavior discrimination model based on a fusion attention mechanism can avoid consuming excessive computing resources, greatly reducing the missed detection rate of abnormal behaviors and effectively improving the reliability of identifying abnormal driving behaviors. Analyzing the environmental road conditions by combining visibility analysis with several frames of road section images collected by several information collection devices makes the analysis of environmental road conditions more comprehensive. Conducting road section congestion analysis based on several frames of road section images, and determining corresponding road condition safety information based on driving speed, abnormal driving behaviors, environmental road condition information, and road section congestion information. In this way, sufficiently comprehensive and accurate road condition information can be obtained, more accurately analyzing the current road safety situation, and thus being able to issue corresponding warnings to the vehicle terminal in a timely and accurate manner, reducing the occurrence of accidents.

[0036] Embodiment III

[0037] Please refer to Figure 3 , Figure 3 which is a schematic structural composition diagram of a road condition safety analysis system based on vehicle-road collaboration in an embodiment of the present invention. The system includes several information collection devices 31 and an information publishing device 32, and the information collection devices 31 are communicatively connected to the information publishing device 32; the system is configured to execute the road condition safety analysis method based on vehicle-road collaboration described in the above embodiment.

[0038] In the specific implementation process of the present invention, the information collection device 31 includes a stroboscopic lamp, a camera, a radar, a roadside unit, a flasher, an equipment box, and a roadside pole. The equipment box houses a network, a power supply, and an edge computing and processing device. The camera is used to collect road section images and driver images. The stroboscopic lamp and the flasher provide supplementary lighting for image collection. The radar is used to collect vehicle speed information for supplementation. The roadside unit can communicate with the vehicle terminal, and after collecting relevant information, it is transmitted to the edge computing and processing device in the equipment box for relevant data processing and analysis. The information publishing device can freely combine a stroboscopic lamp, a camera, a radar, a flasher, an equipment box, a roadside unit, a display screen, and a roadside pole. The equipment box houses a network and a power supply device. For example, information publishing device A and information publishing device B can only use a display screen, and information publishing device C can use a stroboscopic lamp, a camera, a radar, a flasher, an equipment box, a roadside unit, a display screen, and a roadside pole. The equipment box houses a network and a power supply device. After the information collection device 31 collects data and performs relevant data analysis, relevant road condition information is obtained and transmitted to the information publishing device 32. The information publishing device 32 provides a safety warning to the vehicle terminal. At the same time, the information collection device and the information publishing device can also be combined into one whole to form an information collection and publishing device. The applications of the information collection device and the information publishing device can be as Figure 5 shown. The applications of the information collection device and the information publishing device can be like information collection device A, information collection device B, information collection and publishing device C, and information collection device D collecting information for data analysis, and transmitting the analysis results to information publishing device A, information publishing device B, and information collection and publishing device C for display. At the same time, the information publishing device provides a safety warning to the vehicle terminal. Figure 3 The system shown does not constitute a limitation on all components, and may include more or fewer components than shown, or combine certain components.

[0039] In the specific implementation process of the present invention, the specific implementation manners of the system items can refer to the above embodiments and will not be elaborated here.

[0040] In the embodiments of the present invention, each frame of driver image is input into a target positioning model to obtain corresponding facial positioning images and hand positioning images for each frame. By analyzing the target clarity of each frame of facial positioning image and hand positioning image to optimize their clarity, clearer driver facial images and hand images can be obtained, improving the accuracy of driving abnormal behavior analysis. Analyzing the key point position relationship between each frame of driver facial image and driver hand image, and combining the key point position relationship with an abnormal behavior discrimination model based on a fusion attention mechanism to identify driving abnormal behavior can avoid consuming excessive computing resources, greatly reducing the missed detection rate of abnormal behavior and effectively improving the reliability of driving abnormal behavior recognition. Based on the analysis of the environmental road conditions by combining the visibility analysis with several frames of road section images collected by several information collection devices, the analysis of the environmental road conditions is more comprehensive. Based on several frames of road section images, road section congestion analysis is carried out, and corresponding road condition safety information is determined based on the driving speed, driving abnormal behavior, environmental road condition information, and road section congestion information. In this way, sufficiently comprehensive and accurate road condition information can be obtained, and the current road safety situation can be analyzed more accurately, so that corresponding warnings can be sent to the vehicle terminal in a timely and accurate manner, reducing the occurrence of accidents.

[0041] Embodiment 4

[0042] Please refer to Figure 4 , Figure 4 which is a schematic structural diagram of a road condition safety analysis device based on vehicle-road collaboration in the embodiments of the present invention. The device is applied to several information collection devices and information publishing devices, and the information collection devices are communicatively connected to the information publishing devices; the device includes: Speed analysis module 41: configured to perform speed analysis of the current vehicle based on the echo signal collected by the information collection device to obtain the corresponding driving speed; Facial and hand positioning module 42: configured to perform facial positioning, hand positioning, and clarity optimization on several frames of driver images of the current vehicle collected by several information collection devices to obtain corresponding several frames of driver facial images and driver hand images; Abnormal behavior analysis module 43: configured to analyze the key point position relationship between each frame of driver facial image and driver hand image, and combine the key point position relationship with an abnormal behavior discrimination model based on a fusion attention mechanism to identify driving abnormal behavior; Environmental road condition analysis module 44: configured to perform environmental road condition analysis by combining visibility analysis with several frames of road section images collected by several information collection devices to obtain environmental road condition information; Congestion analysis module 45: configured to perform road section congestion analysis based on several frames of road section images to obtain road section congestion information; Safety warning module 46: It is used to determine corresponding road condition safety information based on the driving speed, abnormal driving behavior, environmental road condition information, and road section congestion information. The information publishing device gives a road condition warning to the vehicle terminal based on the road condition safety information.

[0043] In the specific implementation process of the present invention, for the specific implementation manner of the device item, reference can be made to the above-mentioned embodiments, which will not be elaborated here.

[0044] In the embodiment of the present invention, each frame of driver image is input into the target positioning model to obtain the corresponding facial positioning image and hand positioning image of each frame. By analyzing the target clarity of each frame of facial positioning image and hand positioning image to optimize their clarity, a clearer driver facial image and hand image can be obtained, improving the accuracy of abnormal driving behavior analysis. Analyzing the key point position relationship of each frame of driver facial image and driver hand image, and combining the abnormal behavior discrimination model based on the fusion attention mechanism based on the key point position relationship to identify abnormal driving behavior can avoid consuming too much computing resources, greatly reduce the missed detection rate of abnormal behavior, and effectively improve the reliability of abnormal driving behavior recognition. Based on the combination of visibility analysis with several frames of road section images collected by several information collection devices for environmental road condition analysis, the environmental road condition analysis is more comprehensive. Based on several frames of road section images for road section congestion analysis, and determining the corresponding road condition safety information based on the driving speed, abnormal driving behavior, environmental road condition information, and road section congestion information, so as to obtain sufficiently comprehensive and accurate road condition information, more accurately analyze the current road safety situation, and thus be able to issue corresponding warnings to the vehicle terminal in a timely and accurate manner, reducing the occurrence of accidents.

[0045] A computer-readable storage medium provided by an embodiment of the present invention, on which a computer program is stored. When the program is executed by a processor, it implements the road condition safety analysis method based on vehicle-road collaboration in any one of the above-mentioned embodiments. Among them, the computer-readable storage medium includes, but is not limited to, any type of disk (including floppy disk, hard disk, optical disk, CD-ROM, and magneto-optical disk), ROM (Read-Only Memory), RAM (Random Access Memory), EPROM (Erasable Programmable Read-Only Memory), EEPROM (Electrically Erasable Programmable Read-Only Memory), flash memory, magnetic card, or optical card. That is, the storage device includes any medium that can store or transmit information in a readable form by a device (such as a computer, mobile phone), and can be a read-only memory, a magnetic disk, or an optical disk, etc.

[0046] In addition, the above has introduced in detail a road condition safety analysis method and related devices provided by embodiments of the present invention. Specific examples should have been used in this article to elaborate on the principle and implementation manner of the present invention. The description of the above embodiments is only used to help understand the method and its core idea of the present invention. At the same time, for those of ordinary skill in the art, according to the idea of the present invention, there will be changes in the specific implementation manner and application scope. In summary, the content of this specification should not be construed as a limitation to the present invention.

Claims

1. A road condition safety analysis method based on vehicle-road collaboration, characterized in that, Applied to a number of information collection devices and information publishing devices, the information collection devices are communicatively connected to the information publishing devices; the method includes: Performing speed analysis of the current vehicle based on the echo signals collected by the information collection devices to obtain the corresponding driving speed; Performing face localization, hand localization, and clarity optimization on several frames of driver images of the current vehicle collected by several information collection devices to obtain the corresponding several frames of driver face images and driver hand images; Analyzing the key point position relationships of each frame of driver face images and driver hand images, and identifying driving abnormal behaviors based on the key point position relationships in combination with an abnormal behavior discrimination model based on a fusion attention mechanism; Performing environmental road condition analysis based on several frames of road section images collected by several information collection devices in combination with visibility analysis to obtain environmental road condition information; Performing road section congestion analysis based on several frames of road section images to obtain road section congestion information; Determining the corresponding road condition safety information based on the driving speed, driving abnormal behaviors, environmental road condition information, and road section congestion information, and the information publishing device performing road condition early warning to the vehicle terminal based on the road condition safety information.

2. The method for analyzing road condition safety based on vehicle-road cooperation according to claim 1, wherein The performing speed analysis of the current vehicle based on the echo signals collected by the information collection devices to obtain the corresponding driving speed includes: Mixing the echo signals with the local oscillator signals to obtain intermediate frequency signals, and determining the corresponding speed spectrum and spectrum position based on the peak spectral lines of the intermediate frequency signals; Determining the corresponding driving speed of the current vehicle based on the speed spectrum and spectrum position in combination with the radial speed of the radar in the information collection device.

3. The traffic condition safety analysis method based on vehicle-road cooperation according to claim 1, wherein The performing face localization, hand localization, and clarity optimization on several frames of driver images of the current vehicle collected by several information collection devices to obtain the corresponding several frames of driver face images and driver hand images includes: Inputting each frame of driver image into the target localization model for face localization and hand localization to obtain the corresponding face localization images and hand localization images of each frame; Determining the first clarity corresponding to the face localization images and hand localization images of each frame based on the contrast and noise values of each pixel point in the face localization images and hand localization images of each frame; Filtering the face localization images and hand localization images of each frame with filters of different filter coefficients to obtain the filtered face localization images and filtered hand localization images corresponding to different filter coefficients; Performing horizontal clarity analysis and vertical clarity analysis on the intermediate frequency images of the filtered face localization images and filtered hand localization images of each frame to obtain horizontal clarity analysis data and vertical clarity analysis data, and determining the second clarity based on the horizontal clarity analysis data and vertical clarity analysis data; Determining the target clarity of the face localization images and hand localization images of each frame based on the first clarity and the second clarity, and performing clarity optimization on the face localization images and hand localization images of each frame based on the comparison result between the target clarity and the preset clarity threshold to obtain the corresponding several frames of driver face images and driver hand images.

4. The traffic condition safety analysis method based on vehicle-road cooperation according to claim 1, characterized in that Analyzing the key point position relationship between the driver's facial images and hand images in each frame, and identifying driving abnormal behaviors based on the key point position relationship in combination with an abnormal behavior discrimination model based on a fusion attention mechanism, including: Analyzing the eye key points, ear key points, and lip key points of the driver's facial images in each frame and the hand key points of the driver's hand images in each frame, and determining the first key point position distance between the ear key points and the hand key points and the second key point position distance between the lip key points and the hand key points; Analyzing the first distracted driving behavior of the driver based on the first key point position distance and the second key point position distance, and analyzing the second distracted driving behavior of the driver based on the eye key points of the driver's facial images in each frame; Analyzing the fatigue driving behavior of the driver based on the driver's facial images and eye key points in each frame using an abnormal behavior discrimination model based on a fusion attention mechanism.

5. The method for analyzing road condition safety based on vehicle-road cooperation according to claim 1, wherein The environmental road condition analysis based on several frames of road section images collected by several information collection devices in combination with visibility analysis to obtain environmental road condition information, including: Performing grayscale processing on each frame of road section image to obtain each frame of grayscale road section image, and extracting texture features, pixel brightness features, and color features based on each frame of grayscale road section image to obtain texture feature information, pixel brightness feature information, and color feature information; Identifying the road section icing area based on the texture feature information, pixel brightness feature information, and color feature information to obtain road section icing area information; Determining the road section fog concentration based on each frame of road section image using a fog detection model, and determining the road section visibility based on the road section fog concentration. Generating environmental road condition information based on the road section icing area information and the road section visibility.

6. The traffic condition safety analysis method based on vehicle-road cooperation according to claim 1, wherein, The road section congestion analysis based on several frames of road section images to obtain road section congestion information, including: Calculating the total vehicle area based on several frames of road section images, and determining the congested road sections based on the ratio of the total vehicle area to the road section area; Dividing the congested road sections into several grid lane road sections, and counting the vehicle density of each grid lane road section; Determining the road section congestion information based on the vehicle density of each grid lane road section in combination with the statistics of lane-changing vehicles.

7. The method for analyzing road condition safety based on vehicle-road collaboration according to claim 1, characterized in that, Determining the corresponding road condition safety information based on the driving speed, driving abnormal behaviors, environmental road condition information, and road section congestion information, and the information publishing device giving a road condition warning to the vehicle terminal based on the road condition safety information, including: Comparing the driving speed with a preset speed limit value, and determining vehicle speeding information based on the comparison result; Determining the corresponding road condition safety information based on the vehicle speeding information, driving abnormal behaviors, environmental road condition information, and road section congestion information, and matching the corresponding warning level based on the road condition safety information; The information publishing device gives a road condition warning to the vehicle terminal based on the road condition safety information and the corresponding warning level.

8. A road condition safety analysis device based on vehicle-road cooperation, characterized in that, Applied to several information collection devices and information publishing devices, the information collection devices are communicatively connected to the information publishing devices; the device includes: A speed analysis module: for performing speed analysis of the current vehicle based on the echo signal collected by the information collection device to obtain the corresponding driving speed; Facial and Hand Localization Module: It is used to perform facial localization, hand localization, and clarity optimization on several frames of driver images of the current vehicle collected by several information collection devices, and obtain corresponding several frames of driver facial images and driver hand images; Abnormal Behavior Analysis Module: It is used to analyze the key point position relationships of each frame of driver facial images and driver hand images, and identify driving abnormal behaviors based on the key point position relationships in combination with an abnormal behavior discrimination model based on a fusion attention mechanism; Environmental Road Conditions Analysis Module: It is used to perform environmental road conditions analysis based on several frames of road section images collected by several information collection devices in combination with visibility analysis, and obtain environmental road conditions information; Congestion Analysis Module: It is used to perform road section congestion analysis based on several frames of road section images, and obtain road section congestion information; Safety Warning Module: It is used to determine corresponding road conditions safety information based on the driving speed, driving abnormal behaviors, environmental road conditions information, and road section congestion information, and the information publishing device issues road conditions warnings to the vehicle terminal based on the road conditions safety information.

9. A road condition safety analysis system based on vehicle-road cooperation, characterized in that, The system includes several information collection devices and an information publishing device. The information collection devices are communicatively connected to the information publishing device. The system is configured to execute the road conditions safety analysis method based on vehicle-road collaboration described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions. When the computer instructions run on an electronic device, the electronic device is caused to execute the road conditions safety analysis method based on vehicle-road collaboration described in any one of claims 1 to 7.

Images