WO2018103313A1 - 发生交通事故的风险预测方法、装置及系统 - Google Patents

发生交通事故的风险预测方法、装置及系统 Download PDF

Info

Publication number
WO2018103313A1
WO2018103313A1 PCT/CN2017/090874 CN2017090874W WO2018103313A1 WO 2018103313 A1 WO2018103313 A1 WO 2018103313A1 CN 2017090874 W CN2017090874 W CN 2017090874W WO 2018103313 A1 WO2018103313 A1 WO 2018103313A1
Authority
WO
WIPO (PCT)
Prior art keywords
information
traffic accident
vehicle
vehicles
target road
Prior art date
Application number
PCT/CN2017/090874
Other languages
English (en)
French (fr)
Inventor
蒋瑜
Original Assignee
杭州海康威视数字技术股份有限公司
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 杭州海康威视数字技术股份有限公司 filed Critical 杭州海康威视数字技术股份有限公司
Publication of WO2018103313A1 publication Critical patent/WO2018103313A1/zh

Links

Images

Classifications

    • GPHYSICS
    • G08SIGNALLING
    • G08GTRAFFIC CONTROL SYSTEMS
    • G08G1/00Traffic control systems for road vehicles
    • G08G1/01Detecting movement of traffic to be counted or controlled
    • G08G1/0104Measuring and analyzing of parameters relative to traffic conditions
    • G08G1/0125Traffic data processing
    • G08G1/0129Traffic data processing for creating historical data or processing based on historical data
    • GPHYSICS
    • G08SIGNALLING
    • G08GTRAFFIC CONTROL SYSTEMS
    • G08G1/00Traffic control systems for road vehicles
    • G08G1/01Detecting movement of traffic to be counted or controlled

Definitions

  • the present application relates to the field of intelligent traffic management and control technologies, and in particular, to a risk prediction method, device and system for occurrence of a traffic accident.
  • the method for predicting road accident risk on the target road section is specifically: collecting information on the road accident risk, such as people, vehicles, roads and environment on the target road section in real time; and predicting the risk state of the road accident of the target road section according to the collected information.
  • the prior art method for predicting road accident risk is to predict the accident risk based on the road risk of the target road segment, and does not take into account the difference in vehicle conditions of the vehicles on the target road segment, for example, for an old vehicle that has been used for 20 years and is to be repaired.
  • the probability of an accident with the vehicle and a new vehicle that has just passed the factory and is in compliance with the standard is the same.
  • the above method for predicting road accident risk does not predict the specific vehicle risk status.
  • the purpose of the embodiments of the present application is to provide a risk prediction method, device and system for a traffic accident, so as to realize a risk prediction of a traffic accident in a vehicle and improve the accuracy of a road accident risk prediction.
  • the specific technical solutions are as follows:
  • an embodiment of the present application provides a method for predicting a risk of a traffic accident in a vehicle, including:
  • the video structured information includes at least license plate number information of each first vehicle on the target road segment;
  • a risk factor of a traffic accident in the vehicle includes: a traffic accident occurrence probability value and/or a severity value of the occurrence of the traffic accident, the historical information including: a traffic accident occurrence in the history of the target road segment Vehicle condition information and historical behavior information of the two vehicles, and environmental condition information when the traffic accident occurs;
  • the risk factor is determined as a risk prediction result of a traffic accident of the vehicle of the target road segment.
  • the embodiment of the present application provides a risk prediction method for a road accident, including:
  • the video structured information includes at least license plate number information of each first vehicle on the target road segment;
  • the first information Determining, for each of the first vehicles, the first information according to the historical information of the traffic accident occurring in the target road segment, the vehicle condition information and the historical behavior information of each of the first vehicles, and the environmental condition information.
  • a traffic accident occurrence probability value of a traffic accident of the vehicle includes: vehicle condition information and historical behavior information of the second vehicle of a traffic accident occurring in the history of the target road section, and an environment when the traffic accident occurs Status information;
  • Counting information of each first vehicle on the target road segment obtaining average vehicle condition information, average historical behavior information, average value of traffic accident occurrence probability values of occurrence of traffic accidents on the target road section, and Mean current driver behavior information and road condition information on the target road segment;
  • the information of each of the first vehicles includes: the video structure information, the vehicle condition information and the historical behavior information of each of the first vehicles, a traffic accident occurrence probability value of each of the first vehicles in which a traffic accident occurs;
  • the first occurrence probability value is determined as a risk prediction result of a road accident occurring on the road of the target road section.
  • the embodiment of the present application provides a risk prediction device for a traffic accident in a vehicle, including:
  • a first obtaining unit configured to obtain a monitoring video on the target road segment
  • a processing unit configured to obtain video structured information according to the monitoring video;
  • the video structured information includes at least license plate number information of each first vehicle on the target road segment;
  • a searching unit configured to query, according to the license plate number information, vehicle condition information and historical behavior information of each first vehicle on the target road segment in a preset database
  • a second obtaining unit configured to acquire environmental condition information of the target road segment when the monitoring video is captured
  • a first determining unit configured to perform, according to the target road segment, for each of the first vehicles
  • the risk factors include: The accident occurrence probability value and/or the severity value of the occurrence of the traffic accident
  • the historical information includes: vehicle condition information and historical behavior information of the second vehicle of a traffic accident occurring in the history of the target road section, and the occurrence of the traffic Environmental status information at the time of the accident;
  • a second determining unit configured to determine the risk factor as a risk prediction result of a traffic accident of the vehicle of the target road segment.
  • the embodiment of the present application provides a risk prediction device for a road accident, including:
  • a first obtaining unit configured to obtain a monitoring video on the target road segment
  • a processing unit configured to obtain video structured information according to the monitoring video;
  • the video structured information includes at least license plate number information of each first vehicle on the target road segment;
  • a searching unit configured to query, according to the license plate number information, vehicle condition information and historical behavior information of each first vehicle on the target road segment in a preset database
  • a second obtaining unit configured to acquire environmental condition information of the target road segment when the monitoring video is captured
  • a first determining unit configured, for each of the first vehicles, historical information of a traffic accident according to the target road segment, vehicle condition information and historical behavior information of each of the first vehicles, and the environmental condition information, Determining a traffic accident occurrence probability value of each first vehicle in which a traffic accident occurs;
  • the historical information includes: vehicle condition information and historical behavior information of the second vehicle of a traffic accident occurring in the history of the target road section, and the occurrence of the species Environmental status information at the time of traffic accidents;
  • a third obtaining unit configured to perform statistics on information of each first vehicle on the target road segment, obtain average vehicle condition information, average historical behavior information, and traffic accident occurrence of all vehicles on the target road segment An average of the probability values, and average current driver behavior information and road information on the target road segment;
  • the information of each of the first vehicles includes: the video structured information, the vehicle condition of each of the first vehicles Information and historical behavior information, a probability of occurrence of a traffic accident in which each of the first vehicles has a traffic accident;
  • An input unit configured to obtain average vehicle condition information, average historical behavior information, average value of traffic accident occurrence probability values of traffic accidents, and average current driver behavior information and information of all vehicles on the target road segment obtained
  • the road condition information on the target road segment is input into a preset prediction model, which is a traffic accident that occurs in advance for history, and all vehicles on the target road segment for each occurrence of the traffic accident Average vehicle condition information, average historical behavior information, average value of traffic accident occurrence probability of occurrence of traffic accidents, and average current driver behavior information and road condition information on the target road section, and environmental conditions at the time of occurrence of the traffic accident Information obtained through training;
  • a fourth obtaining unit configured to obtain a first occurrence probability value of a road occurrence traffic accident output by the prediction model
  • a second determining unit configured to determine the first occurrence probability value as a risk prediction result of a road accident occurring on the road of the target road segment.
  • an embodiment of the present application provides a risk prediction system for a vehicle accident, including:
  • a video monitoring device for collecting monitoring video on a target road segment
  • a vehicle information database configured to store vehicle condition information and historical behavior information of each first vehicle on the target road segment
  • An intelligent analysis server configured to obtain a monitoring video on a target road segment collected by the video monitoring device; and obtain video structured information according to the monitoring video; the video structured information includes at least each first on the target road segment Vehicle license plate number information; querying, according to the license plate number information, vehicle condition information and historical behavior information of each first vehicle on the target road segment in the vehicle information database; acquiring a target road segment in an external information source system Environmental condition information when the video is monitored;
  • each of the first vehicles Determining each of the first vehicles according to historical information of a traffic accident occurring in the target road segment, vehicle condition information and historical behavior information of each of the first vehicles, and the environmental condition information a risk factor of a traffic accident in a vehicle;
  • the risk factor includes: a traffic accident occurrence probability value and/or a severity value of the occurrence of the traffic accident, the historical information including: a traffic accident occurring in the history of the target road segment Vehicle condition information and historical behavior letter of the second vehicle Interest information, as well as information on the environmental conditions in the event of such a traffic accident;
  • the risk factor is determined as a risk prediction result of a traffic accident of the vehicle of the target road segment.
  • an embodiment of the present application provides a risk prediction system for a road accident, including:
  • a video monitoring device for collecting monitoring video on a target road segment
  • a vehicle information database configured to store vehicle condition information and historical behavior information of each first vehicle on the target road segment
  • An intelligent analysis server configured to obtain a monitoring video on the target road segment collected by the video monitoring device; obtain video structured information according to the monitoring video; and the video structured information includes at least each of the target road segments Vehicle license plate number information of the first vehicle; querying, according to the license plate number information, vehicle condition information and historical behavior information of each first vehicle on the target road segment in the vehicle information database; acquiring a target road segment in an external information source system Environmental condition information when the surveillance video is captured;
  • the first information Determining, for each of the first vehicles, the first information according to the historical information of the traffic accident occurring in the target road segment, the vehicle condition information and the historical behavior information of each of the first vehicles, and the environmental condition information.
  • a traffic accident occurrence probability value of a traffic accident of the vehicle includes: vehicle condition information and historical behavior information of the second vehicle of a traffic accident occurring in the history of the target road section, and an environment when the traffic accident occurs Status information;
  • Counting information of each first vehicle on the target road segment obtaining average vehicle condition information, average historical behavior information, average value of traffic accident occurrence probability values of occurrence of traffic accidents on the target road section, and Mean current driver behavior information and road condition information on the target road segment;
  • the information of each of the first vehicles includes: the video structure information, the vehicle condition information of each of the first vehicles, each of the Historical behavior information of the first vehicle, a probability of occurrence of a traffic accident in which each of the first vehicles has a traffic accident;
  • the prediction The model is an average of the average vehicle condition information, the average historical behavior information, and the traffic accident probability value of all the vehicles on the target road segment in the occurrence of the traffic accident.
  • the value, and the average current driver behavior information and the road condition information on the target road segment, and the environmental condition information when the traffic accident occurs are obtained by training;
  • the present application provides a storage medium, wherein the storage medium is configured to store executable program code for performing a vehicle occurrence as described in the first aspect above at runtime A method for predicting a risk of a traffic accident and a method for predicting a risk of a road accident caused by the second aspect.
  • the present application provides an application, wherein the application is configured to perform, at runtime, a risk prediction method for a vehicle accident caused by the first aspect, and the second aspect A risk prediction method for traffic accidents on roads.
  • the application provides an electronic device, including:
  • processor a memory, a communication interface, and a bus
  • the processor, the memory, and the communication interface are connected by the bus and complete communication with each other;
  • the memory stores executable program code
  • the processor runs a program corresponding to the executable program code by reading executable program code stored in the memory for performing a risk of a traffic accident of a vehicle as described in the first aspect above
  • the prediction method and the risk prediction method for road traffic accidents described in the second aspect are described in the first aspect above.
  • a method, device and system for predicting a risk of a traffic accident provided by an embodiment of the present application, the method comprising: obtaining a monitoring video on a target road segment; obtaining video structured information according to the monitoring video; Information on the license plate number in the information, querying the vehicle condition information and historical behavior information of each first vehicle on the target road segment in the preset database; acquiring the target road Environmental condition information when the monitoring video is captured; for each of the first vehicles, historical information of a traffic accident according to the target road segment, vehicle condition information and historical behavior information of each of the first vehicles, and The environmental condition information is used to determine a risk factor of a traffic accident in each of the first vehicles; and determining the risk factor as a risk prediction result of a traffic accident of the vehicle in the target road segment, which is implemented by applying the embodiment of the present application. Risk prediction of traffic accidents in vehicles.
  • the application of the embodiment of the present application improves the accuracy of the road accident risk; at the same time, the embodiment of the present application does not need to install additional equipment on each vehicle, and the transformation cost is low.
  • implementing any of the products or methods of the present application necessarily does not necessarily require all of the advantages described above to be achieved at the same time.
  • 1a is a flow chart of a method for predicting a risk of a traffic accident in a vehicle according to an embodiment of the present application
  • FIG. 1b is another flowchart of a method for predicting a traffic accident of a vehicle according to an embodiment of the present application
  • 2a is a flow chart of a method for predicting a risk of road traffic accidents according to an embodiment of the present application
  • 2b is another flow chart of a method for predicting a risk of road traffic accidents according to an embodiment of the present application
  • 3a is a structural diagram of a device for predicting a traffic accident of a vehicle according to an embodiment of the present application
  • FIG. 3b is another structural diagram of a risk prediction device for a vehicle accident caused by an embodiment of the present application.
  • 4a is a structural diagram of a risk prediction device for a road accident caused by an embodiment of the present application
  • 4b is another structural diagram of a risk prediction device for road traffic accidents according to an embodiment of the present application.
  • FIG. 5 is a structural diagram of a risk prediction system for a vehicle accident caused by an embodiment of the present application.
  • FIG. 5b is another structural diagram of a risk prediction system for a vehicle accident caused by an embodiment of the present application.
  • 6a is a structural diagram of a risk prediction system for road traffic accidents according to an embodiment of the present application.
  • FIG. 6b is another structural diagram of a risk prediction system for road traffic accidents according to an embodiment of the present application.
  • the embodiment of the present application provides a risk prediction method, device and system for occurrence of a traffic accident, realizes a risk prediction of a traffic accident of a vehicle, and improves the accuracy of a road accident risk prediction.
  • a method for predicting a traffic accident of a vehicle may include the following steps:
  • each camera is used to monitor the motion state of the moving target in the camera monitoring area, and each camera will return to the monitoring video in the monitored area.
  • each camera is responsible for monitoring the area of the area where the camera is located, and all the cameras on the target road section jointly monitor the motion state of the moving target on the target road section, where the moving target includes the vehicle, but is not limited thereto.
  • the video structure information includes at least license plate number information of each first vehicle on the target road segment.
  • the video monitoring technology is obtained by using the video analysis technology on the returned monitoring video, where the video structured information includes at least each of the target road segments.
  • the license plate number information of the first vehicle, and the video structured information may further include license plate number information, whether the driver is wearing seat belt information, whether the driver calls mobile phone information, vehicle model information, vehicle color information, and the like.
  • each car on the target road segment corresponds to a video structured information.
  • Table 1 shows the video structured information of 5 cars, and the video structured information includes: license plate information, whether the driver is wearing seat belt information, whether the driver is calling mobile phone information, vehicle model information, vehicle Color information, but not limited to this.
  • the first vehicle includes all of the vehicles traveling on the target road segment, and each of the first vehicles refers to one of all vehicles on the target road segment.
  • the vehicle condition information and the historical behavior information of each first vehicle corresponding to the license plate number information are queried in the preset database. It can be understood that the license plate number information of each vehicle is in advance.
  • the vehicle condition information includes: vehicle annual inspection information (for example, the number of vehicles not detected annually), vehicle age, etc.
  • Historical behavior information includes: number of red light, speed of overspeed, violation of regulations Change the road and so on.
  • the preset database here may include a vehicle information database of the traffic police system.
  • the environmental condition information of the target road section may include: weather (yellow, rain, snow, etc.) when the monitoring video is acquired, whether there is a curve in the target road section, whether there is a slope, visibility, and the like. Because different environmental conditions have a certain impact on traffic accidents in vehicles. For example, in the case of low visibility, when the current car accident stops, the driver of the rear car may not see the driving condition of the preceding car, and no corresponding measures are taken. , causing traffic accidents. Therefore, when predicting the risk of a traffic accident in a vehicle, it is necessary to consider the environmental condition information of the vehicle at the target road section.
  • each of the first vehicles determine each of the first vehicles according to historical information of a traffic accident occurring in the target road segment, vehicle condition information and historical behavior information of each of the first vehicles, and the environmental condition information.
  • a risk factor of a traffic accident occurring in the first vehicle includes: a traffic accident occurrence probability value and/or a severity value of the occurrence of the traffic accident, the historical information including: a traffic accident occurring in the history of the target road segment Vehicle condition information and historical behavior information of the second vehicle, and environmental condition information when the traffic accident occurs;
  • each of the first vehicles on the target road segment corresponds to the vehicle condition information and the historical behavior information, and it can be understood that the environmental condition information of all the vehicles on the target road segment may be the same.
  • determining the risk factor of the traffic accident of each first vehicle in combination with the historical information of the traffic accident occurring on the target road segment after obtaining the vehicle condition information and the historical behavior information of each first vehicle and the environmental condition information, determining the risk factor of the traffic accident of each first vehicle in combination with the historical information of the traffic accident occurring on the target road segment.
  • the risk factors for traffic accidents in each of the first vehicles include: The probability of occurrence of the traffic accident and/or the severity of the occurrence of the traffic accident, that is, the historical information of the traffic accident according to the target road section, the vehicle condition information and historical behavior information of each of the first vehicles, and the environmental condition information Determining a traffic accident occurrence probability value of each first vehicle in which a traffic accident occurs, or historical information of a traffic accident occurring according to the target road section, vehicle condition information and historical behavior information of each of the first vehicles, and the environmental condition Information to determine the severity of a traffic accident in each of the first vehicles.
  • the risk factors include: the probability of occurrence of a traffic accident and/or the severity of a traffic accident, where the probability of occurrence of the traffic accident and/or the severity of the occurrence of the traffic accident are determined as the risk prediction result.
  • the video structured information is obtained through the monitoring video on the obtained target road segment, and the first license information of each first vehicle on the target road segment is queried in the preset database according to the license plate number information in the video structured information.
  • Vehicle condition information and historical behavior information at the same time, obtaining environmental condition information when the monitoring video is captured, determining the vehicle status information, historical behavior information, environmental condition information, and historical information of traffic accidents on the target road segment, and determining each The risk factor of the traffic accident of the first vehicle can obtain the risk prediction result of the vehicle.
  • a method for predicting a traffic accident of a vehicle may include the following steps:
  • the video structure information includes at least license plate number information of each first vehicle on the target road segment.
  • the steps S201 to S204 are the same as the steps S101 to S104 shown in FIG. 1a, and are not described herein again.
  • S205 Receive a speed of each first vehicle on the target road segment; obtain a road panorama of the target road segment according to the monitoring video; and obtain each of the first vehicles according to the road panoramic image a relative distance of the neighboring vehicles; obtaining a relative speed of each of the first vehicles and their neighboring vehicles according to the road panorama and the speed of each of the first vehicles;
  • the speed of each first vehicle on the target road segment may be different.
  • the speed of each first vehicle is obtained by using a sensor; the sensor may include a radar speed measuring device, but is not limited thereto.
  • a road panorama on the target road segment is constructed, which is used to indicate the relative positional relationship of each first vehicle when the surveillance video on the target road segment obtained by the shooting is captured. Based on the relative positional relationship of each of the first vehicles in the road panorama, the relative distance of each of the first vehicles and the adjacent vehicles is obtained. At the same time, based on the road panorama and the speed of each of the first vehicles, the relative speed of each of the first vehicles and their neighboring vehicles is calculated.
  • the vehicles adjacent to the A car include: B car, C car, the speed of the A car measured by the sensor is 50 km/s, the speed of the B car is 45 km/s, and the speed of the C car is 40km/s, the relative speed of A and B is 5km/s, and the relative speed of A and C is 10km/s.
  • each of the first vehicles determine each of the first vehicles according to historical information of a traffic accident occurring in the target road segment, vehicle condition information and historical behavior information of each of the first vehicles, and the environmental condition information.
  • the initial risk factor for a traffic accident in the first vehicle
  • the video structured information further includes: current driver behavior information of each of the first vehicles;
  • the historical information further includes: current driver behavior information of the second vehicle in each of the traffic accidents occurring historically on the target road segment, and the current driver behavior information of the second vehicle is relative to the behavior of the driver in the event of a historical traffic accident. information.
  • the current driver behavior information of each first vehicle refers to that when the monitoring video on the target road segment is obtained, the video structure information obtained according to the monitoring video further includes current driver behavior information, and the current driver behavior information is as follows.
  • Table 1 shows: whether the driver is wearing seat belt information, whether the driver is using cell phone information, etc., but is not limited thereto, and may include: information on whether the driver smokes, information on whether the driver eats or the like.
  • the step of describing an initial risk factor of a traffic accident for each first vehicle may include: for each of the first vehicles, historical information of a traffic accident according to the target road segment, and vehicle condition information of each of the first vehicles And historical behavior information, and the environmental condition information and the current driver behavior information of each of the first vehicles, determining an initial risk factor of the traffic accident of each of the first vehicles;
  • the current driver behavior information of each first vehicle needs to be considered. Because the current driver behavior information may be a major factor directly causing traffic accidents, for example, when the driver drives while driving, it is easy to cause the driver to lose focus, and does not notice the sudden braking of other neighboring vehicles or the traffic of the road. The situation leads to a traffic accident, so when determining the initial risk factor of a traffic accident, the current driver behavior information should be combined.
  • the history information is relative to the vehicle condition information, the historical behavior information, and the environmental condition information corresponding to each of the first vehicles when the monitoring video is captured, that is, the specific information of the historical information, when a traffic accident occurs in the history on the target road section.
  • Vehicle condition information corresponding to the second vehicle annual inspection status of the second vehicle in the event of a traffic accident, age of the vehicle, etc.
  • historical behavior information number of red light in the second vehicle in the event of a traffic accident, number of speeding, violation of the lane, etc.
  • environmental status information weather, visibility, etc. on the target road section of the second vehicle in the event of an accident.
  • the initial risk factor for a traffic accident for each of the first vehicles obtained herein does not take into account the impact of the vehicle adjacent to each first vehicle on the first vehicle.
  • the initial risk factor of each first vehicle occurrence of the traffic accident is corrected, and the final risk factor of the first vehicle occurrence of the traffic accident is obtained.
  • the relative distance and relative speed of the first vehicle and its neighboring vehicles are considered.
  • the initial factor of each first vehicle occurrence of a traffic accident is corrected, and the final risk of a traffic accident of each first vehicle is obtained. factor.
  • the risk prediction of traffic accidents in vehicles is more accurate.
  • S208 The final risk factor of the traffic accident occurring in each of the first vehicles is used as a risk prediction result for predicting a traffic accident of the vehicle.
  • the monitoring video on the obtained target road segment is analyzed, and the video structured information and the road panorama image are obtained.
  • each first query is obtained in the preset database.
  • Vehicle condition information and historical behavior information at the same time, obtaining environmental condition information on the target road section when shooting the surveillance video, obtaining the speed of each first vehicle, and combining the road panorama to determine each first vehicle and its adjacent vehicle Relative distance and relative speed, the obtained vehicle condition information, historical behavior information, current driver behavior information, environmental condition information, relative distance and relative speed are input to the classifier of the historical information training of the traffic accident in the target road section, And causing the classifier to output an initial risk factor of each first vehicle in which a traffic accident occurs, correcting the initial risk factor according to the relative distance and relative speed of each first vehicle and its neighboring vehicle, and obtaining each first vehicle occurrence
  • the ultimate risk factor for traffic accidents using the final risk factor as a predictor The risk of accidents occur predictions vehicles.
  • historical information of a traffic accident based on the target road segment, vehicle condition information and historical behavior information of each of the first vehicles, and the environmental condition information and the The current driver behavior information of each first vehicle determines an initial risk factor of each of the first vehicles in which a traffic accident occurs, including:
  • a stored score table which is a table for scoring the historical information of each second vehicle in which the traffic accident occurs in advance for a traffic accident occurring in history; for each of the a first vehicle, in the rating table, searching for vehicle condition information and historical behavior information of each first vehicle, and a pair of the environmental condition information and current driver behavior information of the first vehicle The score should be scored; according to the corresponding score, the traffic accident occurrence probability value of each of the first vehicles and/or the severity of the occurrence of the traffic accident is obtained as an initial risk factor of the first vehicle occurrence of the traffic accident.
  • a stored score table is obtained, which is a table for scoring the historical information of the second vehicle in which the traffic accident has occurred, that is, according to the score.
  • the score of the history information of the second vehicle in the history of which the traffic accident has occurred, the vehicle condition information and the historical behavior information of each of the first vehicles, and the environmental condition information and the current driver behavior of each of the first vehicles The information is scored, where the vehicle condition information and historical behavior information of each first vehicle, and the environmental condition information and the current driver behavior information of each of the first vehicles have corresponding scores in the score sheet. Specifically, each traffic accident corresponds to a score sheet.
  • the historical information of a traffic accident according to the target road segment, the vehicle condition information and the historical behavior of each of the first vehicles are determined for each of the first vehicles. And determining, by the information, and the environmental condition information and the current driver behavior information of each of the first vehicles, the step of determining an initial risk factor of each of the first vehicles to generate a traffic accident, comprising:
  • each classifier corresponds to a traffic accident occurring in the history of the target road segment; each of the classifiers is pre-targeted to a historical traffic accident, for each occurrence of the traffic accident
  • the vehicle condition information and historical behavior information of the second vehicle, and the environmental condition information when the traffic accident occurs and the current driver behavior information of the second vehicle are trained;
  • the method further includes: digitizing the vehicle condition information and historical behavior information of each of the first vehicles, and the environmental condition information and current driver behavior information of each of the first vehicles, and digitizing the The vehicle condition information and the historical behavior information of each first vehicle, and the environmental condition information and the current driver behavior information are input into a plurality of preset classifiers, so that the traffic accident occurrence probability value and/or output of each classifier is obtained.
  • the severity value of the traffic accident is the initial risk factor of the traffic accident of the first vehicle, wherein each traffic accident corresponds to one type Classifier.
  • an initial risk factor of a traffic accident occurring in the first vehicle is corrected according to a relative distance and a relative speed of the first vehicle and its neighboring vehicles, and each of the first vehicles is obtained.
  • the steps to the ultimate risk factor for a traffic accident include:
  • the ultimate risk factor for the t-th traffic accident in the ith car The initial risk factor for the t-th traffic accident for the i-th car, N i is the number of vehicles adjacent to the i-th car or the number of vehicles with the distance from the i-th car being less than the first preset value, d is The relative distance between the i-th car and the j-th car in the road panorama information, v is the relative speed between the i-th car and the j-th car in the road panorama information, For the relative distance influence factor between the ith car and the jth car, For the relative speed influence factor between the i-th car and the j-th car, X t ⁇ is the influence factor of the type of accident in the preceding vehicle, which causes the type of t-accident in the rear car or the severity of the car accident.
  • the initial risk factor for the t-th traffic accident in the jth car T is the set of accident types, t is the type of accident in the preceding vehicle, and ⁇ is the type of accident in the after-car, a, ⁇ , ⁇ , ⁇ , X t ⁇ respectively
  • the preset constant is any decimal between 0 and 1.
  • ⁇ and ⁇ are adjusted according to actual conditions, and X t ⁇ is obtained based on historical statistical data, for example, the probability of a type of ⁇ accident that causes rear-end collision to occur in the type of t accident in which the preceding vehicle collides with the guardrail.
  • the final risk factor of each first vehicle occurrence of a traffic accident considers the relative distance and relative speed of each first vehicle and its neighboring vehicles compared to the initial risk factor of each first vehicle having a traffic accident. Therefore, it is more accurate to predict the risk factor of a traffic accident in each of the first vehicles. For example, when each first vehicle is far away from its neighboring vehicle, when the current vehicle has an unexpected situation, the rear vehicle driver may have a relatively long time to take measures to reduce or avoid a traffic accident.
  • the method further includes:
  • each traffic accident occurrence probability value of each of the first vehicle occurrence traffic accidents is greater than a preset first threshold; if the determination result is yes, for each of the first greater than the preset first threshold a vehicle, generating an early warning information output; and/or determining whether a severity value of each of the final risk factors of each of the first vehicles causing a traffic accident is greater than a preset second threshold; if the determination result is yes And generating an early warning information output for each first vehicle that is greater than a preset second threshold.
  • the number of first vehicles greater than the preset first threshold may be one or more vehicles, and the number of first vehicles greater than the preset second threshold may be one or more vehicles.
  • a method for predicting a traffic accident of a road accident includes the following steps:
  • Steps S301 to S304 are the same as steps S101 to S104 shown in FIG. 1a, and therefore, details are not described herein again.
  • the risk prediction method for the road accident is as follows:
  • each of the first vehicles determine each of the first vehicles according to historical information of a traffic accident occurring in the target road segment, vehicle condition information and historical behavior information of each of the first vehicles, and the environmental condition information.
  • the information includes: vehicle condition information and historical behavior information of the second vehicle of a traffic accident occurring in the history of the target road section, and environmental condition information when the traffic accident occurs;
  • Step S305 is similar to S105 shown in FIG. 1a, S305 determines a traffic accident occurrence probability value of each first vehicle occurrence of a traffic accident, and S105 determines a risk factor of each first vehicle occurrence of a traffic accident, the risk factor includes The traffic accident occurrence probability value and/or the severity value of the traffic accident occurs. Therefore, the method of determining the traffic accident occurrence probability value of each first vehicle occurrence of the traffic accident is referred to S105.
  • the information of each of the first vehicles includes: the video structure information, the vehicle condition information of each of the first vehicles, the Historical behavior information of each first vehicle, a traffic accident occurrence probability value of each of the first vehicles in which a traffic accident occurs;
  • the average value of the traffic accident occurrence probability value of the traffic accident of all the vehicles on the target road section is determined, that is, on the target road section.
  • the probability of occurrence of traffic accidents in all vehicles with traffic accidents is averaged, and the average value of the probability of occurrence of traffic accidents in all vehicles on the target road segment is obtained.
  • the average vehicle condition information includes: average vehicle age, average history violation record number, vehicle type distribution, average number of annual inspections, and the like, but is not limited thereto.
  • the average historical behavior information includes: an average number of red lights, an average number of times of speeding, an average number of violations, and the like, but is not limited thereto.
  • the average current driver behavior information includes: an average number of mobile phones, an average number of times of eating, and an average number of people not wearing a seat belt, but is not limited thereto.
  • the road condition information includes an average vehicle speed, a relative distance between the average front and rear vehicles, and a relative vehicle speed of the average front and rear vehicles, but is not limited thereto.
  • the step of obtaining the average vehicle speed comprises: acquiring a speed of each first vehicle on the target road segment, and obtaining an average vehicle speed on the target road segment according to the speed of each first vehicle;
  • Obtaining the relative distance of the average front and rear vehicles includes: obtaining the relative positional relationship of each first vehicle in the road panorama, and obtaining, according to the relative position, the relative of each first vehicle and the adjacent vehicle The distance is obtained according to the relative distance between each first vehicle and the adjacent vehicle, and the relative distance between the average front and rear vehicles of all the vehicles on the target road segment is obtained.
  • Obtaining the relative vehicle speed of the average front and rear vehicles includes: obtaining, according to the relative positional relationship of each first vehicle in the road panorama, and the speed of each first vehicle, obtaining each first vehicle adjacent to the first vehicle
  • the relative vehicle speed of the vehicle is calculated based on the relative vehicle speed, and the relative vehicle speed of the average front and rear vehicles of all the vehicles on the target road segment is calculated.
  • Average vehicle condition information average historical behavior information, average value of traffic accident occurrence probability values of traffic accidents, and average current driver behavior information of all vehicles and the target road section of all the vehicles on the target road segment to be obtained.
  • the road condition information is input to a preset prediction model, which is an average vehicle condition for all the vehicles on the target road section in which the traffic accident occurs in advance for a traffic accident occurring in history.
  • Information, average historical behavior information, average value of traffic accident probability values for occurrence of traffic accidents, and average current driver behavior information and road condition information on the target road segment, and environmental condition information at the time of occurrence of such traffic accidents are trained. acquired.
  • the average vehicle condition information of all the obtained vehicles, the average historical behavior information, the average value of the traffic accident probability values of the traffic accidents, and the average current driver behavior information of all the vehicles and the road condition information on the target road section are digitized. And averaged vehicle condition information, average historical behavior information, average value of traffic accident occurrence probability of occurrence of traffic accidents, and average current driver behavior information of all vehicles and road conditions on the target road section Information, input to a pre-set forecasting model. It should be noted that the information input to the prediction model is numerical information.
  • the prediction model is based on the average vehicle condition information of all vehicles that are numerically input, the average historical behavior information, the average value of the traffic accident occurrence probability value of the traffic accident, and the average current driver behavior information of all the vehicles and the target road segment.
  • the road condition information calculates the first occurrence probability value of the road accident.
  • the probability of occurrence of a traffic accident in which all the vehicles have a traffic accident is calculated.
  • the average value of the values combined with the average vehicle condition information of all vehicles, the average historical behavior information, and the average current driver behavior information and road condition information of all vehicles, determines the first probability value of the road traffic accident. In this way, when determining the first occurrence probability value of a road accident, the probability of occurrence of a traffic accident in all vehicles is considered, and the accuracy of the risk prediction of the road accident is improved.
  • a method for predicting a traffic accident of a road accident includes the following steps:
  • Steps S401 to S405 are the same as steps S201 to S205 shown in FIG. 1b, and therefore, details are not described herein again.
  • step S406 the risk prediction method for the road accident is as follows:
  • S406. Determine, for each of the first vehicles, the historical information according to the target road segment, the vehicle condition information and the historical behavior information of each of the first vehicles, and the environmental condition information.
  • S406 is similar to S206 shown in FIG. 1b, S406 determines a traffic accident occurrence probability value of each first vehicle occurrence of a traffic accident, and S206 determines an initial risk factor of each first vehicle occurrence of a traffic accident, the risk factor The traffic accident occurrence probability value and/or the severity value of the traffic accident is included, so the step of determining the traffic accident occurrence probability value of each first vehicle occurrence traffic accident in S406 refers to S206.
  • each first vehicle For each first vehicle, correct an initial traffic accident occurrence probability value of a traffic accident of the first vehicle according to a relative distance and a relative speed of the first vehicle and its neighboring vehicle, and obtain each of the first The final traffic accident occurrence probability value of a traffic accident occurred in a vehicle; the final traffic accident occurrence probability value of each obtained first vehicle occurrence traffic accident is taken as the traffic accident occurrence probability value of each first vehicle occurrence traffic accident;
  • S407 is similar to S207 shown in FIG. 1b, and S407 considers the initial traffic accident probability of a traffic accident for each first vehicle on the basis of considering the relative distance and relative speed of each first vehicle and its neighboring vehicles.
  • the value is corrected, and in S207, based on the relative distance and relative speed of each first vehicle and its neighboring vehicles, the initial risk factor of each first vehicle occurrence of a traffic accident is corrected, and the risk factor includes Traffic accident probability value and / or traffic occurrence
  • the severity value of the accident is determined in S407 in S407 for correcting the initial traffic accident occurrence probability value of each of the first vehicles.
  • the information of each of the first vehicles includes: the video structure information, the vehicle condition information of each of the first vehicles, the Historical behavior information of each first vehicle, a traffic accident occurrence probability value of each of the first vehicles in which a traffic accident occurs;
  • the S408 is the same as the step S306 in FIG. 2a, and details are not described herein again.
  • the average vehicle condition information, the average historical behavior information, the average value of the traffic accident occurrence probability values of the traffic accidents, and the average current driver behavior information and road conditions of all the vehicles are obtained for all the vehicles on the target road segment obtained.
  • the information is digitized, and the average vehicle condition information, the average historical behavior information of all the vehicles on the target road segment, the average value of the traffic accident probability value of the traffic accident, and the average current driver behavior information of all the vehicles are calculated.
  • the road condition information is input to a preset prediction model such that the prediction model outputs the numerical value.
  • the initial traffic accident occurrence probability value of each of the obtained first vehicle occurrence traffic accidents is corrected, and the final traffic event of each first vehicle occurrence traffic accident is obtained. Therefore, on the basis of the probability value, the average value of the probability of occurrence of the traffic accident in which the vehicle is involved in the traffic accident is calculated, and the average vehicle condition information, the average historical behavior information, and the average current driver behavior information and the road of all the vehicles are combined. Status information determines the first probability value of a road accident. In this way, when determining the first occurrence probability value of a road accident, the probability of occurrence of a traffic accident for each first vehicle is considered, and the accuracy of the risk prediction of the road accident is improved.
  • the video structured information further includes: current driver behavior information of each of the first vehicles; and the historical information further includes: a current current of the second vehicle when each traffic accident occurs Driver behavior information;
  • the step of generating an initial traffic accident probability value of the traffic accident of the vehicle includes: for each of the first vehicles, historical information of a traffic accident according to the target road segment, vehicle condition information and history of each of the first vehicles.
  • the behavior information, and the environmental condition information and the current driver behavior information of each of the first vehicles determine an initial traffic accident occurrence probability value of each of the first vehicles in which a traffic accident occurs.
  • the current driver behavior information includes: the driver calls and plays Mobile phones, etc., when the driver drives, play the mobile phone, affecting the driver's attention, making the current traffic accident probability value increase.
  • a stored score table which is a table for scoring the historical information of each second vehicle in which the traffic accident occurs in advance for a traffic accident occurring in history; for each of the a first vehicle, in the rating table, searching for vehicle condition information and historical behavior information of each of the first vehicles, and the environmental condition information and the current driver behavior letter of each of the first vehicles Corresponding scoring of the interest rate; according to the corresponding scoring, obtaining a traffic accident occurrence probability value of each of the first vehicle occurrence traffic accidents as an initial traffic accident occurrence probability value of the first vehicle occurrence traffic accident.
  • the step of determining the initial traffic accident occurrence probability value of each of the first vehicles in the traffic accident including: the environmental condition information and the current driver behavior information of each of the first vehicles, including:
  • each classifier corresponds to a traffic accident occurring in the history of the target road segment; each of the classifiers is pre-targeted to a historical traffic accident, for each occurrence of the traffic accident
  • the vehicle condition information and the historical behavior information of the second vehicle, and the environmental condition information when the traffic accident occurs and the current driver behavior information of the second vehicle are obtained by training; obtaining the traffic accident outputted by each classifier
  • the occurrence probability value is an initial traffic accident occurrence probability value of each of the first vehicles in which a traffic accident occurs.
  • the initial traffic accident occurrence probability value of the first vehicle occurrence traffic accident is corrected, Obtaining a final traffic accident occurrence probability value of each of the first vehicles in a traffic accident, comprising:
  • the probability of the final traffic accident occurrence of the t-th traffic accident for the ith car The initial traffic accident probability value of the t-th traffic accident for the i-th car, where N i is the number of vehicles adjacent to the i-th car or the number of vehicles less than the second preset value from the i-th car, d is the relative distance between the i-th car and the j-th car in the road panorama information, and v is the relative speed between the i-th car and the j-th car,
  • N i is the number of vehicles adjacent to the i-th car or the number of vehicles less than the second preset value from the i-th car
  • d is the relative distance between the i-th car and the j-th car in the road panorama information
  • v is the relative speed between the i-th car and the j-th car
  • X t ⁇ is the influence factor of the type of accident in the preceding vehicle, which causes the type of t-accident in the rear car or the severity
  • the initial traffic accident probability value of the t-th traffic accident for the jth car T is the set of accident types, t is the type of accident occurred in the preceding vehicle, and ⁇ is the type of accident occurred in the after-car, a, ⁇ , ⁇ , ⁇ , X t ⁇ is a preset constant, which is any decimal between 0 and 1;
  • the final traffic accident occurrence probability value of each of the obtained first vehicle occurrence traffic accidents is taken as the traffic accident occurrence probability value of each first vehicle occurrence traffic accident.
  • the method for predicting a traffic accident of a road accident further includes: determining whether a first occurrence probability value of a road traffic accident output by the prediction model is greater than a preset third threshold If the judgment result is yes, an early warning message is generated.
  • the early warning information is generated and output.
  • Road administrators, drivers, traffic police and other personnel on the target road section receive early warning information through the road electronic screen, and take measures to avoid traffic accidents, thereby effectively reducing vehicles and roads. The risk of a traffic accident.
  • a risk prediction device for a traffic accident occurred in a vehicle provided by an embodiment of the present application, where the device includes:
  • the first obtaining unit 301a is configured to obtain a monitoring video on the target road segment
  • the processing unit 302a is configured to obtain video structured information according to the monitoring video; the video structured information includes at least license plate number information of each first vehicle on the target road segment;
  • the searching unit 303a is configured to query, according to the license plate number information, vehicle condition information and historical behavior information of each first vehicle on the target road segment in a preset database;
  • a second obtaining unit 304a configured to acquire environmental condition information of the target road segment when the monitoring video is captured
  • a first determining unit 305a configured, for each of the first vehicles, historical information of a traffic accident according to the target road segment, vehicle condition information and historical behavior information of each of the first vehicles, and the environmental condition information Determining a risk factor of a traffic accident in each of the first vehicles;
  • the risk factor includes: a traffic accident occurrence probability value and/or a severity value of the occurrence of the traffic accident, the historical information including: the target road segment history Vehicle condition information and historical behavior information of a second vehicle in which a traffic accident occurs, and environmental condition information when such a traffic accident occurs;
  • the second determining unit 306a is configured to determine the risk factor as a risk prediction result of a traffic accident of the vehicle of the target road segment.
  • the video structured information is obtained through the monitoring video on the obtained target road segment, and the first license information of each first vehicle on the target road segment is queried in the preset database according to the license plate number information in the video structured information.
  • Vehicle condition information and historical behavior information at the same time, obtaining environmental condition information when the monitoring video is captured, determining the vehicle status information, historical behavior information, environmental condition information, and historical information of traffic accidents on the target road segment, and determining each The risk factor of the traffic accident of the first vehicle can obtain the risk prediction result of the vehicle.
  • a risk prediction device for a traffic accident of a vehicle provided by an embodiment of the present application, the device includes:
  • the first obtaining unit 301b, the processing unit 302b, the searching unit 303b, and the second obtaining The unit 304b is the same as the first obtaining unit 301a, the processing unit 302a, the searching unit 303a, and the second obtaining unit 304a shown in FIG. 3a, and details are not described herein again.
  • a receiving unit 305b configured to receive a speed of each first vehicle on the target road segment
  • a third obtaining unit 306b configured to obtain a road panorama of the target road segment according to the monitoring video
  • a fourth obtaining unit 307b configured to obtain, according to the road panorama, a relative distance between each of the first vehicles and its neighboring vehicles; and obtain a location according to the road panorama and the speed of each of the first vehicles Describe the relative speed of each first vehicle and its neighboring vehicles;
  • a first determining unit 308b configured, for each of the first vehicles, historical information of a traffic accident according to the target road segment, vehicle condition information and historical behavior information of each of the first vehicles, and the environmental condition information Determining an initial risk factor of a traffic accident in each of the first vehicles;
  • a correction unit 309b configured to correct, for each first vehicle, an initial risk factor of a traffic accident occurring in the first vehicle according to a relative distance and a relative speed of the first vehicle and its neighboring vehicles, to obtain each of the The ultimate risk factor for a traffic accident in the first vehicle;
  • the second determining unit 310b is configured to determine a final risk factor of the traffic accident of each of the first vehicles as a risk prediction result of a traffic accident of the vehicle of the target road segment.
  • the initial risk factor of each first vehicle occurrence traffic accident is corrected, and a traffic accident occurs in each of the first vehicles.
  • the ultimate risk factor is the final risk factor as a predictor of the risk of a vehicle accident. It can be seen that applying the implementation of the present application improves the accuracy of the risk prediction of a traffic accident in a vehicle.
  • the video structured information further includes: current driver behavior information of each of the first vehicles; and the historical information further includes: a current current of the second vehicle when each traffic accident occurs Driver behavior information;
  • the first determining unit 308b is specifically configured to: for each of the first vehicles, historical information of a traffic accident according to the target road segment, vehicle condition information and historical behavior of each of the first vehicles The information, and the environmental condition information and the current driver behavior information of each of the first vehicles, determine an initial risk factor of the traffic accident for each of the first vehicles.
  • the first determining sub-unit 308b is specifically configured to acquire a stored score table, which is a kind of traffic accident that occurs in history for the second vehicle that has occurred in each of the traffic accidents. a table for scoring historical information; for each of the first vehicles, searching for the vehicle condition information and historical behavior information of each of the first vehicles in the rating table, and the environmental condition information and the first vehicle Corresponding scoring of current driver behavior information; obtaining, according to the corresponding scoring, a traffic accident occurrence probability value of each of the first vehicles and/or a severity of a traffic accident as an initial risk factor of the first vehicle occurrence traffic accident .
  • the first determining sub-unit 308b is specifically configured to, for each of the first vehicles, vehicle condition information and historical behavior information of each of the first vehicles, and the environmental condition information and each of the first vehicles
  • the current driver behavior information is input to a plurality of preset classifiers, each classifier corresponding to a traffic accident occurring in the history of the target road segment; each of the classifiers is pre-targeted to a historical occurrence a traffic accident obtained by training the vehicle condition information and historical behavior information of the second vehicle in which the traffic accident occurs, and the environmental condition information when the traffic accident occurs and the current driver behavior information of the second vehicle; Obtaining a traffic accident occurrence probability value and/or a severity value of a traffic accident outputted by each of the classifiers as an initial risk factor of the first vehicle occurrence of the traffic accident.
  • the modifying subunit 309b is specifically configured to:
  • the ultimate risk factor for the t-th traffic accident in the ith car The initial risk factor for the t-th traffic accident for the i-th car, N i is the number of vehicles adjacent to the i-th car or the number of vehicles with the distance from the i-th car being less than the first preset value, d is The relative distance between the i-th car and the j-th car in the road panorama information, v is the relative speed between the i-th car and the j-th car in the road panorama information, For the relative distance influence factor between the ith car and the jth car, For the relative speed influence factor between the i-th car and the j-th car, X t ⁇ is the influence factor of the type of accident in the preceding vehicle, which causes the type of t-accident in the rear car or the severity of the car accident.
  • the initial risk factor for the t-th traffic accident in the jth car T is the set of accident types, t is the type of accident in the preceding vehicle, and ⁇ is the type of accident in the after-car, a, ⁇ , ⁇ , ⁇ , X t ⁇ respectively
  • the preset constant is any decimal between 0 and 1.
  • the device further includes:
  • a determining unit configured to determine whether each of the final risk factors of the traffic accidents of each of the first vehicles is greater than a preset first threshold; if the determination result is yes, And generating, by each of the first vehicles of the threshold, an early warning information output; and/or determining whether a severity value of each of the final risk factors of each of the first vehicles in which the traffic accident occurred is greater than a preset second threshold If the judgment result is YES, the warning information output is generated for each of the first vehicles that are greater than the preset second threshold.
  • the apparatus comprising:
  • the first obtaining unit 401a, the processing unit 402a, the searching unit 403a, and the second obtaining unit 404a in the embodiment of the present application are respectively associated with the first obtaining unit 301a, the processing unit 302a, and the searching unit 303a in the embodiment shown in FIG. 3a. It is the same as the second obtaining unit 304a, and therefore, it will not be described again here.
  • a first determining unit 405a configured, for each of the first vehicles, historical information of a traffic accident according to the target road segment, vehicle condition information and historical behavior information of each of the first vehicles, and the environmental condition information Determining a traffic accident occurrence probability value of each first vehicle in which a traffic accident occurs;
  • the historical information includes: vehicle condition information and historical behavior information of the second vehicle of a traffic accident occurring in the history of the target road section, and occurrence of the Environmental status information at the time of a traffic accident;
  • the third obtaining unit 406a is configured to perform statistics on information of each first vehicle on the target road segment, obtain average vehicle condition information, average historical behavior information, and traffic accidents of traffic accidents on all the vehicles on the target road segment. An average of occurrence probability values, and average current driver behavior information and road information on the target road segment; the information of each of the first vehicles, including: the video structuring information, each of the first vehicles Vehicle condition information, historical behavior information of each of the first vehicles, and a traffic accident occurrence probability value of each of the first vehicles in which a traffic accident occurs;
  • the input unit 407a is configured to obtain average vehicle condition information, average historical behavior information, average value of traffic accident occurrence probability values of traffic accidents, and average current driver behavior information of all vehicles on the target road section obtained, and
  • the road condition information on the target road segment is input into a preset prediction model, which is a kind of traffic accident that occurs in history for all of the target road segments for each occurrence of the traffic accident.
  • a fourth obtaining unit 408a configured to obtain a first occurrence probability value of a road occurrence traffic accident output by the prediction model
  • the second determining unit 409a is configured to determine the first occurrence probability value as a risk prediction result of a road accident of the target road segment.
  • the video structured information is obtained by using the monitoring video on the obtained target road segment, and each first vehicle on the target road segment is queried in the preset database according to the license plate number information in the video structured information.
  • the vehicle condition information and the historical behavior information obtain the environmental condition information when the monitoring video is taken, and obtain the vehicle condition information, the historical behavior information, the environmental condition information, and the historical information of the traffic accident occurring on the target road section, and determine each The probability of occurrence of a traffic accident in a first vehicle, and the average value of the probability of occurrence of a traffic accident in all vehicles, combined with the average vehicle condition information, average historical behavior information, and all vehicles of all vehicles.
  • the current driver behavior information and road condition information are averaged, and the first probability value of the road traffic accident is determined as the risk prediction result of the road traffic accident on the target road section.
  • the average value of the probability of occurrence of a traffic accident in which all vehicles have a traffic accident is considered, and the accuracy of the risk prediction of a road accident is improved.
  • a risk prediction device for road traffic accidents provided by an embodiment of the present application includes:
  • first obtaining unit 401b, the processing unit 402b, the searching unit 403b, and the second obtaining unit 404b in this embodiment are respectively associated with the first obtaining unit 301a, the processing unit 302a, the searching unit 303a, and the first obtaining unit 301a shown in FIG. 3a.
  • the second obtaining unit 304a is the same, and therefore, details are not described herein again.
  • a receiving unit 405b configured to receive a speed of each first vehicle on the target road segment
  • a third obtaining unit 406b configured to obtain a road panorama of the target road segment according to the monitoring video
  • a fourth obtaining unit 407b specifically configured to obtain, according to the road panorama, a relative distance between each of the first vehicles and its neighboring vehicles; according to the road panorama and the speed of each of the first vehicles, The relative speed of each of the first vehicles and their neighboring vehicles;
  • a first determining unit 408b configured, for each of the first vehicles, historical information of a traffic accident according to the target road segment, vehicle condition information and historical behavior information of each of the first vehicles, and the environmental condition information Determining an initial traffic accident occurrence probability value of each of the first vehicles in which a traffic accident occurs;
  • a correction unit 409b configured, for each first vehicle, an initial traffic accident in which a traffic accident occurs in the first vehicle according to a relative speed and a relative distance of the first vehicle and its neighboring vehicles The probability value is corrected to obtain a final traffic accident occurrence probability value of each of the first vehicles in which a traffic accident occurs; and the obtained final traffic accident occurrence probability value of each first vehicle occurrence traffic accident is taken as each first The probability of occurrence of a traffic accident in which a vehicle has a traffic accident.
  • the fifth obtaining unit 410b is configured to perform statistics on information of each first vehicle on the target road segment, obtain average vehicle condition information, average historical behavior information, and traffic accidents of traffic accidents on all the vehicles on the target road segment. An average of occurrence probability values, and average current driver behavior information and road information on the target road segment; the information of each of the first vehicles, including: the video structuring information, each of the first vehicles Vehicle condition information and historical behavior information, a probability of occurrence of a traffic accident in which each of the first vehicles has a traffic accident;
  • the input unit 411b is configured to obtain average vehicle condition information, average historical behavior information, average value of traffic accident occurrence probability values of traffic accidents, and average current driver behavior information of all vehicles on the target road segment obtained, and
  • the road condition information on the target road segment is input into a preset prediction model, which is a kind of traffic accident that occurs in history for all of the target road segments for each occurrence of the traffic accident.
  • a sixth obtaining unit 412b configured to obtain a first occurrence probability value of a road occurrence traffic accident output by the prediction model
  • the second determining unit 413b is configured to determine the first occurrence probability value as a risk prediction result of a road accident of the target road segment.
  • the initial traffic accident occurrence probability value of each first vehicle occurrence traffic accident is corrected, and each first vehicle occurrence is obtained.
  • the probability of traffic accident occurrence of traffic accidents combined with the average vehicle condition information of all vehicles, the average historical behavior information, and the average current driver behavior information and road condition information of all vehicles, determine the first probability value of the road traffic accident as the target road section.
  • the risk prediction result of a traffic accident on the road Applying the embodiments of the present application, the accuracy of the risk prediction of road traffic accidents is improved.
  • the video structured information further includes: current driver behavior information of each of the first vehicles; and the historical information further includes: current driver behavior information of the second vehicle when each traffic accident occurs;
  • the first determining unit 408b is specifically configured to, for each of the first vehicles, history information of a traffic accident according to the target road segment, vehicle condition information and historical behavior information of each of the first vehicles, and the The environmental condition information and the current driver behavior information of each of the first vehicles determine an initial traffic accident occurrence probability value of each of the first vehicles in which a traffic accident occurs.
  • the first determining unit 408b is specifically configured to obtain a stored score table, where the score table is in advance for a traffic accident occurring in history, for each second vehicle in which the traffic accident occurs. a table for scoring historical information; for each of the first vehicles, searching for the vehicle condition information and historical behavior information of each of the first vehicles, and the environmental status information and each of the first Corresponding scoring of current driver behavior information of a vehicle; obtaining, according to the corresponding scoring, a traffic accident occurrence probability value of each of the first vehicle occurrence traffic accidents as an initial traffic accident occurrence probability value of the first vehicle occurrence traffic accident .
  • the first determining subunit 408b is specifically configured to, for each of the first vehicles, the vehicle condition information and historical behavior information of each of the first vehicles, and the environmental condition information and each of the first
  • the current driver behavior information of the vehicle is input to a plurality of preset classifiers, each classifier corresponding to a traffic accident occurring in the history of the target road segment; each of the classifiers is pre-targeted for one of the historical occurrences a traffic accident, which is obtained by training the vehicle condition information and historical behavior information of the second vehicle in which the traffic accident occurs, and the environmental condition information when the traffic accident occurs and the current driver behavior information of the second vehicle.
  • the correction subunit 409b is specifically configured to calculate an initial traffic accident probability value for each first vehicle occurrence of a traffic accident, and perform calculations by using the following formula to obtain a final traffic accident occurrence of each first vehicle occurrence traffic accident. Probability value:
  • the probability of the final traffic accident occurrence of the t-th traffic accident for the ith car The initial traffic accident probability value of the t-th traffic accident for the i-th car, where N i is the number of vehicles adjacent to the i-th car or the number of vehicles less than the second preset value from the i-th car, d is the relative distance between the i-th car and the j-th car in the road panorama information, and v is the relative speed between the i-th car and the j-th car,
  • N i is the number of vehicles adjacent to the i-th car or the number of vehicles less than the second preset value from the i-th car
  • d is the relative distance between the i-th car and the j-th car in the road panorama information
  • v is the relative speed between the i-th car and the j-th car
  • X t ⁇ is the influence factor of the type of accident in the preceding vehicle, which causes the type of t-accident in the rear car or the severity
  • the initial traffic accident probability value of the t-th traffic accident for the jth car T is the set of accident types, t is the type of accident occurred in the preceding vehicle, and ⁇ is the type of accident occurred in the after-car, a, ⁇ , ⁇ , ⁇ , X t ⁇ is a preset constant, and is any decimal between 0 and 1.
  • the device further includes:
  • the determining unit is configured to determine whether the first occurrence probability value of the road occurrence traffic accident output by the prediction model is greater than a preset third threshold value; and if the determination result is yes, generate the early warning information output.
  • a risk prediction system for a vehicle accident occurred in an embodiment of the present application the system 510 includes:
  • the video monitoring device 511 is configured to collect monitoring video on the target road segment
  • a vehicle information database 512 configured to store vehicle condition information of each first vehicle on the target road segment And historical behavior information
  • the intelligent analysis server 513 is configured to obtain a monitoring video on the target road segment collected by the video monitoring device 511, and obtain video structured information according to the monitoring video.
  • the video structured information includes at least each of the target road segments.
  • the license plate number information of the first vehicle querying the vehicle condition information and the historical behavior information of each first vehicle on the target road segment in the vehicle information database 512 according to the license plate number information; acquiring in the external information source system 514 Environmental condition information of the target road segment when the surveillance video is captured;
  • each of the first vehicles Determining each of the first vehicles according to historical information of a traffic accident occurring in the target road segment, vehicle condition information and historical behavior information of each of the first vehicles, and the environmental condition information a risk factor of a traffic accident in a vehicle;
  • the risk factor includes: a traffic accident occurrence probability value and/or a severity value of the occurrence of the traffic accident, the historical information including: a traffic accident occurring in the history of the target road segment Vehicle condition information and historical behavior information of the second vehicle, and environmental condition information when the traffic accident occurs;
  • the risk factor is determined as a risk prediction result of a traffic accident of the vehicle of the target road segment.
  • the video structured information is obtained through the monitoring video on the obtained target road segment, and the first license information of each first vehicle on the target road segment is queried in the preset database according to the license plate number information in the video structured information.
  • Vehicle condition information and historical behavior information at the same time, obtaining environmental condition information when the monitoring video is captured, determining the vehicle status information, historical behavior information, environmental condition information, and historical information of traffic accidents on the target road segment, and determining each The risk factor of the traffic accident of the first vehicle can obtain the risk prediction result of the vehicle.
  • a risk prediction system for a vehicle accident occurred in the embodiment of the present application includes:
  • the video monitoring device 521, the vehicle information database 522, and the smart analysis server 523 in FIG. 5b are the same as the video monitoring device 511, the vehicle information database 512, and the smart analysis server 513 in FIG. 5a, and are not described herein again.
  • the external information source system 526 in Figure 5b does not belong to the system 520.
  • a sensor 524 configured to measure a speed of each of the first vehicles on the target road segment
  • the video structured information further includes: current driver behavior information of each of the first vehicles;
  • the historical information further includes: current driver behavior information of the second vehicle when each traffic accident occurs;
  • the smart analysis server 523 is specifically configured to receive a speed of each of the first vehicles on the target road segment sent by the sensor 524, and obtain a road panorama of the target road segment according to the monitoring video; a road panorama, obtaining a relative distance between each of the first vehicles and its neighboring vehicles; obtaining, according to the road panorama and the speed of each of the first vehicles, the first vehicle and its neighboring vehicles Relative velocity;
  • a stored score table which is a table for scoring the historical information of each second vehicle in which the traffic accident occurs in advance for a traffic accident occurring in history; for each of the a first vehicle, in the rating table, searching for the vehicle condition information and the historical behavior information of each of the first vehicles, and the corresponding rating of the environmental condition information and the current driver behavior information of the first vehicle; Corresponding to the score, obtaining the traffic accident occurrence probability value of each of the first vehicles and/or the severity of the occurrence of the traffic accident as an initial risk factor of the first vehicle occurrence of the traffic accident;
  • each classifier corresponds to a traffic accident occurring in the history of the target road segment; each of the classifiers is pre-targeted to a historical traffic accident, for each occurrence of the traffic.
  • the vehicle condition information and historical behavior information of the second vehicle of the accident, and the environmental condition information when the traffic accident occurs and the current driver behavior information of the second vehicle are obtained by training; obtaining the traffic output by each classifier
  • the probability of occurrence of the accident and/or the severity of the occurrence of the traffic accident as an initial risk factor for the traffic accident of the first vehicle;
  • the ultimate risk factor for the t-th traffic accident in the ith car The initial risk factor for the t-th traffic accident for the i-th car, N i is the number of vehicles adjacent to the i-th car or the number of vehicles with the distance from the i-th car being less than the first preset value, d is The relative distance between the i-th car and the j-th car in the road panorama information, v is the relative speed between the i-th car and the j-th car in the road panorama information, For the relative distance influence factor between the ith car and the jth car, For the relative speed influence factor between the i-th car and the j-th car, X t ⁇ is the influence factor of the type of accident in the preceding vehicle, which causes the type of t-accident in the rear car or the severity of the car accident.
  • T is the set of accident types
  • t is the type of accident in the preceding vehicle
  • is the type of accident in the after-car
  • the preset constant is any decimal between 0 and 1;
  • the first vehicle generates an early warning information output.
  • the early warning device 525 is configured to receive the early warning information output by the intelligent analysis server.
  • the monitoring video on the obtained target road segment is analyzed, and the video structured information and the road panorama image are obtained.
  • each first query is obtained in the preset database.
  • Vehicle condition information and historical behavior information at the same time, obtaining environmental condition information on the target road section when shooting the surveillance video, obtaining the speed of each first vehicle, and combining the road panorama to determine each first vehicle and its adjacent vehicle Relative distance and relative speed, the obtained vehicle condition information, historical behavior information, current driver behavior information, environmental condition information, relative distance and relative speed are input to the classifier of the historical information training of the traffic accident in the target road section, And causing the classifier to output an initial risk factor of each first vehicle in which a traffic accident occurs, correcting the initial risk factor according to the relative distance and relative speed of each first vehicle and its neighboring vehicle, and obtaining each first vehicle occurrence
  • the ultimate risk factor for traffic accidents using the final risk factor as a predictor The risk of accidents occur predictions vehicles.
  • a risk prediction system for road traffic accidents provided by an embodiment of the present application, the system 610 includes:
  • the video monitoring device 611 is configured to collect monitoring video on the target road segment
  • a vehicle information database 612 configured to store vehicle condition information and historical behavior information of each first vehicle on the target road segment
  • the intelligent analysis server 613 is configured to obtain a monitoring video on the target road segment sent by the video monitoring device 611, and obtain video structured information according to the monitoring video.
  • the video structured information includes at least the target road segment.
  • Vehicle license plate number information of each first vehicle querying vehicle condition information and historical behavior information of each first vehicle on the target road segment in the vehicle information database 612 according to the license plate number information; in the external information source system 614 Obtaining environmental condition information of the target road segment when the monitoring video is captured;
  • the first information Determining, for each of the first vehicles, the first information according to the historical information of the traffic accident occurring in the target road segment, the vehicle condition information and the historical behavior information of each of the first vehicles, and the environmental condition information.
  • the probability of occurrence of a traffic accident in which a vehicle has a traffic accident the historical information packet Included: vehicle condition information and historical behavior information of a second vehicle in a history of the target road segment, and environmental condition information when the traffic accident occurs;
  • Counting information of each first vehicle on the target road segment obtaining average vehicle condition information, average historical behavior information, average value of traffic accident occurrence probability values of occurrence of traffic accidents on the target road section, and Mean current driver behavior information and road condition information on the target road segment;
  • the information of each of the first vehicles includes: the video structure information, the vehicle condition information of each of the first vehicles, each of the Historical behavior information of the first vehicle, a probability of occurrence of a traffic accident in which each of the first vehicles has a traffic accident;
  • the prediction model is a kind of traffic accident occurring in history, and average vehicle condition information, average historical behavior information, occurrence of all vehicles on the target road section of each occurrence of the traffic accident.
  • the average value of the traffic accident occurrence probability value of the traffic accident, the average current driver behavior information and the road condition information on the target road section, and the environmental condition information when the traffic accident occurs are obtained by training;
  • an average value of the probability of occurrence of a traffic accident in which all the vehicles have a traffic accident is calculated, and combined with all the vehicles.
  • the average vehicle condition information, the average historical behavior information, and the average current driver behavior information and road condition information of all vehicles determine the first probability value of the road traffic accident. In this way, when determining the first occurrence probability value of a road accident, the probability of occurrence of a traffic accident in all vehicles is considered, and the accuracy of the risk prediction of the road accident is improved.
  • a risk prediction system for road traffic accidents provided by an embodiment of the present application, the system 620 includes:
  • the video monitoring device 621, the vehicle information database 622, and the smart analysis server 623 in FIG. 6b The same as the video monitoring device 611, the vehicle information database 612, and the smart analysis server 613 in FIG. 6a, and details are not described herein again, wherein the external information source system in FIG. 6b does not belong to the system 620.
  • a sensor 624 configured to measure a speed of each of the first vehicles on the target road segment
  • the video structured information further includes: current driver behavior information of each of the first vehicles;
  • the historical information further includes: current driver behavior information of the second vehicle when each traffic accident occurs;
  • the smart analysis server 623 is specifically configured to receive a speed of each of the first vehicles on the target road segment sent by the sensor 624, and obtain a road panorama of the target road segment according to the monitoring video; a road panorama, obtaining a relative speed of each of the first vehicles and their neighboring vehicles; obtaining, according to the road panorama and the speed of each of the first vehicles, the first vehicle and its neighboring vehicles Relative velocity;
  • a stored score table which is a table for scoring the historical information of each second vehicle in which the traffic accident occurs in advance for a traffic accident occurring in history; for each of the a first vehicle, in the rating table, searching for vehicle condition information and historical behavior information of each of the first vehicles, and corresponding scores of the environmental condition information and current driver behavior information of the first vehicle; Scoring, obtaining a traffic accident occurrence probability value of each of the first vehicles as an initial traffic accident occurrence probability value of the first vehicle occurrence traffic accident; or
  • each classifier corresponds to a traffic accident occurring in the history of the target road segment; each of the classifiers is pre-targeted to a historical traffic accident, for each occurrence of the traffic accident
  • the vehicle condition information and the historical behavior information of the second vehicle, and the environmental condition information when the traffic accident occurs and the current driver behavior information of the second vehicle are obtained by training; obtaining the traffic accident outputted by each classifier
  • the occurrence probability value is an initial traffic accident occurrence probability value of each of the first vehicle occurrence traffic accidents; wherein the historical information includes: vehicle condition information of the second vehicle of a traffic accident occurring in the history of the target road section and Historical behavior information, and environmental condition information when the traffic accident occurs and current driver behavior information of the second vehicle;
  • the probability of the final traffic accident occurrence of the t-th traffic accident for the ith car The initial traffic accident probability value of the t-th traffic accident for the i-th car, where N i is the number of vehicles adjacent to the i-th car or the number of vehicles less than the second preset value from the i-th car, d is the relative distance between the i-th car and the j-th car in the road panorama information, and v is the relative speed between the i-th car and the j-th car,
  • N i is the number of vehicles adjacent to the i-th car or the number of vehicles less than the second preset value from the i-th car
  • d is the relative distance between the i-th car and the j-th car in the road panorama information
  • v is the relative speed between the i-th car and the j-th car
  • X t ⁇ is the influence factor of the type of accident in the preceding vehicle, which causes the type of t-accident in the rear car or the severity
  • the initial traffic accident probability value of the t-th traffic accident for the jth car T is the set of accident types, t is the type of accident occurred in the preceding vehicle, and ⁇ is the type of accident occurred in the after-car, a, ⁇ , ⁇ , ⁇ , X t ⁇ is a preset constant, which is any decimal between 0 and 1;
  • the final traffic accident occurrence probability value of each of the obtained first vehicle occurrence traffic accidents is taken as the traffic accident occurrence probability value of each first vehicle occurrence traffic accident;
  • the early warning device 625 is configured to receive the early warning information output by the intelligent analysis server.
  • the initial traffic accident occurrence probability value of each of the obtained first vehicle occurrence traffic accidents is corrected, and the final traffic accident occurrence probability value of each first vehicle occurrence traffic accident is obtained.
  • Calculate the average value of the probability of occurrence of traffic accidents in which the vehicle has a traffic accident and combine the average vehicle condition information of all vehicles, the average historical behavior information, and the average current driver behavior information and road condition information of all vehicles to determine the road traffic accident.
  • the first probability value is used. In this way, when determining the first occurrence probability value of a road accident, the probability of occurrence of a traffic accident for each first vehicle is considered, and the accuracy of the risk prediction of the road accident is improved.
  • the present application further provides a storage medium, wherein the storage medium is used to store executable program code for performing a traffic accident of a vehicle described in the present application at runtime.
  • Risk prediction method and a risk prediction method for road traffic accidents includes:
  • the video structured information includes at least license plate number information of each first vehicle on the target road segment;
  • a risk factor of a traffic accident in the vehicle includes: a traffic accident occurrence probability value and/or a severity value of the occurrence of the traffic accident, the historical information including: a traffic accident occurrence in the history of the target road segment Vehicle condition information and historical behavior information of the two vehicles, and environmental condition information when the traffic accident occurs;
  • the video structured information is obtained through the monitoring video on the obtained target road segment, and the first license information of each first vehicle on the target road segment is queried in the preset database according to the license plate number information in the video structured information.
  • Vehicle condition information and historical behavior information at the same time, obtaining environmental condition information when the monitoring video is captured, determining the vehicle status information, historical behavior information, environmental condition information, and historical information of traffic accidents on the target road segment, and determining each The risk factor of the traffic accident of the first vehicle can obtain the risk prediction result of the vehicle.
  • the method for predicting the risk of a road accident caused by the present application includes:
  • the video structured information includes at least license plate number information of each first vehicle on the target road segment;
  • the first information Determining, for each of the first vehicles, the first information according to the historical information of the traffic accident occurring in the target road segment, the vehicle condition information and the historical behavior information of each of the first vehicles, and the environmental condition information.
  • a traffic accident occurrence probability value of a traffic accident of the vehicle includes: vehicle condition information and historical behavior information of the second vehicle of a traffic accident occurring in the history of the target road section, and an environment when the traffic accident occurs Status information;
  • Counting information of each first vehicle on the target road segment obtaining average vehicle condition information, average historical behavior information, average value of traffic accident occurrence probability values of occurrence of traffic accidents on the target road section, and Mean current driver behavior information and road condition information on the target road segment;
  • the information of each of the first vehicles includes: the video structure information, the vehicle condition information and the historical behavior information of each of the first vehicles, a traffic accident occurrence probability value of each of the first vehicles in which a traffic accident occurs;
  • Average vehicle condition information, average historical behavior information, average value of traffic accident probability values of traffic accidents, and average current driver behavior of all vehicles on the target road segment to be obtained The information and the road condition information on the target road segment are input into a preset prediction model, which is a kind of traffic accident that occurs in advance for the history, and the target road segment for each occurrence of the traffic accident Average vehicle condition information, average historical behavior information, average value of traffic accident occurrence probability of occurrence of traffic accidents, average current driver behavior information, and road condition information on the target road section, and when such a traffic accident occurs
  • a preset prediction model which is a kind of traffic accident that occurs in advance for the history, and the target road segment for each occurrence of the traffic accident
  • Average vehicle condition information, average historical behavior information, average value of traffic accident occurrence probability of occurrence of traffic accidents, average current driver behavior information, and road condition information on the target road section, and when such a traffic accident occurs Environmental condition information obtained by training;
  • the first occurrence probability value is determined as a risk prediction result of a road accident occurring on the road of the target road section.
  • the monitoring video on the obtained target road segment is analyzed, and the video structured information and the road panorama image are obtained.
  • each first query is obtained in the preset database.
  • Vehicle condition information and historical behavior information at the same time, obtaining environmental condition information on the target road section when shooting the surveillance video, obtaining the speed of each first vehicle, and combining the road panorama to determine each first vehicle and its adjacent vehicle Relative distance and relative speed, the obtained vehicle condition information, historical behavior information, current driver behavior information, environmental condition information, relative distance and relative speed are input to the classifier of the historical information training of the traffic accident in the target road section, And causing the classifier to output an initial risk factor of each first vehicle in which a traffic accident occurs, correcting the initial risk factor according to the relative distance and relative speed of each first vehicle and its neighboring vehicle, and obtaining each first vehicle occurrence
  • the ultimate risk factor for traffic accidents using the final risk factor as a predictor The risk of accidents occur predictions vehicles.
  • the present application further provides an application, wherein the application is used to perform a risk prediction method for a vehicle accident caused by a vehicle and a risk prediction of a road accident at the time of operation.
  • the method for predicting a risk of a traffic accident in a vehicle according to the present application includes:
  • the video structured information includes at least license plate number information of each first vehicle on the target road segment;
  • a risk factor of a traffic accident in the vehicle includes: a traffic accident occurrence probability value and/or a severity value of the occurrence of the traffic accident, the historical information including: a traffic accident occurrence in the history of the target road segment Vehicle condition information and historical behavior information of the two vehicles, and environmental condition information when the traffic accident occurs;
  • the risk factor is determined as a risk prediction result of a traffic accident of the vehicle of the target road segment.
  • the video structured information is obtained through the monitoring video on the obtained target road segment, and the first license information of each first vehicle on the target road segment is queried in the preset database according to the license plate number information in the video structured information.
  • Vehicle condition information and historical behavior information at the same time, obtaining environmental condition information when the monitoring video is captured, determining the vehicle status information, historical behavior information, environmental condition information, and historical information of traffic accidents on the target road segment, and determining each The risk factor of the traffic accident of the first vehicle can obtain the risk prediction result of the vehicle.
  • the method for predicting the risk of a road accident caused by the present application includes:
  • the video structured information includes at least license plate number information of each first vehicle on the target road segment;
  • the historical information For each of the first vehicles, historical information of a traffic accident according to the target road segment, Determining, by the vehicle condition information and the historical behavior information of each first vehicle, the environmental condition information, a traffic accident occurrence probability value of each of the first vehicles in which a traffic accident occurs; the historical information comprising: the target road segment History information and historical behavior information of a second vehicle in a historical accident, and environmental condition information when such a traffic accident occurs;
  • Counting information of each first vehicle on the target road segment obtaining average vehicle condition information, average historical behavior information, average value of traffic accident occurrence probability values of occurrence of traffic accidents on the target road section, and Mean current driver behavior information and road condition information on the target road segment;
  • the information of each of the first vehicles includes: the video structure information, the vehicle condition information and the historical behavior information of each of the first vehicles, a traffic accident occurrence probability value of each of the first vehicles in which a traffic accident occurs;
  • the first occurrence probability value is determined as a risk prediction result of a road accident occurring on the road of the target road section.
  • the monitoring video on the obtained target road segment is analyzed, and the video structured information and the road panorama image are obtained.
  • each first query is obtained in the preset database.
  • Vehicle condition information and historical behavior information at the same time, obtaining environmental condition information on the target road section when shooting the surveillance video, obtaining the speed of each first vehicle, and combining the road panorama to determine each first vehicle and its adjacent vehicle Relative distance and relative speed, the obtained vehicle condition information, historical behavior information, current driver behavior information, environmental condition information, relative distance and relative speed are input to the classifier of the historical information training of the traffic accident in the target road section, Causing the classifier to output an initial risk factor for each of the first vehicles in the event of a traffic accident, According to the relative distance and relative speed of each first vehicle and its neighboring vehicles, the initial risk factor is corrected, and the final risk factor of each first vehicle occurrence traffic accident is obtained, and the final risk factor is used as a predicted traffic accident of the vehicle.
  • Risk prediction results It can be seen that, in the implementation of the present application, based on the relative speed and relative distance of each first vehicle and its neighboring vehicles, the initial risk factor of each first vehicle occurrence traffic accident is corrected, and the risk prediction of the vehicle traffic accident is improved accurately. Sex.
  • an electronic device including:
  • processor a memory, a communication interface, and a bus
  • the processor, the memory, and the communication interface are connected by the bus and complete communication with each other;
  • the memory stores executable program code
  • the processor runs a program corresponding to the executable program code by reading executable program code stored in the memory for performing a risk prediction method for a vehicle accident as described in the present application And a risk prediction method for traffic accidents on the road.
  • the method for predicting a risk of a traffic accident in a vehicle according to the present application includes:
  • the video structured information includes at least license plate number information of each first vehicle on the target road segment;
  • a risk factor of a traffic accident in the vehicle includes: a traffic accident occurrence probability value and/or a severity value of the occurrence of the traffic accident, the historical information including: a traffic accident occurrence in the history of the target road segment Vehicle condition information and historical behavior information of the two vehicles, and environmental condition information when the traffic accident occurs;
  • the risk factor is determined as a risk prediction result of a traffic accident of the vehicle of the target road segment.
  • the video structured information is obtained through the monitoring video on the obtained target road segment, and the first license information of each first vehicle on the target road segment is queried in the preset database according to the license plate number information in the video structured information.
  • Vehicle condition information and historical behavior information at the same time, obtaining environmental condition information when the monitoring video is captured, determining the vehicle status information, historical behavior information, environmental condition information, and historical information of traffic accidents on the target road segment, and determining each The risk factor of the traffic accident of the first vehicle can obtain the risk prediction result of the vehicle.
  • the method for predicting the risk of a road accident caused by the present application includes:
  • the video structured information includes at least license plate number information of each first vehicle on the target road segment;
  • the first information Determining, for each of the first vehicles, the first information according to the historical information of the traffic accident occurring in the target road segment, the vehicle condition information and the historical behavior information of each of the first vehicles, and the environmental condition information.
  • a traffic accident occurrence probability value of a traffic accident of the vehicle includes: vehicle condition information and historical behavior information of the second vehicle of a traffic accident occurring in the history of the target road section, and an environment when the traffic accident occurs Status information;
  • Counting information of each first vehicle on the target road segment obtaining average vehicle condition information, average historical behavior information, average value of traffic accident occurrence probability values of occurrence of traffic accidents on the target road section, and Mean current driver behavior information and road condition information on the target road segment;
  • the information of each of the first vehicles includes: the video structure information, the vehicle condition information and the historical behavior information of each of the first vehicles, a traffic accident occurrence probability value of each of the first vehicles in which a traffic accident occurs;
  • Average vehicle condition information and average historical behavior letter of all vehicles on the target road segment to be obtained The average value of the traffic accident probability value of the traffic accident, and the average current driver behavior information and the road condition information on the target road segment are input into a preset prediction model, which is pre-targeted in history.
  • a traffic accident that occurs, the average vehicle condition information, the average historical behavior information of each vehicle on the target road section of each of the traffic accidents, the average value of the probability of occurrence of a traffic accident, and the average current driver The behavior information and the road condition information on the target road segment, and the environmental condition information when the traffic accident occurs are obtained by training;
  • the first occurrence probability value is determined as a risk prediction result of a road accident occurring on the road of the target road section.
  • the monitoring video on the obtained target road segment is analyzed, and the video structured information and the road panorama image are obtained.
  • each first query is obtained in the preset database.
  • Vehicle condition information and historical behavior information at the same time, obtaining environmental condition information on the target road section when shooting the surveillance video, obtaining the speed of each first vehicle, and combining the road panorama to determine each first vehicle and its adjacent vehicle Relative distance and relative speed, the obtained vehicle condition information, historical behavior information, current driver behavior information, environmental condition information, relative distance and relative speed are input to the classifier of the historical information training of the traffic accident in the target road section, And causing the classifier to output an initial risk factor of each first vehicle in which a traffic accident occurs, correcting the initial risk factor according to the relative distance and relative speed of each first vehicle and its neighboring vehicle, and obtaining each first vehicle occurrence
  • the ultimate risk factor for traffic accidents using the final risk factor as a predictor The risk of accidents occur predictions vehicles.

Abstract

一种发生交通事故的风险预测方法、装置及系统,能实现车辆发生交通事故的风险预测,提高道路发生交通事故风险预测的准确性。该方法包括:获得目标路段上的监控视频(S101);根据该监控视频,获得视频结构化信息(S102);根据视频结构化信息中的车牌号码信息,在预设数据库中查询该目标路段上每个第一车辆的车况信息和历史行为信息(S103);获取目标路段在拍摄监控视频时的环境状况信息(S104);针对每个第一车辆,根据该目标路段发生交通事故的历史信息、车况信息和历史行为信息,以及环境状况信息,确定每个第一车辆发生交通事故的风险因子(S105);将该风险因子确定为目标路段的车辆发生交通事故的风险预测结果(S106)。

Description

发生交通事故的风险预测方法、装置及系统
本申请要求于2016年12月06日提交中国专利局、申请号为201611111798.2发明名称为“发生交通事故的风险预测方法、装置及系统”的中国专利申请的优先权,其全部内容通过引用结合在本申请中。
技术领域
本申请涉及智能交通管理和控制技术领域,特别是涉及一种发生交通事故的风险预测方法、装置及系统。
背景技术
随着社会的发展,我国机动车保有量呈现急速膨胀,随之而来,道路事故的发生频率也呈急剧上升趋势。对特定路段和车辆风险状态进行有效预测,可降低交通事故发生率,减少人们生命财产损失,具有巨大的现实意义和价值。
目前,预测目标路段上的道路事故风险的方法具体为:实时收集目标路段上的人、车、路以及环境等影响道路事故风险的信息;根据收集的信息,预测目标路段道路事故的风险状态。
现有技术的预测道路事故风险的方法是基于目标路段的道路风险来预测事故风险,并没有考虑到目标路段上各车辆的车况差异,比如,对于一辆已经使用二十年且待维修的旧车和一辆刚出厂且质检符合标准的新车,使用上述预测道路事故风险方法得到的发生事故的概率是相同的。这样,上述预测道路事故风险方法,没有针对具体车辆风险状态进行预测。
发明内容
本申请实施例的目的在于提供一种发生交通事故的风险预测方法、装置及系统,以实现车辆发生交通事故的风险预测,提高道路发生交通事故风险预测的准确性。具体技术方案如下:
第一方面,本申请实施例提供了一种车辆发生交通事故的风险预测方法,包括:
获得目标路段上的监控视频;
根据所述监控视频,获得视频结构化信息;所述视频结构化信息至少包括所述目标路段上每个第一车辆的车牌号码信息;
根据所述车牌号码信息,在预设数据库中查询所述目标路段上每个第一车辆的车况信息和历史行为信息;
获取目标路段在拍摄所述监控视频时的环境状况信息;
针对所述每个第一车辆,根据所述目标路段发生交通事故的历史信息、所述每个第一车辆的车况信息和历史行为信息,以及所述环境状况信息,确定所述每个第一车辆发生交通事故的风险因子;所述风险因子包括:交通事故发生概率值和/或发生交通事故的严重程度值,所述历史信息包括:所述目标路段历史上发生的一种交通事故的第二车辆的车况信息和历史行为信息,以及发生该种交通事故时的环境状况信息;
将所述风险因子确定为所述目标路段的车辆发生交通事故的风险预测结果。
第二方面,本申请实施例提供了一种道路发生交通事故的风险预测方法,包括:
获得目标路段上的监控视频;
根据所述监控视频,获得视频结构化信息;所述视频结构化信息至少包括所述目标路段上每个第一车辆的车牌号码信息;
根据所述车牌号码信息,在预设数据库中查询所述目标路段上每个第一车辆的车况信息和历史行为信息;
获取目标路段在拍摄所述监控视频时的环境状况信息;
针对所述每个第一车辆,根据所述目标路段发生交通事故的历史信息、所述每个第一车辆的车况信息和历史行为信息,以及所述环境状况信息,确定所述每个第一车辆发生交通事故的交通事故发生概率值;所述历史信息包括:所述目标路段历史上发生的一种交通事故的第二车辆的车况信息和历史行为信息,以及发生该种交通事故时的环境状况信息;
对所述目标路段上的每个第一车辆的信息进行统计,获得所述目标路段上的所有车辆的平均车况信息、平均历史行为信息、发生交通事故的交通事故发生概率值的平均值、以及平均当前司机行为信息和所述目标路段上的道路状况信息;所述每个第一车辆的信息,包括:所述视频结构化信息、所述每个第一车辆的车况信息和历史行为信息、所述每个第一车辆发生交通事故的交通事故发生概率值;
将获得的所述目标路段上的所有车辆的平均车况信息、平均历史行为信息、发生交通事故的交通事故发生概率值的平均值、以及平均当前司机行为信息和所述目标路段上的道路状况信息,输入至预先设置的预测模型中,所述预测模型是预先针对历史上发生的一种交通事故,对每个发生该种交通事故所述目标路段上的所有车辆的平均车况信息、平均历史行为信息、发生交通事故的交通事故发生概率值的平均值、以及平均当前司机行为信息和所述目标路段上的道路状况信息,以及发生该种交通事故时的环境状况信息进行训练获得的;
获得所述预测模型输出的道路发生交通事故的第一发生概率值;
将所述第一发生概率值确定为所述目标路段的道路发生交通事故的风险预测结果。
第三方面,本申请实施例提供了一种车辆发生交通事故的风险预测装置,包括:
第一获得单元,用于获得目标路段上的监控视频;
处理单元,用于根据所述监控视频,获得视频结构化信息;所述视频结构化信息至少包括所述目标路段上每个第一车辆的车牌号码信息;
查找单元,用于根据所述车牌号码信息,在预设数据库中查询所述目标路段上每个第一车辆的车况信息和历史行为信息;
第二获得单元,用于获取目标路段在拍摄所述监控视频时的环境状况信息;
第一确定单元,用于针对所述每个第一车辆,根据所述目标路段发生交 通事故的历史信息、所述每个第一车辆的车况信息和历史行为信息,以及所述环境状况信息,确定所述每个第一车辆发生交通事故的风险因子;所述风险因子包括:交通事故发生概率值和/或发生交通事故的严重程度值,所述历史信息包括:所述目标路段历史上发生的一种交通事故的第二车辆的车况信息和历史行为信息,以及发生该种交通事故时的环境状况信息;
第二确定单元,用于将所述风险因子确定为所述目标路段的车辆发生交通事故的风险预测结果。
第四方面,本申请实施例提供了一种道路发生交通事故的风险预测装置,包括:
第一获得单元,用于获得目标路段上的监控视频;
处理单元,用于根据所述监控视频,获得视频结构化信息;所述视频结构化信息至少包括所述目标路段上每个第一车辆的车牌号码信息;
查找单元,用于根据所述车牌号码信息,在预设数据库中查询所述目标路段上每个第一车辆的车况信息和历史行为信息;
第二获得单元,用于获取目标路段在拍摄所述监控视频时的环境状况信息;
第一确定单元,用于针对所述每个第一车辆,根据所述目标路段发生交通事故的历史信息、所述每个第一车辆的车况信息和历史行为信息,以及所述环境状况信息,确定每个第一车辆发生交通事故的交通事故发生概率值;所述历史信息包括:所述目标路段历史上发生的一种交通事故的第二车辆的车况信息和历史行为信息,以及发生该种交通事故时的环境状况信息;
第三获得单元,用于对所述目标路段上的每个第一车辆的信息进行统计,获得所述目标路段上的所有车辆的平均车况信息、平均历史行为信息、发生交通事故的交通事故发生概率值的平均值、以及平均当前司机行为信息和所述目标路段上的道路信息;所述每个第一车辆的信息,包括:所述视频结构化信息、所述每个第一车辆的车况信息和历史行为信息、所述每个第一车辆发生交通事故的交通事故发生概率值;
输入单元,用于将获得的所述目标路段上的所有车辆的平均车况信息、平均历史行为信息、发生交通事故的交通事故发生概率值的平均值、以及所有车辆的平均当前司机行为信息和所述目标路段上的道路状况信息,输入至预先设置的预测模型中,所述预测模型是预先针对历史上发生的一种交通事故,对每个发生该种交通事故所述目标路段上的所有车辆的平均车况信息、平均历史行为信息、发生交通事故的交通事故发生概率值的平均值、以及平均当前司机行为信息和所述目标路段上的道路状况信息,以及发生该种交通事故时的环境状况信息进行训练获得的;
第四获得单元,用于获得所述预测模型输出的道路发生交通事故的第一发生概率值;
第二确定单元,用于将所述第一发生概率值确定为所述目标路段的道路发生交通事故的风险预测结果。
第五方面,本申请实施例提供了一种车辆发生交通事故的风险预测系统,包括:
视频监控设备,用于采集目标路段上的监控视频;
车辆信息数据库,用于存储所述目标路段上每个第一车辆的车况信息和历史行为信息;
智能分析服务器,用于获得所述视频监控设备采集的目标路段上的监控视频;根据所述监控视频,获得视频结构化信息;所述视频结构化信息至少包括所述目标路段上每个第一车辆的车牌号码信息;根据所述车牌号码信息,在所述车辆信息数据库中查询所述目标路段上每个第一车辆的车况信息和历史行为信息;在外部信息源系统中获取目标路段在拍摄所述监控视频时的环境状况信息;
针对所述每个第一车辆,根据所述目标路段发生交通事故的历史信息、所述每个第一车辆的车况信息和历史行为信息,以及所述的环境状况信息,确定所述每个第一车辆发生交通事故的风险因子;所述风险因子包括:交通事故发生概率值和/或发生交通事故的严重程度值,所述历史信息包括:所述目标路段历史上发生的一种交通事故的第二车辆的车况信息和历史行为信 息,以及发生该种交通事故时的环境状况信息;
将所述风险因子确定为所述目标路段的车辆发生交通事故的风险预测结果。
第六方面,本申请实施例提供了一种道路发生交通事故的风险预测系统,包括:
视频监控设备,用于采集目标路段上的监控视频;
车辆信息数据库,用于存储所述目标路段上每个第一车辆的车况信息和历史行为信息;
智能分析服务器,用于获得所述视频监控设备采集的所述目标路段上的监控视频;根据所述监控视频,获得视频结构化信息;所述视频结构化信息至少包括所述目标路段上每个第一车辆的车牌号码信息;根据所述车牌号码信息,在所述车辆信息数据库中查询所述目标路段上每个第一车辆的车况信息和历史行为信息;在外部信息源系统中获取目标路段在拍摄所述监控视频时的环境状况信息;
针对所述每个第一车辆,根据所述目标路段发生交通事故的历史信息、所述每个第一车辆的车况信息和历史行为信息,以及所述环境状况信息,确定所述每个第一车辆发生交通事故的交通事故发生概率值;所述历史信息包括:所述目标路段历史上发生的一种交通事故的第二车辆的车况信息和历史行为信息,以及发生该种交通事故时的环境状况信息;
对所述目标路段上的每个第一车辆的信息进行统计,获得所述目标路段上的所有车辆的平均车况信息、平均历史行为信息、发生交通事故的交通事故发生概率值的平均值、以及平均当前司机行为信息和所述目标路段上的道路状况信息;所述每个第一车辆的信息,包括:所述视频结构化信息、所述每个第一车辆的车况信息、所述每个第一车辆的历史行为信息、所述每个第一车辆发生交通事故的交通事故发生概率值;
将获得的所述目标路段上的所有车辆的平均车况信息、平均历史行为信息、发生交通事故的交通事故发生概率值的平均值、以及所有车辆的平均当前司机行为信息和道路状况信息,输入至预先设置的预测模型中,所述预测 模型是预先针对历史上发生的一种交通事故,对每个发生该种交通事故所述目标路段上的所有车辆的平均车况信息、平均历史行为信息、发生交通事故的交通事故发生概率值的平均值、以及平均当前司机行为信息和所述目标路段上的道路状况信息,以及发生该种交通事故时的环境状况信息进行训练获得的;
获得所述预测模型输出的道路发生交通事故的第一发生概率值;将所述第一发生概率值确定为所述目标路段的道路发生交通事故的风险预测结果。
第七方面,本申请提供了一种存储介质,其中,该存储介质用于存储可执行程序代码,所述可执行程序代码用于在运行时执行如上述第一方面所述的一种车辆发生交通事故的风险预测方法及第二方面所述的一种道路发生交通事故的风险预测方法。
第八方面,本申请提供了一种应用程序,其中,该应用程序用于在运行时执行如上述第一方面所述的一种车辆发生交通事故的风险预测方法及第二方面所述的一种道路发生交通事故的风险预测方法。
第九方面,本申请提供了一种电子设备,包括:
处理器、存储器、通信接口和总线;
所述处理器、所述存储器和所述通信接口通过所述总线连接并完成相互间的通信;
所述存储器存储可执行程序代码;
所述处理器通过读取所述存储器中存储的可执行程序代码来运行与所述可执行程序代码对应的程序,以用于执行如上述第一方面所述的一种车辆发生交通事故的风险预测方法及第二方面所述的一种道路发生交通事故的风险预测方法。
本申请实施例提供的一种发生交通事故的风险预测方法、装置及系统,所述方法包括:获得目标路段上的监控视频;根据所述监控视频,获得视频结构化信息;根据所述视频结构化信息中的车牌号码信息,在预设数据库中查询所述目标路段上每个第一车辆的车况信息和历史行为信息;获取目标路 段在拍摄所述监控视频时的环境状况信息;针对所述每个第一车辆,根据所述目标路段发生交通事故的历史信息、所述每个第一车辆的车况信息和历史行为信息,以及所述环境状况信息,确定所述每个第一车辆发生交通事故的风险因子;将所述风险因子确定为所述目标路段的车辆发生交通事故的风险预测结果,应用本申请实施例,实现了车辆发生交通事故的风险预测。
另外,在所述每个第一车辆发生交通事故的风险因子的基础上,结合目标路段发生交通事故的历史信息,确定目标路段的道路发生交通事故的交通事故发生概率值;将所述道路发生交通事故的交通事故发生概率值确定为所述目标路段的道路发生交通事故的风险预测结果。应用本申请实施例,提高了道路发生交通事故风险的准确性;同时,本申请实施例无需在每一车辆上安装额外设备,改造成本低。当然,实施本申请的任一产品或方法必不一定需要同时达到以上所述的所有优点。
附图说明
为了更清楚地说明本申请实施例和现有技术的技术方案,下面对实施例和现有技术中所需要使用的附图作简单地介绍,显而易见地,下面描述中的附图仅仅是本申请的一些实施例,对于本领域普通技术人员来讲,在不付出创造性劳动的前提下,还可以根据这些附图获得其他的附图。
图1a为本申请实施例所提供的车辆发生交通事故的风险预测方法的一种流程图;
图1b为本申请实施例所提供的车辆发生交通事故的风险预测方法的另一种流程图;
图2a为本申请实施例所提供的道路发生交通事故的风险预测方法的一种流程图;
图2b为本申请实施例所提供的道路发生交通事故的风险预测方法的另一种流程图;
图3a为本申请实施例所提供的车辆发生交通事故的风险预测装置的一种结构图;
图3b为本申请实施例所提供的车辆发生交通事故的风险预测装置的另一种结构图;
图4a为本申请实施例所提供的道路发生交通事故的风险预测装置的一种结构图;
图4b为本申请实施例所提供的道路发生交通事故的风险预测装置的另一种结构图;
图5a为本申请实施例所提供的车辆发生交通事故的风险预测系统的一种结构图;
图5b为本申请实施例所提供的车辆发生交通事故的风险预测系统的另一种结构图;
图6a为本申请实施例所提供的道路发生交通事故的风险预测系统的一种结构图;
图6b为本申请实施例所提供的道路发生交通事故的风险预测系统的另一种结构图。
具体实施方式
下面将结合本申请实施例中的附图,对本申请实施例中的技术方案进行清楚、完整地描述,显然,所描述的实施例仅仅是本申请一部分实施例,而不是全部的实施例。基于本申请中的实施例,本领域普通技术人员在没有做出创造性劳动前提下所获得的所有其他实施例,都属于本申请保护的范围。
本申请实施例所提供了发生交通事故的风险预测方法、装置及系统,实现了车辆发生交通事故的风险预测,提高了道路发生交通事故风险预测的准确性。
下面首先对本申请实施例所提供的一种车辆发生交通事故的风险预测方法进行介绍。
如图1a所示,本申请实施例所提供的一种车辆发生交通事故的风险预测方法,可以包括如下步骤:
S101,获得目标路段上的监控视频;
实际应用中,在目标路段上预先安装了多个摄像头,每个摄像头用于的监控该摄像头监控区域内移动目标的运动状态,同时每个摄像头将返回所监控区域内的监控视频。这里,每个摄像头负责监控该摄像头所在的区域范围,目标路段上的所有摄像头共同监控该目标路段上的移动目标的运动状态,这里的移动目标包括车辆,但不限于此。
S102,根据所述监控视频,获得视频结构化信息;所述视频结构化信息至少包括所述目标路段上每个第一车辆的车牌号码信息;
具体的,在获得目标路段上的各个摄像头返回的监控视频时,对所返回的监控视频,采用视频分析技术,获得视频结构化信息,所述视频结构化信息至少包括所述目标路段上每个第一车辆的车牌号码信息,且所述视频结构化信息还可以包括车牌号码信息、司机是否系安全带信息、司机是否打手机信息、车辆型号信息、车辆颜色的信息等。
需要注意的是,目标路段上的可以同时存在至少一辆车或者不存在任何一辆车。当目标路段上存在至少一辆车时,通过视频分析技术对所述监控视频进行结构化分析,获得所述至少一辆车中的每辆车的视频结构化信息。其中,目标路段上的每一辆车对应一个视频结构化信息。例如,如表1所示,表1示出了5辆车的视频结构化信息,且视频结构化信息包括:车牌信息、司机是否系安全带信息、司机是否打手机信息、车辆型号信息、车辆颜色的信息,但不限于此。
表1
Figure PCTCN2017090874-appb-000001
Figure PCTCN2017090874-appb-000002
S103,根据所述车牌号码信息,在预设数据库中查询所述目标路段上每个第一车辆的车况信息和历史行为信息;
第一车辆包括行驶在目标路段上的所有车辆,每个第一车辆是指目标路段上所有车辆中的一辆车。根据视频结构化信息中的车牌号码信息,在预设数据库中查询该车牌号码信息对应的每个第一车辆的车况信息以及历史行为信息,可以理解的是,每辆车的车牌号码信息在预设数据库中对应该辆车的车况信息和历史行为信息,具体的,车况信息包括:车辆年检信息(例如,车辆未年检次数)、车龄等,历史行为信息包括:闯红灯次数、超速次数、违规变道等。这里的预设数据库可以包括交警系统的车辆信息数据库。
S104,获取目标路段在拍摄所述监控视频时的环境状况信息;
所述目标路段的环境状况信息可以包括:获取监控视频时的天气(阴、雨、雪等)、目标路段是否存在弯道、是否存在斜坡、能见度等。因为,不同的环境状况信息对车辆发生交通事故有一定的影响,例如,在能见度低的情况下,当前车意外停止时,后车司机可能看不清前车的行驶状况,没有采取相应的措施,导致交通事故发生。所以,在预测车辆发生交通事故的风险时,要考虑车辆在目标路段所处的环境状况信息。
S105,针对所述每个第一车辆,根据所述目标路段发生交通事故的历史信息、所述每个第一车辆的车况信息和历史行为信息,以及所述环境状况信息,确定所述每个第一车辆发生交通事故的风险因子;所述风险因子包括:交通事故发生概率值和/或发生交通事故的严重程度值,所述历史信息包括:所述目标路段历史上发生的一种交通事故的第二车辆的车况信息和历史行为信息,以及发生该种交通事故时的环境状况信息;
具体的,目标路段上的每个第一车辆对应其车况信息和历史行为信息,可以理解的是,目标路段上的所有车辆的环境状况信息可以是相同的。在实际应用中,当获得每个第一车辆的车况信息和历史行为信息,以及所述环境状况信息后,结合目标路段发生交通事故的历史信息,确定每一第一车辆发生交通事故的风险因子。这里的每个第一车辆发生交通事故的风险因子包括: 交通事故发生概率值和/或发生交通事故的严重程度,也就是根据所述目标路段发生交通事故的历史信息、所述每个第一车辆的车况信息和历史行为信息,以及所述环境状况信息,确定每个第一车辆发生交通事故的交通事故发生概率值,或者根据所述目标路段发生交通事故的历史信息、所述每个第一车辆的车况信息和历史行为信息,以及所述环境状况信息,确定每个第一车辆发生交通事故的严重程度。
S106,将所述风险因子确定为所述目标路段的车辆发生交通事故的风险预测结果。
可以理解的是,风险因子包括:交通事故发生概率值和/或发生交通事故的严重程度,这里将交通事故发生概率值和/或发生交通事故的严重程度确定为风险预测结果。
本申请实施例,通过所获得目标路段上的监控视频,获得视频结构化信息,并根据该视频结构化信息中的车牌号码信息,在预设数据库中查询该目标路段上每个第一车辆的车况信息和历史行为信息,同时,获得拍摄所述监控视频时的环境状况信息,由所获得的车况信息和历史行为信息、环境状况信息,结合目标路段上发生交通事故的历史信息,确定每个第一车辆发生交通事故的风险因子,就能够获得车辆的风险预测结果。
如图1b所示,本申请实施例所提供的一种车辆发生交通事故的风险预测方法,可以包括如下步骤:
S201,获得目标路段上的监控视频;
S202,根据所述监控视频,获得视频结构化信息;所述视频结构化信息至少包括所述目标路段上每个第一车辆的车牌号码信息;
S203,根据所述车牌号码信息,在预设数据库中查询所述目标路段上每个第一车辆的车况信息和历史行为信息;
S204,获取目标路段上在拍摄所述监控视频时的环境状况信息;
其中,步骤S201~步骤S204与图1a所示的步骤S101~步骤S104的相同,在此不再赘述。
S205,接收所述目标路段上的每个第一车辆的速度;根据所述监控视频,获得所述目标路段的道路全景图;根据所述道路全景图,获得所述每个第一车辆与其相邻车辆的相对距离;根据所述道路全景图和所述每个第一车辆的速度,获得所述每个第一车辆与其相邻车辆的相对速度;
可以理解的是,目标路段上的每个第一车辆的速度可以不相同,通常,采用传感器获得每个第一车辆的速度;该传感器可以包括雷达测速装置,但不限于此。
通过对监控视频进行分析,构建目标路段上的道路全景图,该道路全景图用于表示拍摄所获得的该目标路段上的监控视频时,每个第一车辆的相对位置关系。根据该道路全景图中每个第一车辆的相对位置关系,得到每个第一车辆和相邻车辆的相对距离。同时,根据该道路全景图及每个第一车辆的速度,计算获得每个第一车辆与其相邻车辆的相对速度。例如,道路全景图中,与A车相邻的车辆包括:B车、C车,由传感器测量得到的A车的速度为50km/s,B车的速度为45km/s,C车的速度为40km/s,则A车与B车的相对速度为5km/s,A车与C车的相对速度为10km/s。
S206,针对所述每个第一车辆,根据所述目标路段发生交通事故的历史信息、所述每个第一车辆的车况信息和历史行为信息,以及所述环境状况信息,确定所述每个第一车辆发生交通事故的初始的风险因子;
具体的,所述视频结构化信息还包括:所述每个第一车辆的当前司机行为信息;
所述历史信息还包括:目标路段上历史上发生的每种交通事故时,第二车辆的当前司机行为信息,该第二车辆当前司机行为信息是相对于历史上发生交通事故时的司机的行为信息。
具体的,每个第一车辆的当前司机行为信息,是指在获得目标路段上的监控视频时,根据该监控视频获得视频结构化信息中还包括当前司机行为信息,所述当前司机行为信息如表1所示,包括:司机是否系安全带信息、司机是否打手机信息,等但不限于此,还可以包括:司机是否抽烟的信息、司机是否吃东西的信息等。
可以理解的是,针对所述每个第一车辆,根据所述目标路段发生交通事故的历史信息、所述每个第一车辆的车况信息和历史行为信息,以及所述环境状况信息,确定所述每个第一车辆发生交通事故的初始的风险因子的步骤可以包括:针对所述每个第一车辆,根据所述目标路段发生交通事故的历史信息、所述每个第一车辆的车况信息和历史行为信息,以及所述环境状况信息和所述每个第一车辆的当前司机行为信息,确定所述每个第一车辆发生交通事故的初始的风险因子;
可以理解的是,在确定每个第一车辆发生交通事故的初始的风险因子时,需要将每个第一车辆的当前司机行为信息予以考虑。因为当前司机行为信息可能是直接导致交通事故发生的一大因素,例如,当司机一边开车,一边打电话,容易引起司机注意力不集中,没有注意到其他邻近车辆的急刹车、或者道路的交通状况,导致交通事故发生,所以在确定车辆发生交通事故的初始的风险因子时,要结合当前司机行为信息。
这里,历史信息是相对于拍摄监控视频时的每个第一车辆对应的车况信息、历史行为信息、环境状况信息而言,也就是历史信息具体的指,在目标路段上历史上发生交通事故时的第二车辆对应的车况信息(第二车辆在发生交通事故时的年检情况、车龄等)、历史行为信息(第二车辆在发生交通事故时的闯红灯次数、超速次数、违规变道等)、环境状况信息(第二车辆在发生事故时目标路段上的天气、能见度等)。这里所获的每个第一车辆发生交通事故的初始的风险因子并没有考虑与每个第一车辆相邻车辆对该第一车辆的影响。
S207,针对每个第一车辆,根据该第一车辆与其相邻车辆的相对距离和相对速度,对该第一车辆发生交通事故的初始的风险因子进行修正,获得所述每个第一车辆发生交通事故的最终的风险因子;
基于所获得每个第一车辆与其相邻车辆的相对距离和相对速度,修正每个第一车辆发生交通事故的初始的风险因子,得到该第一车辆发生交通事故的最终的风险因子。
可以理解的是,在目标路段上的相邻车辆越靠近,发生交通事故的交通事故发生概率和/或发生交通事故的严重程度越大,相邻车辆的相对速度越接 近,发生交通事故的交通事故发生概率和/或发生交通事故的严重程度越大。所以,在确定每个第一车辆的风险因子时,要考虑该第一车辆与其相邻车辆的相对距离和相对速度。本申请实施例,根据每个第一车辆与其相邻车辆的相对距离和相对速度,修正了每个第一车辆发生交通事故的初始的因子,得到每个第一车辆发生交通事故的最终的风险因子。使得车辆发生交通事故的风险预测更为准确。
S208,将所述每个第一车辆发生交通事故的最终的风险因子作为预测车辆发生交通事故的风险预测结果。
本申请实施例中,对所获得的目标路段上监控视频进行分析,得到视频结构化信息和道路全景图,基于视频结构化信息中的车牌号码信息,在预设数据库中查询获得每个第一车辆的车况信息和历史行为信息,同时,获取目标路段上在拍摄监控视频时的环境状况信息,获得每个第一车辆的速度,并结合道路全景图,确定每个第一车辆与其相邻车辆的相对距离和相对速度,将所获得车况信息、历史行为信息、当前司机行为信息、环境状况信息、相对距离以及相对速度,输入到所述目标路段发生交通事故的历史信息训练的分类器,以使分类器输出每个第一车辆的发生交通事故的初始的风险因子,根据每个第一车辆与其相邻车辆的相对距离和相对速度,修正该初始的风险因子,得到每个第一车辆发生交通事故的最终的风险因子,将最终的风险因子作为预测车辆发生交通事故的风险预测结果。可见,本申请实施中,基于每个第一车辆与其相邻车辆的相对速度和相对距离,修正每个第一车辆发生交通事故的初始的风险因子,提高了车辆发生交通事故的风险预测的准确性。
在本申请实施例的一种可能的实现方式中,基于所述目标路段发生交通事故的历史信息、所述每个第一车辆的车况信息和历史行为信息,以及所述环境状况信息和所述每个第一车辆的当前司机行为信息,确定所述每个第一车辆发生交通事故的初始的风险因子,包括:
获取已存储的评分表,所述评分表是预先针对历史上发生的一种交通事故,对每个发生该种交通事故的第二车辆的所述历史信息进行评分的表;针对所述每个第一车辆,在所述评分表中查找每个第一车辆的车况信息和历史行为信息,以及所述环境状况信息和所述第一车辆的当前司机行为信息的对 应评分;根据所述对应评分,获得所述每个第一车辆的交通事故发生概率值和/或发生交通事故的严重程度作为该第一车辆发生交通事故的初始的风险因子。实际应用中,获取以存储的评分表,该评分表是针对历史上已经发生的一种交通事故,对发生该种交通事故的第二车辆的所述历史信息进行评分的表,也就是根据评分表中历史上已发生交通事故的第二车辆的历史信息的分数,为每个第一车辆的车况信息和历史行为信息,以及所述环境状况信息和所述每个第一车辆的当前司机行为信息进行评分,这里,每个第一车辆的车况信息和历史行为信息,以及所述环境状况信息和所述每个第一车辆的当前司机行为信息,在评分表有对应的评分。具体的,每种交通事故对应一个评分表。
在本申请实施例的一种可能的实现方式中,所述针对所述每个第一车辆,根据所述目标路段发生交通事故的历史信息、所述每个第一车辆的车况信息和历史行为信息,以及所述环境状况信息和所述每个第一车辆的当前司机行为信息,确定所述每个第一车辆发生交通事故的初始的风险因子的步骤,包括:
针对所述每个第一车辆,将所述每个第一车辆的车况信息和历史行为信息,以及所述环境状况信息和所述每个第一车辆的当前司机行为信息,输入至多个预设的分类器中,每个分类器对应所述目标路段历史上发生的一种交通事故;所述每个分类器,是预先针对历史上发生的一种交通事故,对每个发生该种交通事故的第二车辆的车况信息和历史行为信息,以及发生该种交通事故时的环境状况信息和所述第二车辆的当前司机行为信息进行训练获得的;
实际应用中,在将所述每个第一车辆的车况信息和历史行为信息,以及所述环境状况信息和所述每个第一车辆的当前司机行为信息,输入至多个预设的分类器之前,所述方法还包括:将所述每个第一车辆的车况信息和历史行为信息,以及所述环境状况信息和所述每个第一车辆的当前司机行为信息数值化,将数值化后的每个第一车辆的车况信息和历史行为信息,以及环境状况信息和当前司机行为信息,输入至多个预设的分类器中,这样,获得每个分类器输出的交通事故发生概率值和/或发生交通事故的严重程度值作为该第一车辆发生交通事故的初始的风险因子,其中,每一种交通事故对应一种 分类器。
具体的,针对每个第一车辆,根据该第一车辆与其相邻车辆的相对距离和相对速度,对该第一车辆发生交通事故的初始的风险因子进行修正,获得所述每个第一车辆发生交通事故的最终的风险因子的步骤包括:
针对每个第一车辆发生交通事故的初始的风险因子中的每种交通事故发生概率值和/或发生交通事故的严重程度值,用如下公式分别进行计算,获得所述每个第一车辆发生交通事故的最终的风险因子:
Figure PCTCN2017090874-appb-000003
Figure PCTCN2017090874-appb-000004
Figure PCTCN2017090874-appb-000005
其中,
Figure PCTCN2017090874-appb-000006
为第i辆车发生第t种交通事故的最终的风险因子,
Figure PCTCN2017090874-appb-000007
为第i辆车发生第t种交通事故的初始的风险因子,Ni为与第i辆车相邻的车辆数量或者与第i辆车的距离小于第一预设值的车辆数目,d为所述道路全景图信息中的第i辆车和第j辆车之间的相对距离,v为所述道路全景图信息中第i辆车和第j辆车之间的相对速度,
Figure PCTCN2017090874-appb-000008
为第i辆车和第j辆车之间的相对距离影响因子,
Figure PCTCN2017090874-appb-000009
为第i辆车和第j辆车之间的相对速度影响因子,X为前车发生事故t类型导致后车发生τ事故类型或加剧后车事故严重程度的影响因子,
Figure PCTCN2017090874-appb-000010
为第j辆车发生第τ种交通事故的初始的风险因子,T为事故类型集合,t为前车发生事故类型,τ为后车发生事故类型,a、β、γ、η、X分别为预设常数,均为0~1之间的任一小数。
需要注意的是,γ、η根据实际情况进行调整,X是基于历史统计数据获得的,例如,在前车发生撞护栏的t事故类型下,导致后车发生追尾的τ事故类型的概率。
可见,与每个第一车辆发生交通事故的初始的风险因子相比,每个第一车辆发生交通事故的最终的风险因子考虑了每个第一车辆与其相邻车辆的相对距离和相对速度,所以,对于预测每个第一车辆发生交通事故的风险因子更为准确。例如,每个第一车辆与其相邻车辆的距离较远时,当前车发生意外情况时,后车司机可以有相对较长的时间来采取措施,减少或避免发生交通事故。
在本申请实施例的一种可能的实现方式中,所述方法还包括:
判断所述每个第一车辆发生交通事故的最终的风险因子中的每种交通事故发生概率值是否大于预设第一阈值;如果判断结果为是,针对大于预设第一阈值的每个第一车辆,生成预警信息输出;和/或判断所述每个第一车辆发生交通事故的最终的风险因子中的每种交通事故的严重程度值是否大于预设第二阈值;如果判断结果为是,针对大于预设第二阈值的每个第一车辆,生成预警信息输出。
可以理解的是,大于预设第一阈值的第一车辆的数目可以是一辆或多辆,大于预设第二阈值的第一车辆的数目可以是一辆或多辆。
如图2a所示,本申请实施例所提供的一种道路发生交通事故的风险预测方法,包括以下步骤:
步骤S301~步骤S304与图1a所示的步骤S101~步骤S104相同,因此,这里不再赘述。
从步骤S305开始,所述道路发生交通事故的风险预测方法具体如下:
S305,针对所述每个第一车辆,根据所述目标路段发生交通事故的历史信息、所述每个第一车辆的车况信息和历史行为信息,以及所述环境状况信息,确定所述每个第一车辆发生交通事故的交通事故发生概率值;所述历史 信息包括:所述目标路段历史上发生的一种交通事故的第二车辆的车况信息和历史行为信息,以及发生该种交通事故时的环境状况信息;
步骤S305和图1a所示的S105类似,S305确定了每个第一车辆发生交通事故的交通事故发生概率值,而S105中确定了每个第一车辆发生交通事故的风险因子,该风险因子包括交通事故发生概率值和/或发生交通事故的严重程度值,所以,S305确定每个第一车辆发生交通事故的交通事故发生概率值的方法参照S105。
S306,对所述目标路段上的每个第一车辆的信息进行统计,获得所述目标路段上的所有车辆的平均车况信息、平均历史行为信息、发生交通事故的交通事故发生概率值的平均值、以及平均当前司机行为信息和所述目标路段上的道路状况信息;所述每个第一车辆的信息,包括:所述视频结构化信息、所述每个第一车辆的车况信息、所述每个第一车辆的历史行为信息、所述每个第一车辆发生交通事故的交通事故发生概率值;
具体的,根据S305中获得的每个第一车辆发生交通事故的交通事故发生概率值,确定目标路段上所有车辆发生交通事故的交通事故发生概率值的平均值,也就是,对目标路段上的所有车辆发生交通事故的交通事故发生概率值求平均值,得到目标路段上所有车辆发生交通事故的交通事故发生概率值的平均值。所述平均车况信息包括:平均车龄、平均历史违章记录次数、车型分布、平均年检次数等,但不限于此。所述平均历史行为信息包括:平均闯红灯次数、平均超速次数、平均违规变道次数等,但不限于此。所述平均当前司机行为信息包括:平均打手机次数、平均吃东西次数,平均不系安全带人数等,但不限于此。所述道路状况信息包括:平均车速、平均前后车的相对距离及平均前后车的相对车速,但不限于此。
实际应用中,获得所述平均车速的步骤包括:获取目标路段上每个第一车辆的速度,根据每个第一车辆的速度,获得目标路段上的平均车速;
获得所述平均前后车的相对距离的步骤包括:获得所述在道路全景图中,获得每个第一车辆的相对位置关系,根据该相对位置,获得每个第一车辆与相邻车辆的相对距离,根据每个第一车辆与相邻车辆的相对距离,获得目标路段上所有车辆的平均前后车的相对距离。
获得所述平均前后车的相对车速的步骤包括:根据道路全景图中每个第一车辆的相对位置关系,及每个第一车辆的速度,获得每个第一车辆与该第一车辆相邻车辆的相对车速,基于所述相对车速,计算获得目标路段上所有车辆的平均前后车的相对车速。
S307,将获得的所述目标路段上的所有车辆的平均车况信息、平均历史行为信息、发生交通事故的交通事故发生概率值的平均值、以及所有车辆的平均当前司机行为信息和所述目标路段上的道路状况信息,输入至预先设置的预测模型中,所述预测模型是预先针对历史上发生的一种交通事故,对每个发生该种交通事故所述目标路段上的所有车辆的平均车况信息、平均历史行为信息、发生交通事故的交通事故发生概率值的平均值、以及平均当前司机行为信息和所述目标路段上的道路状况信息,以及发生该种交通事故时的环境状况信息进行训练获得的。
将所获得的所有车辆的平均车况信息、平均历史行为信息、发生交通事故的交通事故发生概率值的平均值、以及所有车辆的平均当前司机行为信息和所述目标路段上的道路状况信息数值化,并将数值化后的所有车辆的平均车况信息、平均历史行为信息、发生交通事故的交通事故发生概率值的平均值、以及所有车辆的平均当前司机行为信息和所述目标路段上的道路状况信息,输入到预先设置的预测模型。需要注意的是,输入到该预测模型的信息都是数值化的信息。
S308,获得所述预测模型输出的道路发生交通事故的第一发生概率值;
预测模型根据输入其的数值化的所有车辆的平均车况信息、平均历史行为信息、发生交通事故的交通事故发生概率值的平均值、以及所有车辆的平均当前司机行为信息和所述目标路段上的道路状况信息,计算获得道路发生交通事故的第一发生概率值。
S309,将所述第一发生概率值确定为所述目标路段的道路发生交通事故的风险预测结果。
本申请实施例中,在所获得的每个第一车辆发生交通事故的交通事故发生概率值的基础上,计算获得该所有车辆发生交通事故的交通事故发生概率 值的平均值,并结合所有车辆的平均车况信息、平均历史行为信息、以及所有车辆的平均当前司机行为信息和道路状况信息,确定道路发生交通事故的第一概率值。这样,在确定道路发生交通事故的第一发生概率值时,考虑了所有车辆发生交通事故的交通事故发生概率值,提高了道路发生交通事故的风险预测的准确性。
如图2b所示,本申请实施例所提供的一种道路发生交通事故的风险预测方法,包括以下步骤:
步骤S401~步骤S405与图1b所示的步骤S201~步骤S205相同,因此,这里不再赘述。
从步骤S406开始,所述道路发生交通事故的风险预测方法具体如下:
S406,针对所述每个第一车辆,根据所述目标路段发生交通事故的历史信息、所述每个第一车辆的车况信息和历史行为信息,以及所述的环境状况信息,确定所述每个第一车辆发生交通事故的初始的交通事故发生概率值;
其中,S406和图1b所示的S206类似,S406确定每个第一车辆发生交通事故的交通事故发生概率值,而S206中确定每个第一车辆发生交通事故的初始的风险因子,该风险因子包括交通事故发生概率值和/或发生交通事故的严重程度值,所以S406中确定每个第一车辆发生交通事故的交通事故发生概率值的步骤参照S206。
S407,针对每个第一车辆,根据该第一车辆与其相邻车辆的相对距离和相对速度,对该第一车辆发生交通事故的初始的交通事故发生概率值进行修正,获得所述每个第一车辆发生交通事故的最终的交通事故发生概率值;将所获得的每个第一车辆发生交通事故的最终的交通事故发生概率值作为每个第一车辆发生交通事故的交通事故发生概率值;
其中,S407和图1b所示的S207类似,S407在考虑每个第一车辆与其相邻车辆的相对距离和相对速度的基础上,对每个第一车辆发生交通事故的初始的交通事故发生概率值进行修正,而S207中是在考虑每个第一车辆与其相邻车辆的相对距离和相对速度的基础上,对每个第一车辆发生交通事故的初始的风险因子进行修正,该风险因子包括交通事故发生概率值和/或发生交通 事故的严重程度值,所以S407中对每个第一车辆发生交通事故的初始的交通事故发生概率值进行修正的参照S207。
S408,对所述目标路段上的每个第一车辆的信息进行统计,获得所述目标路段上的所有车辆的平均车况信息、平均历史行为信息、发生交通事故的交通事故发生概率值的平均值、以及平均当前司机行为信息和所述目标路段上的道路状况信息;所述每个第一车辆的信息,包括:所述视频结构化信息、所述每个第一车辆的车况信息、所述每个第一车辆的历史行为信息、所述每个第一车辆发生交通事故的交通事故发生概率值;
其中,S408与图2a中的步骤S306相同,此处不再赘述。
S409,将获得的所述目标路段上的所有车辆的平均车况信息、平均历史行为信息、发生交通事故的交通事故发生概率值的平均值、以及所有车辆的平均当前司机行为信息和道路状况信息,输入至预先设置的预测模型中,所述预测模型是预先针对历史上发生的一种交通事故,对每个发生该种交通事故所述目标路段上的所有车辆的平均车况信息、平均历史行为信息、发生交通事故的交通事故发生概率值的平均值、以及平均当前司机行为信息和所述目标路段上的道路状况信息,以及发生该种交通事故时的环境状况信息进行训练获得的;
这里,需要对所获得的所述目标路段上的所有车辆的平均车况信息、平均历史行为信息、发生交通事故的交通事故发生概率值的平均值、以及所有车辆的平均当前司机行为信息和道路状况信息数值化,并将数值化后的所述目标路段上的所有车辆的平均车况信息、平均历史行为信息、发生交通事故的交通事故发生概率值的平均值、以及所有车辆的平均当前司机行为信息和道路状况信息输入至预先设置的预测模型,使得预测模型输出数值化的值。
S410,获得所述预测模型输出的道路发生交通事故的第一发生概率值;
S411,将所述第一发生概率值确定为所述目标路段的道路发生交通事故的风险预测结果。
本申请实施例中,对所获得的每个第一车辆发生交通事故的初始的交通事故发生概率值进行修正,获得每个第一车辆发生交通事故的最终的交通事 故发生概率值的基础上,计算获得该车辆发生交通事故的交通事故发生概率值的平均值,并结合所有车辆的平均车况信息、平均历史行为信息、以及所有车辆的平均当前司机行为信息和道路状况信息,确定道路发生交通事故的第一概率值。这样,在确定道路发生交通事故的第一发生概率值时,考虑了每个第一车辆发生交通事故的交通事故发生概率值,提高了道路发生交通事故的风险预测的准确性。
在本申请实施例中,所述视频结构化信息还包括:所述每个第一车辆的当前司机行为信息;所述历史信息还包括:发生每种交通事故时,所述第二车辆的当前司机行为信息;
针对所述每个第一车辆,根据所述目标路段发生交通事故的历史信息、所述每个第一车辆的车况信息和历史行为信息,以及所述环境状况信息,确定所述每个第一车辆发生交通事故的初始的交通事故发生概率值的步骤,包括:针对所述每个第一车辆,根据所述目标路段发生交通事故的历史信息、所述每个第一车辆的车况信息和历史行为信息,以及所述环境状况信息和所述每个第一车辆的当前司机行为信息,确定所述每个第一车辆发生交通事故的初始的交通事故发生概率值。可见,在确定每个第一车辆发生交通事故的初始的交通事故发生概率值时,需要将每个第一车辆的当前司机行为信息考虑进去,例如,当前司机行为信息包括:司机打电话,玩手机等,当司机开车的时候,玩手机,影响司机的注意力,使得当前的交通事故发生概率值的增加。
在本申请实施例的一中可能的实现方式中,针对所述每个第一车辆,根据所述目标路段发生交通事故的历史信息、所述每个第一车辆的车况信息和历史行为信息,以及所述环境状况信息和所述每个第一车辆的当前司机行为信息,确定所述每个第一车辆发生交通事故的初始的交通事故发生概率值,包括:
获取已存储的评分表,所述评分表是预先针对历史上发生的一种交通事故,对每个发生该种交通事故的第二车辆的所述历史信息进行评分的表;针对所述每个第一车辆,在所述评分表中查找该每个第一车辆的车况信息和历史行为信息,以及所述环境状况信息和所述每个第一车辆的当前司机行为信 息的对应评分;根据所述对应评分,获得所述每个第一车辆发生交通事故的交通事故发生概率值作为该第一车辆发生交通事故的初始的交通事故发生概率值。
在本申请实施例的一中可能的实现方式中,针对所述每个第一车辆,根据所述目标路段发生交通事故的历史信息、每个第一车辆的车况信息和历史行为信息,以及所述环境状况信息和所述每个第一车辆的当前司机行为信息,确定所述每个第一车辆发生交通事故的初始的交通事故发生概率值的步骤,包括:
针对所述每个第一车辆,将所述每个第一车辆的车况信息和历史行为信息,以及所述环境状况信息和所述每个第一车辆的当前司机行为信息,输入至多个预设的分类器中,每个分类器对应所述目标路段历史上发生的一种交通事故;所述每个分类器,是预先针对历史上发生的一种交通事故,对每个发生该种交通事故的第二车辆的车况信息和历史行为信息,以及发生该种交通事故时的环境状况信息和所述第二车辆的当前司机行为信息进行训练获得的;获得所述每个分类器输出的交通事故发生概率值作为该每个第一车辆发生交通事故的初始的交通事故发生概率值。
在本申请实施例,所述针对每个第一车辆,根据该第一车辆与其相邻车辆的相对距离和速度关系,对该第一车辆发生交通事故的初始的交通事故发生概率值进行修正,获得所述每个第一车辆发生交通事故的最终的交通事故发生概率值的步骤,包括:
针对每个第一车辆发生交通事故的初始的交通事故发生概率值,用如下公式分别进行计算,获得每个第一车辆发生交通事故的最终的交通事故发生概率值:
Figure PCTCN2017090874-appb-000011
Figure PCTCN2017090874-appb-000012
Figure PCTCN2017090874-appb-000013
其中,
Figure PCTCN2017090874-appb-000014
为第i辆车发生第t种交通事故的最终的交通事故发生概率值,
Figure PCTCN2017090874-appb-000015
为第i辆车发生第t种交通事故的初始的交通事故发生概率值,Ni为与第i辆车相邻的车辆数量或者与第i辆车距离小于第二预设值的车辆数目,d为所述道路全景图信息中的第i辆车和第j辆车之间的相对距离,v为第i辆车和第j辆车之间的相对速度,
Figure PCTCN2017090874-appb-000016
为第i辆车和第j辆车之间的相对距离影响因子,
Figure PCTCN2017090874-appb-000017
为第i辆车和第j辆车之间的相对速度影响因子,X为前车发生事故t类型导致后车发生τ事故类型或加剧后车事故严重程度的影响因子,
Figure PCTCN2017090874-appb-000018
为第j辆车发生第τ种交通事故的初始的交通事故发生概率值,T为事故类型集合,t为前车发生事故类型,τ为后车发生事故类型,a、β、γ、η、X分别为预设常数,均为0~1之间的任一小数;
将所获得的每个第一车辆发生交通事故的最终的交通事故发生概率值作为每个第一车辆发生交通事故的交通事故发生概率值。
在本申请实施例的一种可能的实现方式中,该道路发生交通事故的风险预测方法还包括:判断所述预测模型输出的道路发生交通事故的第一发生概率值是否大于预设第三阈值;如果判断结果为是,生成预警信息输出。
实际应用中,当确定的第一发生概率大于预设第三阈值时,就生成并输出预警信息。目标路段上的道路管理员、司机、交警等人员通过道路电子屏接收到预警信息,采取规避交通事故发生的措施,从而有效降低车辆及道路 发生交通事故的风险。
如图3a所示,本申请实施例所提供的一种车辆发生交通事故的风险预测装置,所述装置包括:
第一获得单元301a,用于获得目标路段上的监控视频;
处理单元302a,用于根据所述监控视频,获得视频结构化信息;所述视频结构化信息至少包括所述目标路段上每个第一车辆的车牌号码信息;
查找单元303a,用于根据所述车牌号码信息,在预设数据库中查询所述目标路段上每个第一车辆的车况信息和历史行为信息;
第二获得单元304a,用于获取目标路段在拍摄所述监控视频时的环境状况信息;
第一确定单元305a,用于针对所述每个第一车辆,根据所述目标路段发生交通事故的历史信息、所述每个第一车辆的车况信息和历史行为信息,以及所述环境状况信息,确定所述每个第一车辆发生交通事故的风险因子;所述风险因子包括:交通事故发生概率值和/或发生交通事故的严重程度值,所述历史信息包括:所述目标路段历史上发生的一种交通事故的第二车辆的车况信息和历史行为信息,以及发生该种交通事故时的环境状况信息;
第二确定单元306a,用于将所述风险因子确定为所述目标路段的车辆发生交通事故的风险预测结果。
本申请实施例,通过所获得目标路段上的监控视频,获得视频结构化信息,并根据该视频结构化信息中的车牌号码信息,在预设数据库中查询该目标路段上每个第一车辆的车况信息和历史行为信息,同时,获得拍摄所述监控视频时的环境状况信息,由所获得的车况信息和历史行为信息、环境状况信息,结合目标路段上发生交通事故的历史信息,确定每个第一车辆发生交通事故的风险因子,就能够获得车辆的风险预测结果。
如图3b所示,本申请实施例所提供的一种车辆发生交通事故的风险预测装置,该装置包括;
其中,第一获得单元301b、处理单元302b、查找单元303b及第二获得 单元304b与图3a所示的第一获得单元301a、处理单元302a、查找单元303a及第二获得单元304a相同,此处不再赘述。
接收单元305b,用于接收所述目标路段上的每个第一车辆的速度;
第三获得单元306b,用于根据所述监控视频,获得所述目标路段的道路全景图;
第四获得单元307b,用于根据所述道路全景图,获得所述每个第一车辆与其相邻车辆的相对距离;根据所述道路全景图和所述每个第一车辆的速度,获得所述每个第一车辆与其相邻车辆的相对速度;
第一确定单元308b,用于针对所述每个第一车辆,根据所述目标路段发生交通事故的历史信息、所述每个第一车辆的车况信息和历史行为信息,以及所述环境状况信息,确定所述每个第一车辆发生交通事故的初始的风险因子;
修正单元309b,用于针对每个第一车辆,根据该第一车辆与其相邻车辆的相对距离和相对速度,对该第一车辆发生交通事故的初始的风险因子进行修正,获得所述每个第一车辆发生交通事故的最终的风险因子;
第二确定单元310b,用于将所述每个第一车辆发生交通事故的最终的风险因子确定为所述目标路段的车辆发生交通事故的风险预测结果。
本申请实施例中,根据所获得的每个第一车辆与其相邻车辆的相对距离和相对速度,修正每个第一车辆发生交通事故的初始的风险因子,得到每个第一车辆发生交通事故的最终的风险因子,将最终的风险因子作为预测车辆发生交通事故的风险预测结果。可见,应用本申请实施,提高了车辆发生交通事故的风险预测的准确性。
在本申请实施例中,所述视频结构化信息还包括:所述每个第一车辆的当前司机行为信息;所述历史信息还包括:发生每种交通事故时,所述第二车辆的当前司机行为信息;
所述第一确定单元308b具体用于,针对所述每个第一车辆,根据所述目标路段发生交通事故的历史信息、所述每个第一车辆的车况信息和历史行为 信息,以及所述环境状况信息和所述每个第一车辆的当前司机行为信息,确定所述每个第一车辆发生交通事故的初始的风险因子。
所述第一确定子单元308b具体用于,获取已存储的评分表,所述评分表是预先针对历史上发生的一种交通事故,对每个发生该种交通事故的第二车辆的所述历史信息进行评分的表;针对所述每个第一车辆,在所述评分表中查找该每个第一车辆的车况信息和历史行为信息,以及所述环境状况信息和所述第一车辆的当前司机行为信息的对应评分;根据所述对应评分,获得所述每个第一车辆的交通事故发生概率值和/或发生交通事故的严重程度作为该第一车辆发生交通事故的初始的风险因子。
所述第一确定子单元308b具体用于针对所述每个第一车辆,将所述每个第一车辆的车况信息和历史行为信息,以及所述环境状况信息和所述每个第一车辆的当前司机行为信息,输入至多个预设的分类器中,每个分类器对应所述目标路段历史上发生的一种交通事故;所述每个分类器,是预先针对历史上发生的一种交通事故,对每个发生该种交通事故的第二车辆的车况信息和历史行为信息,以及发生该种交通事故时的环境状况信息和所述第二车辆的当前司机行为信息进行训练获得的;获得所述每个分类器输出的交通事故发生概率值和/或发生交通事故的严重程度值作为该第一车辆发生交通事故的初始的风险因子。
在本申请实施例的一种可能的实现方式中,所述修正子单元309b具体用于,
针对每个第一车辆发生交通事故的初始的风险因子中的每种交通事故发生概率值和/或发生交通事故的严重程度值,用如下公式分别进行计算,获得所述每个第一车辆发生交通事故的最终的风险因子:
Figure PCTCN2017090874-appb-000019
Figure PCTCN2017090874-appb-000020
Figure PCTCN2017090874-appb-000021
其中,
Figure PCTCN2017090874-appb-000022
为第i辆车发生第t种交通事故的最终的风险因子,
Figure PCTCN2017090874-appb-000023
为第i辆车发生第t种交通事故的初始的风险因子,Ni为与第i辆车相邻的车辆数量或者与第i辆车的距离小于第一预设值的车辆数目,d为所述道路全景图信息中的第i辆车和第j辆车之间的相对距离,v为所述道路全景图信息中第i辆车和第j辆车之间的相对速度,
Figure PCTCN2017090874-appb-000024
为第i辆车和第j辆车之间的相对距离影响因子,
Figure PCTCN2017090874-appb-000025
为第i辆车和第j辆车之间的相对速度影响因子,X为前车发生事故t类型导致后车发生τ事故类型或加剧后车事故严重程度的影响因子,
Figure PCTCN2017090874-appb-000026
为第j辆车发生第τ种交通事故的初始的风险因子,T为事故类型集合,t为前车发生事故类型,τ为后车发生事故类型,a、β、γ、η、X分别为预设常数,均为0~1之间的任一小数。
在本申请实施例的一种可能的实现方式中,所述装置还包括:
判断单元,用于判断所述每个第一车辆发生交通事故的最终的风险因子中的每种交通事故发生概率值是否大于预设第一阈值;如果判断结果为是,针对大于预设第一阈值的每个第一车辆,生成预警信息输出;和/或,判断所述每个第一车辆发生交通事故的最终的风险因子中的每种交通事故的严重程度值是否大于预设第二阈值;如果判断结果为是,针对大于预设第二阈值的每个第一车辆,生成预警信息输出。
如图4a所示,本申请实施例所提供的一种道路发生交通事故的风险预测 装置,所述装置包括:
其中,本申请实施例中的第一获得单元401a、处理单元402a、查找单元403a和第二获得单元404a分别与图3a所示实施例中的第一获得单元301a、处理单元302a、查找单元303a和第二获得单元304a相同,因此,这里不再赘述。
第一确定单元405a,用于针对所述每个第一车辆,根据所述目标路段发生交通事故的历史信息、所述每个第一车辆的车况信息和历史行为信息,以及所述环境状况信息,确定每个第一车辆发生交通事故的交通事故发生概率值;所述历史信息包括:所述目标路段历史上发生的一种交通事故的第二车辆的车况信息和历史行为信息,以及发生该种交通事故时的环境状况信息;
第三获得单元406a,用于对所述目标路段上的每个第一车辆的信息进行统计,获得所述目标路段上的所有车辆的平均车况信息、平均历史行为信息、发生交通事故的交通事故发生概率值的平均值、以及平均当前司机行为信息和所述目标路段上的道路信息;所述每个第一车辆的信息,包括:所述视频结构化信息、所述每个第一车辆的车况信息、所述每个第一车辆的历史行为信息、所述每个第一车辆发生交通事故的交通事故发生概率值;
输入单元407a,用于将获得的所述目标路段上的所有车辆的平均车况信息、平均历史行为信息、发生交通事故的交通事故发生概率值的平均值、以及所有车辆的平均当前司机行为信息和所述目标路段上的道路状况信息,输入至预先设置的预测模型中,所述预测模型是预先针对历史上发生的一种交通事故,对每个发生该种交通事故所述目标路段上的所有车辆的平均车况信息、平均历史行为信息、发生交通事故的交通事故发生概率值的平均值、以及平均当前司机行为信息和所述目标路段上的道路状况信息,以及发生该种交通事故时的环境状况信息进行训练获得的;
第四获得单元408a,用于获得所述预测模型输出的道路发生交通事故的第一发生概率值;
第二确定单元409a,用于将所述第一发生概率值确定为所述目标路段的道路发生交通事故的风险预测结果。
本申请实施例中,通过所获得目标路段上的监控视频,获得视频结构化信息,并根据该视频结构化信息中的车牌号码信息,在预设数据库中查询该目标路段上每个第一车辆的车况信息和历史行为信息,同时,获得拍摄所述监控视频时的环境状况信息,由所获得的车况信息和历史行为信息、环境状况信息,结合目标路段上发生交通事故的历史信息,确定每个第一车辆发生交通事故的交通事故发生概率值,并计算得到所有车辆发生交通事故的交通事故发生概率值的平均值,并结合所有车辆的平均车况信息、平均历史行为信息、以及所有车辆的平均当前司机行为信息和道路状况信息,确定道路发生交通事故的第一概率值作为目标路段的道路发生交通事故的风险预测结果。本申请实施例中,考虑了所有车辆发生交通事故的交通事故发生概率值的平均值,提高了道路发生交通事故的风险预测的准确性。
如图4b所示,本申请实施例所提供的一种道路发生交通事故的风险预测装置,该装置包括:
需要说明的是,本实施例中的第一获得单元401b、处理单元402b、查找单元403b和第二获得单元404b分别与图3a所示的第一获得单元301a、处理单元302a、查找单元303a和第二获得单元304a相同,因此,这里不再赘述。
接收单元405b,用于接收所述目标路段上的每个第一车辆的速度;
第三获得单元406b,用于根据所述监控视频,获得所述目标路段的道路全景图;
第四获得单元407b,具体用于根据所述道路全景图,获得所述每个第一车辆与其相邻车辆的相对距离;根据所述道路全景图和所述每个第一车辆的速度,获得所述每个第一车辆与其相邻车辆的相对速度;
第一确定单元408b,用于针对所述每个第一车辆,根据所述目标路段发生交通事故的历史信息、所述每个第一车辆的车况信息和历史行为信息,以及所述环境状况信息,确定所述每个第一车辆发生交通事故的初始的交通事故发生概率值;
修正单元409b,用于针对每个第一车辆,根据该第一车辆与其相邻车辆的相对速度和相对距离,对该第一车辆发生交通事故的初始的交通事故发生 概率值进行修正,获得所述每个第一车辆发生交通事故的最终的交通事故发生概率值;将所获得的每个第一车辆发生交通事故的最终的交通事故发生概率值作为每个第一车辆发生交通事故的交通事故发生概率值。
第五获得单元410b,用于对所述目标路段上的每个第一车辆的信息进行统计,获得所述目标路段上的所有车辆的平均车况信息、平均历史行为信息、发生交通事故的交通事故发生概率值的平均值、以及平均当前司机行为信息和所述目标路段上的道路信息;所述每个第一车辆的信息,包括:所述视频结构化信息、所述每个第一车辆的车况信息和历史行为信息、所述每个第一车辆发生交通事故的交通事故发生概率值;
输入单元411b,用于将获得的所述目标路段上的所有车辆的平均车况信息、平均历史行为信息、发生交通事故的交通事故发生概率值的平均值、以及所有车辆的平均当前司机行为信息和所述目标路段上的道路状况信息,输入至预先设置的预测模型中,所述预测模型是预先针对历史上发生的一种交通事故,对每个发生该种交通事故所述目标路段上的所有车辆的平均车况信息、平均历史行为信息、发生交通事故的交通事故发生概率值的平均值、以及平均当前司机行为信息和所述目标路段上的道路状况信息,以及发生该种交通事故时的环境状况信息进行训练获得的;
第六获得单元412b,用于获得所述预测模型输出的道路发生交通事故的第一发生概率值;
第二确定单元413b,用于将所述第一发生概率值确定为所述目标路段的道路发生交通事故的风险预测结果。
本申请实施例,根据所获得的每个第一车辆与其相邻车辆的相对距离和相对速度,修正每个第一车辆发生交通事故的初始的交通事故发生概率值,得到每个第一车辆发生交通事故的交通事故发生概率值,并结合所有车辆的平均车况信息、平均历史行为信息、以及所有车辆的平均当前司机行为信息和道路状况信息,确定道路发生交通事故的第一概率值作为目标路段的道路发生交通事故的风险预测结果。应用本申请实施例,提高了道路发生交通事故的风险预测的准确性。
具体的,所述视频结构化信息还包括:所述每个第一车辆的当前司机行为信息;所述历史信息还包括:发生每种交通事故时,所述第二车辆的当前司机行为信息;
所述第一确定单元408b,具体用于针对所述每个第一车辆,根据所述目标路段发生交通事故的历史信息、所述每个第一车辆的车况信息和历史行为信息,以及所述环境状况信息和所述每个第一车辆的当前司机行为信息,确定所述每个第一车辆发生交通事故的初始的交通事故发生概率值。
其中,所述第一确定单元408b,具体用于获取已存储的评分表,所述评分表是预先针对历史上发生的一种交通事故,对每个发生该种交通事故的第二车辆的所述历史信息进行评分的表;针对所述每个第一车辆,在所述评分表中查找该每个第一车辆的车况信息和历史行为信息,以及所述环境状况信息和所述每个第一车辆的当前司机行为信息的对应评分;根据所述对应评分,获得所述每个第一车辆发生交通事故的交通事故发生概率值作为该第一车辆发生交通事故的初始的交通事故发生概率值。
所述第一确定子单元408b具体用于,针对所述每个第一车辆,将所述每个第一车辆的车况信息和历史行为信息,以及所述环境状况信息和所述每个第一车辆的当前司机行为信息,输入至多个预设的分类器中,每个分类器对应所述目标路段历史上发生的一种交通事故;所述每个分类器,是预先针对历史上发生的一种交通事故,对每个发生该种交通事故的第二车辆的车况信息和历史行为信息,以及发生该种交通事故时的环境状况信息和所述第二车辆的当前司机行为信息进行训练获得的;获得所述每个分类器输出的交通事故发生概率值作为该每个第一车辆发生交通事故的初始的交通事故发生概率值。
所述修正子单元409b,具体用于针对每个第一车辆发生交通事故的初始的交通事故发生概率值,用如下公式分别进行计算,获得每个第一车辆发生交通事故的最终的交通事故发生概率值:
Figure PCTCN2017090874-appb-000027
Figure PCTCN2017090874-appb-000028
Figure PCTCN2017090874-appb-000029
其中,
Figure PCTCN2017090874-appb-000030
为第i辆车发生第t种交通事故的最终的交通事故发生概率值,
Figure PCTCN2017090874-appb-000031
为第i辆车发生第t种交通事故的初始的交通事故发生概率值,Ni为与第i辆车相邻的车辆数量或者与第i辆车距离小于第二预设值的车辆数目,d为所述道路全景图信息中的第i辆车和第j辆车之间的相对距离,v为第i辆车和第j辆车之间的相对速度,
Figure PCTCN2017090874-appb-000032
为第i辆车和第j辆车之间的相对距离影响因子,
Figure PCTCN2017090874-appb-000033
为第i辆车和第j辆车之间的相对速度影响因子,X为前车发生事故t类型导致后车发生τ事故类型或加剧后车事故严重程度的影响因子,
Figure PCTCN2017090874-appb-000034
为第j辆车发生第τ种交通事故的初始的交通事故发生概率值,T为事故类型集合,t为前车发生事故类型,τ为后车发生事故类型,a、β、γ、η、X分别为预设常数,均为0~1之间的任一小数。
在本申请实施例的一种可能的实现方式中,所述装置还包括:
判断单元,用于判断所述预测模型输出的道路发生交通事故的第一发生概率值是否大于预设第三阈值;用于如果判断结果为是,生成预警信息输出。
如图5a所示,本申请实施例所提供的一种车辆发生交通事故的风险预测系统,所述系统510包括:
视频监控设备511,用于采集目标路段上的监控视频;
车辆信息数据库512,用于存储所述目标路段上每个第一车辆的车况信息 和历史行为信息;
智能分析服务器513,用于获得所述视频监控设备511采集的目标路段上的监控视频;根据所述监控视频,获得视频结构化信息;所述视频结构化信息至少包括所述目标路段上每个第一车辆的车牌号码信息;根据所述车牌号码信息,在所述车辆信息数据库512中查询所述目标路段上每个第一车辆的车况信息和历史行为信息;在外部信息源系统514中获取目标路段在拍摄所述监控视频时的环境状况信息;
针对所述每个第一车辆,根据所述目标路段发生交通事故的历史信息、所述每个第一车辆的车况信息和历史行为信息,以及所述的环境状况信息,确定所述每个第一车辆发生交通事故的风险因子;所述风险因子包括:交通事故发生概率值和/或发生交通事故的严重程度值,所述历史信息包括:所述目标路段历史上发生的一种交通事故的第二车辆的车况信息和历史行为信息,以及发生该种交通事故时的环境状况信息;
将所述风险因子确定为所述目标路段的车辆发生交通事故的风险预测结果。
本申请实施例,通过所获得目标路段上的监控视频,获得视频结构化信息,并根据该视频结构化信息中的车牌号码信息,在预设数据库中查询该目标路段上每个第一车辆的车况信息和历史行为信息,同时,获得拍摄所述监控视频时的环境状况信息,由所获得的车况信息和历史行为信息、环境状况信息,结合目标路段上发生交通事故的历史信息,确定每个第一车辆发生交通事故的风险因子,就能够获得车辆的风险预测结果。
如图5b所示,本申请实施例所提供的一种车辆发生交通事故的风险预测系统,所述系统520包括:
其中,图5b中的视频监控设备521、车辆信息数据库522及智能分析服务器523与图5a中的视频监控设备511、车辆信息数据库512及智能分析服务器513相同,在此处不再赘述,其中,图5b中的外部信息源系统526不属于系统520。
传感器524,用于测量所述目标路段上的每个第一车辆的速度;
所述视频结构化信息还包括:所述每个第一车辆的当前司机行为信息;
所述历史信息还包括:发生每种交通事故时,所述第二车辆的当前司机行为信息;
所述智能分析服务器523具体用于,接收所述传感器524发送的所述目标路段上的每个第一车辆的速度;根据所述监控视频,获得所述目标路段的道路全景图;根据所述道路全景图,获得所述每个第一车辆与其相邻车辆的相对距离;根据所述道路全景图和所述每个第一车辆的速度,获得所述每个第一车辆与其相邻车辆的相对速度;
获取已存储的评分表,所述评分表是预先针对历史上发生的一种交通事故,对每个发生该种交通事故的第二车辆的所述历史信息进行评分的表;针对所述每个第一车辆,在所述评分表中查找该每个第一车辆的车况信息和历史行为信息,以及所述的环境状况信息和所述第一车辆的当前司机行为信息的对应评分;根据所述对应评分,获得所述每个第一车辆的交通事故发生概率值和/或发生交通事故的严重程度作为该第一车辆发生交通事故的初始的风险因子;
针对所述每个第一车辆,将所述每个第一车辆的车况信息和历史行为信息,以及所述的环境状况信息和所述每个第一车辆的当前司机行为信息,输入至多个预设的分类器中,每个分类器对应所述目标路段历史上发生的一种交通事故;所述每个分类器,是预先针对历史上发生的一种交通事故,对每个发生该种交通事故的第二车辆的车况信息和历史行为信息,以及发生该种交通事故时的环境状况信息和所述第二车辆的当前司机行为信息进行训练获得的;获得所述每个分类器输出的交通事故发生概率值和/或发生交通事故的严重程度值作为该第一车辆发生交通事故的初始的风险因子;
针对每个第一车辆发生交通事故的初始的风险因子中的每种交通事故发生概率值和/或发生交通事故的严重程度值,用如下公式分别进行计算,获得所述每个第一车辆发生交通事故的最终的风险因子:
Figure PCTCN2017090874-appb-000035
Figure PCTCN2017090874-appb-000036
Figure PCTCN2017090874-appb-000037
其中,
Figure PCTCN2017090874-appb-000038
为第i辆车发生第t种交通事故的最终的风险因子,
Figure PCTCN2017090874-appb-000039
为第i辆车发生第t种交通事故的初始的风险因子,Ni为与第i辆车相邻的车辆数量或者与第i辆车的距离小于第一预设值的车辆数目,d为所述道路全景图信息中的第i辆车和第j辆车之间的相对距离,v为所述道路全景图信息中第i辆车和第j辆车之间的相对速度,
Figure PCTCN2017090874-appb-000040
为第i辆车和第j辆车之间的相对距离影响因子,
Figure PCTCN2017090874-appb-000041
为第i辆车和第j辆车之间的相对速度影响因子,X为前车发生事故t类型导致后车发生τ事故类型或加剧后车事故严重程度的影响因子,
Figure PCTCN2017090874-appb-000042
为第j辆车发生第τ种交通事故的初始的风险因子,T为事故类型集合,t为前车发生事故类型,τ为后车发生事故类型,a、β、γ、η、X分别为预设常数,均为0~1之间的任一小数;
判断所述每个第一车辆发生交通事故的最终的风险因子中的每种交通事故发生概率值是否大于预设第一阈值;如果判断结果为是,针对大于预设第一阈值的每个第一车辆,生成预警信息输出;
和/或
判断所述每个第一车辆发生交通事故的最终的风险因子中的每种交通事故的严重程度值是否大于预设第二阈值;如果判断结果为是,针对大于预设第二阈值的每个第一车辆,生成预警信息输出。
预警设备525,用于接收所述智能分析服务器输出的预警信息。
本申请实施例中,对所获得的目标路段上监控视频进行分析,得到视频结构化信息和道路全景图,基于视频结构化信息中的车牌号码信息,在预设数据库中查询获得每个第一车辆的车况信息和历史行为信息,同时,获取目标路段上在拍摄监控视频时的环境状况信息,获得每个第一车辆的速度,并结合道路全景图,确定每个第一车辆与其相邻车辆的相对距离和相对速度,将所获得车况信息、历史行为信息、当前司机行为信息、环境状况信息、相对距离以及相对速度,输入到所述目标路段发生交通事故的历史信息训练的分类器,以使分类器输出每个第一车辆的发生交通事故的初始的风险因子,根据每个第一车辆与其相邻车辆的相对距离和相对速度,修正该初始的风险因子,得到每个第一车辆发生交通事故的最终的风险因子,将最终的风险因子作为预测车辆发生交通事故的风险预测结果。可见,本申请实施中,基于每个第一车辆与其相邻车辆的相对速度和相对距离,修正每个第一车辆发生交通事故的初始的风险因子,提高了车辆发生交通事故的风险预测的准确性。
如图6a所示,本申请实施例所提供的一种道路发生交通事故的风险预测系统,该系统610包括:
视频监控设备611,用于采集目标路段上的监控视频;
车辆信息数据库612,用于存储所述目标路段上每个第一车辆的车况信息和历史行为信息;
智能分析服务器613,用于获得所述视频监控设备611发送的所述目标路段上的监控视频;根据所述监控视频,获得视频结构化信息;所述视频结构化信息至少包括所述目标路段上每个第一车辆的车牌号码信息;根据所述车牌号码信息,在所述车辆信息数据库612中查询所述目标路段上每个第一车辆的车况信息和历史行为信息;在外部信息源系统614中获取目标路段在拍摄所述监控视频时的环境状况信息;
针对所述每个第一车辆,根据所述目标路段发生交通事故的历史信息、所述每个第一车辆的车况信息和历史行为信息,以及所述环境状况信息,确定所述每个第一车辆发生交通事故的交通事故发生概率值;所述历史信息包 括:所述目标路段历史上发生的一种交通事故的第二车辆的车况信息和历史行为信息,以及发生该种交通事故时的环境状况信息;
对所述目标路段上的每个第一车辆的信息进行统计,获得所述目标路段上的所有车辆的平均车况信息、平均历史行为信息、发生交通事故的交通事故发生概率值的平均值、以及平均当前司机行为信息和所述目标路段上的道路状况信息;所述每个第一车辆的信息,包括:所述视频结构化信息、所述每个第一车辆的车况信息、所述每个第一车辆的历史行为信息、所述每个第一车辆发生交通事故的交通事故发生概率值;
将获得的所述目标路段上的所有车辆的平均车况信息、平均历史行为信息、发生交通事故的交通事故发生概率值的平均值、以及所有车辆的平均当前司机行为信息和道路状况信息,输入至预先设置的预测模型中,所述预测模型是预先针对历史上发生的一种交通事故,对每个发生该种交通事故所述目标路段上的所有车辆的平均车况信息、平均历史行为信息、发生交通事故的交通事故发生概率值的平均值、以及平均当前司机行为信息和所述目标路段上的道路状况信息,以及发生该种交通事故时的环境状况信息进行训练获得的;
获得所述预测模型输出的道路发生交通事故的第一发生概率值;将所述第一发生概率值确定为所述目标路段的道路发生交通事故的风险预测结果。
本申请实施例中,在所获得的每个第一车辆发生交通事故的交通事故发生概率值的基础上,计算获得该所有车辆发生交通事故的交通事故发生概率值的平均值,并结合所有车辆的平均车况信息、平均历史行为信息、以及所有车辆的平均当前司机行为信息和道路状况信息,确定道路发生交通事故的第一概率值。这样,在确定道路发生交通事故的第一发生概率值时,考虑了所有车辆发生交通事故的交通事故发生概率值,提高了道路发生交通事故的风险预测的准确性。
如图6b所示,本申请实施例所提供的一种道路发生交通事故的风险预测系统,该系统620包括:
图6b中的视频监控设备621、车辆信息数据库622及智能分析服务器623 与图6a中的视频监控设备611、车辆信息数据库612及智能分析服务器613相同,在此处不再赘述,其中,图6b中的外部信息源系统不属于系统620。
传感器624,用于测量所述目标路段上的每个第一车辆的速度;
所述视频结构化信息还包括:所述每个第一车辆的当前司机行为信息;
所述历史信息还包括:发生每种交通事故时,所述第二车辆的当前司机行为信息;
所述智能分析服务器623具体用于,接收所述传感器624发送的所述目标路段上的每个第一车辆的速度;根据所述监控视频,获得所述目标路段的道路全景图;根据所述道路全景图,获得所述每个第一车辆与其相邻车辆的相对速度;根据所述道路全景图和所述每个第一车辆的速度,获得所述每个第一车辆与其相邻车辆的相对速度;
获取已存储的评分表,所述评分表是预先针对历史上发生的一种交通事故,对每个发生该种交通事故的第二车辆的所述历史信息进行评分的表;针对所述每个第一车辆,在所述评分表中查找该每个第一车辆的车况信息和历史行为信息,以及所述环境状况信息和所述第一车辆的当前司机行为信息的对应评分;根据所述对应评分,获得所述每个第一车辆的交通事故发生概率值作为该第一车辆发生交通事故的初始的交通事故发生概率值;或者
针对所述每个第一车辆,将所述每个第一车辆的车况信息和历史行为信息,以及所述环境状况信息和所述每个第一车辆的当前司机行为信息,输入至多个预设的分类器中,每个分类器对应所述目标路段历史上发生的一种交通事故;所述每个分类器,是预先针对历史上发生的一种交通事故,对每个发生该种交通事故的第二车辆的车况信息和历史行为信息,以及发生该种交通事故时的环境状况信息和所述第二车辆的当前司机行为信息进行训练获得的;获得所述每个分类器输出的交通事故发生概率值作为该每个第一车辆发生交通事故的初始的交通事故发生概率值;其中,所述历史信息包括:所述目标路段历史上发生的一种交通事故的第二车辆的车况信息和历史行为信息,以及发生该种交通事故时的环境状况信息和所述第二车辆的当前司机行为信息;
针对每个第一车辆发生交通事故的初始的交通事故发生概率值,用如下公式分别进行计算,获得每个第一车辆发生交通事故的最终的交通事故发生概率值:
Figure PCTCN2017090874-appb-000043
Figure PCTCN2017090874-appb-000044
Figure PCTCN2017090874-appb-000045
其中,
Figure PCTCN2017090874-appb-000046
为第i辆车发生第t种交通事故的最终的交通事故发生概率值,
Figure PCTCN2017090874-appb-000047
为第i辆车发生第t种交通事故的初始的交通事故发生概率值,Ni为与第i辆车相邻的车辆数量或者与第i辆车距离小于第二预设值的车辆数目,d为所述道路全景图信息中的第i辆车和第j辆车之间的相对距离,v为第i辆车和第j辆车之间的相对速度,
Figure PCTCN2017090874-appb-000048
为第i辆车和第j辆车之间的相对距离影响因子,
Figure PCTCN2017090874-appb-000049
为第i辆车和第j辆车之间的相对速度影响因子,X为前车发生事故t类型导致后车发生τ事故类型或加剧后车事故严重程度的影响因子,
Figure PCTCN2017090874-appb-000050
为第j辆车发生第τ种交通事故的初始的交通事故发生概率值,T为事故类型集合,t为前车发生事故类型,τ为后车发生事故类型,a、β、γ、η、X分别为预设常数,均为0~1之间的任一小数;
将所获得的每个第一车辆发生交通事故的最终的交通事故发生概率值作为每个第一车辆发生交通事故的交通事故发生概率值;
判断所述预测模型输出的道路发生交通事故的第一发生概率值是否大于预设第三阈值;如果判断结果为是,生成预警信息输出。
预警设备625,用于接收所述智能分析服务器输出的预警信息。
本申请实施例中,对所获得的每个第一车辆发生交通事故的初始的交通事故发生概率值进行修正,获得每个第一车辆发生交通事故的最终的交通事故发生概率值的基础上,计算获得该车辆发生交通事故的交通事故发生概率值的平均值,并结合所有车辆的平均车况信息、平均历史行为信息、以及所有车辆的平均当前司机行为信息和道路状况信息,确定道路发生交通事故的第一概率值。这样,在确定道路发生交通事故的第一发生概率值时,考虑了每个第一车辆发生交通事故的交通事故发生概率值,提高了道路发生交通事故的风险预测的准确性。
相应的,本申请还提供了一种存储介质,其中,该存储介质用于存储可执行程序代码,所述可执行程序代码用于在运行时执行本申请所述的一种车辆发生交通事故的风险预测方法及一种道路发生交通事故的风险预测方法。其中,本申请所述的一种车辆发生交通事故的风险预测方法,包括:
获得目标路段上的监控视频;
根据所述监控视频,获得视频结构化信息;所述视频结构化信息至少包括所述目标路段上每个第一车辆的车牌号码信息;
根据所述车牌号码信息,在预设数据库中查询所述目标路段上每个第一车辆的车况信息和历史行为信息;
获取目标路段在拍摄所述监控视频时的环境状况信息;
针对所述每个第一车辆,根据所述目标路段发生交通事故的历史信息、所述每个第一车辆的车况信息和历史行为信息,以及所述环境状况信息,确定所述每个第一车辆发生交通事故的风险因子;所述风险因子包括:交通事故发生概率值和/或发生交通事故的严重程度值,所述历史信息包括:所述目标路段历史上发生的一种交通事故的第二车辆的车况信息和历史行为信息,以及发生该种交通事故时的环境状况信息;
将所述风险因子确定为所述目标路段的车辆发生交通事故的风险预测结 果。
本申请实施例,通过所获得目标路段上的监控视频,获得视频结构化信息,并根据该视频结构化信息中的车牌号码信息,在预设数据库中查询该目标路段上每个第一车辆的车况信息和历史行为信息,同时,获得拍摄所述监控视频时的环境状况信息,由所获得的车况信息和历史行为信息、环境状况信息,结合目标路段上发生交通事故的历史信息,确定每个第一车辆发生交通事故的风险因子,就能够获得车辆的风险预测结果。
其中,本申请所述的一种道路发生交通事故的风险预测方法,包括:
获得目标路段上的监控视频;
根据所述监控视频,获得视频结构化信息;所述视频结构化信息至少包括所述目标路段上每个第一车辆的车牌号码信息;
根据所述车牌号码信息,在预设数据库中查询所述目标路段上每个第一车辆的车况信息和历史行为信息;
获取目标路段在拍摄所述监控视频时的环境状况信息;
针对所述每个第一车辆,根据所述目标路段发生交通事故的历史信息、所述每个第一车辆的车况信息和历史行为信息,以及所述环境状况信息,确定所述每个第一车辆发生交通事故的交通事故发生概率值;所述历史信息包括:所述目标路段历史上发生的一种交通事故的第二车辆的车况信息和历史行为信息,以及发生该种交通事故时的环境状况信息;
对所述目标路段上的每个第一车辆的信息进行统计,获得所述目标路段上的所有车辆的平均车况信息、平均历史行为信息、发生交通事故的交通事故发生概率值的平均值、以及平均当前司机行为信息和所述目标路段上的道路状况信息;所述每个第一车辆的信息,包括:所述视频结构化信息、所述每个第一车辆的车况信息和历史行为信息、所述每个第一车辆发生交通事故的交通事故发生概率值;
将获得的所述目标路段上的所有车辆的平均车况信息、平均历史行为信息、发生交通事故的交通事故发生概率值的平均值、以及平均当前司机行为 信息和所述目标路段上的道路状况信息,输入至预先设置的预测模型中,所述预测模型是预先针对历史上发生的一种交通事故,对每个发生该种交通事故所述目标路段上的所有车辆的平均车况信息、平均历史行为信息、发生交通事故的交通事故发生概率值的平均值、以及平均当前司机行为信息和所述目标路段上的道路状况信息,以及发生该种交通事故时的环境状况信息进行训练获得的;
获得所述预测模型输出的道路发生交通事故的第一发生概率值;
将所述第一发生概率值确定为所述目标路段的道路发生交通事故的风险预测结果。
本申请实施例中,对所获得的目标路段上监控视频进行分析,得到视频结构化信息和道路全景图,基于视频结构化信息中的车牌号码信息,在预设数据库中查询获得每个第一车辆的车况信息和历史行为信息,同时,获取目标路段上在拍摄监控视频时的环境状况信息,获得每个第一车辆的速度,并结合道路全景图,确定每个第一车辆与其相邻车辆的相对距离和相对速度,将所获得车况信息、历史行为信息、当前司机行为信息、环境状况信息、相对距离以及相对速度,输入到所述目标路段发生交通事故的历史信息训练的分类器,以使分类器输出每个第一车辆的发生交通事故的初始的风险因子,根据每个第一车辆与其相邻车辆的相对距离和相对速度,修正该初始的风险因子,得到每个第一车辆发生交通事故的最终的风险因子,将最终的风险因子作为预测车辆发生交通事故的风险预测结果。可见,本申请实施中,基于每个第一车辆与其相邻车辆的相对速度和相对距离,修正每个第一车辆发生交通事故的初始的风险因子,提高了车辆发生交通事故的风险预测的准确性。
相应的,本申请还提供了一种应用程序,其中,该应用程序用于在运行时执行如本申请所述的一种车辆发生交通事故的风险预测方法及一种道路发生交通事故的风险预测方法。其中,本申请所述的车辆发生交通事故的风险预测方法,包括:
获得目标路段上的监控视频;
根据所述监控视频,获得视频结构化信息;所述视频结构化信息至少包括所述目标路段上每个第一车辆的车牌号码信息;
根据所述车牌号码信息,在预设数据库中查询所述目标路段上每个第一车辆的车况信息和历史行为信息;
获取目标路段在拍摄所述监控视频时的环境状况信息;
针对所述每个第一车辆,根据所述目标路段发生交通事故的历史信息、所述每个第一车辆的车况信息和历史行为信息,以及所述环境状况信息,确定所述每个第一车辆发生交通事故的风险因子;所述风险因子包括:交通事故发生概率值和/或发生交通事故的严重程度值,所述历史信息包括:所述目标路段历史上发生的一种交通事故的第二车辆的车况信息和历史行为信息,以及发生该种交通事故时的环境状况信息;
将所述风险因子确定为所述目标路段的车辆发生交通事故的风险预测结果。
本申请实施例,通过所获得目标路段上的监控视频,获得视频结构化信息,并根据该视频结构化信息中的车牌号码信息,在预设数据库中查询该目标路段上每个第一车辆的车况信息和历史行为信息,同时,获得拍摄所述监控视频时的环境状况信息,由所获得的车况信息和历史行为信息、环境状况信息,结合目标路段上发生交通事故的历史信息,确定每个第一车辆发生交通事故的风险因子,就能够获得车辆的风险预测结果。
其中,本申请所述的一种道路发生交通事故的风险预测方法,包括:
获得目标路段上的监控视频;
根据所述监控视频,获得视频结构化信息;所述视频结构化信息至少包括所述目标路段上每个第一车辆的车牌号码信息;
根据所述车牌号码信息,在预设数据库中查询所述目标路段上每个第一车辆的车况信息和历史行为信息;
获取目标路段在拍摄所述监控视频时的环境状况信息;
针对所述每个第一车辆,根据所述目标路段发生交通事故的历史信息、 所述每个第一车辆的车况信息和历史行为信息,以及所述环境状况信息,确定所述每个第一车辆发生交通事故的交通事故发生概率值;所述历史信息包括:所述目标路段历史上发生的一种交通事故的第二车辆的车况信息和历史行为信息,以及发生该种交通事故时的环境状况信息;
对所述目标路段上的每个第一车辆的信息进行统计,获得所述目标路段上的所有车辆的平均车况信息、平均历史行为信息、发生交通事故的交通事故发生概率值的平均值、以及平均当前司机行为信息和所述目标路段上的道路状况信息;所述每个第一车辆的信息,包括:所述视频结构化信息、所述每个第一车辆的车况信息和历史行为信息、所述每个第一车辆发生交通事故的交通事故发生概率值;
将获得的所述目标路段上的所有车辆的平均车况信息、平均历史行为信息、发生交通事故的交通事故发生概率值的平均值、以及平均当前司机行为信息和所述目标路段上的道路状况信息,输入至预先设置的预测模型中,所述预测模型是预先针对历史上发生的一种交通事故,对每个发生该种交通事故所述目标路段上的所有车辆的平均车况信息、平均历史行为信息、发生交通事故的交通事故发生概率值的平均值、以及平均当前司机行为信息和所述目标路段上的道路状况信息,以及发生该种交通事故时的环境状况信息进行训练获得的;
获得所述预测模型输出的道路发生交通事故的第一发生概率值;
将所述第一发生概率值确定为所述目标路段的道路发生交通事故的风险预测结果。
本申请实施例中,对所获得的目标路段上监控视频进行分析,得到视频结构化信息和道路全景图,基于视频结构化信息中的车牌号码信息,在预设数据库中查询获得每个第一车辆的车况信息和历史行为信息,同时,获取目标路段上在拍摄监控视频时的环境状况信息,获得每个第一车辆的速度,并结合道路全景图,确定每个第一车辆与其相邻车辆的相对距离和相对速度,将所获得车况信息、历史行为信息、当前司机行为信息、环境状况信息、相对距离以及相对速度,输入到所述目标路段发生交通事故的历史信息训练的分类器,以使分类器输出每个第一车辆的发生交通事故的初始的风险因子, 根据每个第一车辆与其相邻车辆的相对距离和相对速度,修正该初始的风险因子,得到每个第一车辆发生交通事故的最终的风险因子,将最终的风险因子作为预测车辆发生交通事故的风险预测结果。可见,本申请实施中,基于每个第一车辆与其相邻车辆的相对速度和相对距离,修正每个第一车辆发生交通事故的初始的风险因子,提高了车辆发生交通事故的风险预测的准确性。
相应的,本申请还提供了一种电子设备,包括:
处理器、存储器、通信接口和总线;
所述处理器、所述存储器和所述通信接口通过所述总线连接并完成相互间的通信;
所述存储器存储可执行程序代码;
所述处理器通过读取所述存储器中存储的可执行程序代码来运行与所述可执行程序代码对应的程序,以用于执行如本申请所述的一种车辆发生交通事故的风险预测方法及一种道路发生交通事故的风险预测方法。其中,本申请所述的车辆发生交通事故的风险预测方法,包括:
获得目标路段上的监控视频;
根据所述监控视频,获得视频结构化信息;所述视频结构化信息至少包括所述目标路段上每个第一车辆的车牌号码信息;
根据所述车牌号码信息,在预设数据库中查询所述目标路段上每个第一车辆的车况信息和历史行为信息;
获取目标路段在拍摄所述监控视频时的环境状况信息;
针对所述每个第一车辆,根据所述目标路段发生交通事故的历史信息、所述每个第一车辆的车况信息和历史行为信息,以及所述环境状况信息,确定所述每个第一车辆发生交通事故的风险因子;所述风险因子包括:交通事故发生概率值和/或发生交通事故的严重程度值,所述历史信息包括:所述目标路段历史上发生的一种交通事故的第二车辆的车况信息和历史行为信息,以及发生该种交通事故时的环境状况信息;
将所述风险因子确定为所述目标路段的车辆发生交通事故的风险预测结果。
本申请实施例,通过所获得目标路段上的监控视频,获得视频结构化信息,并根据该视频结构化信息中的车牌号码信息,在预设数据库中查询该目标路段上每个第一车辆的车况信息和历史行为信息,同时,获得拍摄所述监控视频时的环境状况信息,由所获得的车况信息和历史行为信息、环境状况信息,结合目标路段上发生交通事故的历史信息,确定每个第一车辆发生交通事故的风险因子,就能够获得车辆的风险预测结果。
其中,本申请所述的一种道路发生交通事故的风险预测方法,包括:
获得目标路段上的监控视频;
根据所述监控视频,获得视频结构化信息;所述视频结构化信息至少包括所述目标路段上每个第一车辆的车牌号码信息;
根据所述车牌号码信息,在预设数据库中查询所述目标路段上每个第一车辆的车况信息和历史行为信息;
获取目标路段在拍摄所述监控视频时的环境状况信息;
针对所述每个第一车辆,根据所述目标路段发生交通事故的历史信息、所述每个第一车辆的车况信息和历史行为信息,以及所述环境状况信息,确定所述每个第一车辆发生交通事故的交通事故发生概率值;所述历史信息包括:所述目标路段历史上发生的一种交通事故的第二车辆的车况信息和历史行为信息,以及发生该种交通事故时的环境状况信息;
对所述目标路段上的每个第一车辆的信息进行统计,获得所述目标路段上的所有车辆的平均车况信息、平均历史行为信息、发生交通事故的交通事故发生概率值的平均值、以及平均当前司机行为信息和所述目标路段上的道路状况信息;所述每个第一车辆的信息,包括:所述视频结构化信息、所述每个第一车辆的车况信息和历史行为信息、所述每个第一车辆发生交通事故的交通事故发生概率值;
将获得的所述目标路段上的所有车辆的平均车况信息、平均历史行为信 息、发生交通事故的交通事故发生概率值的平均值、以及平均当前司机行为信息和所述目标路段上的道路状况信息,输入至预先设置的预测模型中,所述预测模型是预先针对历史上发生的一种交通事故,对每个发生该种交通事故所述目标路段上的所有车辆的平均车况信息、平均历史行为信息、发生交通事故的交通事故发生概率值的平均值、以及平均当前司机行为信息和所述目标路段上的道路状况信息,以及发生该种交通事故时的环境状况信息进行训练获得的;
获得所述预测模型输出的道路发生交通事故的第一发生概率值;
将所述第一发生概率值确定为所述目标路段的道路发生交通事故的风险预测结果。
本申请实施例中,对所获得的目标路段上监控视频进行分析,得到视频结构化信息和道路全景图,基于视频结构化信息中的车牌号码信息,在预设数据库中查询获得每个第一车辆的车况信息和历史行为信息,同时,获取目标路段上在拍摄监控视频时的环境状况信息,获得每个第一车辆的速度,并结合道路全景图,确定每个第一车辆与其相邻车辆的相对距离和相对速度,将所获得车况信息、历史行为信息、当前司机行为信息、环境状况信息、相对距离以及相对速度,输入到所述目标路段发生交通事故的历史信息训练的分类器,以使分类器输出每个第一车辆的发生交通事故的初始的风险因子,根据每个第一车辆与其相邻车辆的相对距离和相对速度,修正该初始的风险因子,得到每个第一车辆发生交通事故的最终的风险因子,将最终的风险因子作为预测车辆发生交通事故的风险预测结果。可见,本申请实施中,基于每个第一车辆与其相邻车辆的相对速度和相对距离,修正每个第一车辆发生交通事故的初始的风险因子,提高了车辆发生交通事故的风险预测的准确性。
对于装置/系统/存储介质/应用程序/电子设备实施例而言,由于其基本相似于方法实施例,所以描述的比较简单,相关之处参见方法实施例的部分说明即可。
需要说明的是,在本文中,诸如第一和第二等之类的关系术语仅仅用来将一个实体或者操作与另一个实体或操作区分开来,而不一定要求或者暗示这些实体或操作之间存在任何这种实际的关系或者顺序。而且,术语“包括”、 “包含”或者其任何其他变体意在涵盖非排他性的包含,从而使得包括一系列要素的过程、方法、物品或者设备不仅包括那些要素,而且还包括没有明确列出的其他要素,或者是还包括为这种过程、方法、物品或者设备所固有的要素。在没有更多限制的情况下,由语句“包括一个……”限定的要素,并不排除在包括所述要素的过程、方法、物品或者设备中还存在另外的相同要素。
本说明书中的各个实施例均采用相关的方式描述,各个实施例之间相同相似的部分互相参见即可,每个实施例重点说明的都是与其他实施例的不同之处。尤其,对于装置实施例而言,由于其基本相似于方法实施例,所以描述的比较简单,相关之处参见方法实施例的部分说明即可。
本领域普通技术人员可以理解实现上述方法实施方式中的全部或部分步骤是可以通过程序来指令相关的硬件来完成,所述的程序可以存储于计算机可读取存储介质中,这里所称得的存储介质,如:ROM/RAM、磁碟、光盘等。
以上所述仅为本申请的较佳实施例而已,并非用于限定本申请的保护范围。凡在本申请的精神和原则之内所作的任何修改、等同替换、改进等,均包含在本申请的保护范围内。

Claims (25)

  1. 一种车辆发生交通事故的风险预测方法,其特征在于,包括:
    获得目标路段上的监控视频;
    根据所述监控视频,获得视频结构化信息;所述视频结构化信息至少包括所述目标路段上每个第一车辆的车牌号码信息;
    根据所述车牌号码信息,在预设数据库中查询所述目标路段上每个第一车辆的车况信息和历史行为信息;
    获取目标路段在拍摄所述监控视频时的环境状况信息;
    针对所述每个第一车辆,根据所述目标路段发生交通事故的历史信息、所述每个第一车辆的车况信息和历史行为信息,以及所述环境状况信息,确定所述每个第一车辆发生交通事故的风险因子;所述风险因子包括:交通事故发生概率值和/或发生交通事故的严重程度值,所述历史信息包括:所述目标路段历史上发生的一种交通事故的第二车辆的车况信息和历史行为信息,以及发生该种交通事故时的环境状况信息;
    将所述风险因子确定为所述目标路段的车辆发生交通事故的风险预测结果。
  2. 根据权利要求1所述的方法,其特征在于,在所述针对所述每个第一车辆,根据所述目标路段发生交通事故的历史信息、所述每个第一车辆的车况信息和历史行为信息,以及所述环境状况信息,确定所述每个第一车辆发生交通事故的风险因子的步骤前,所述方法还包括:
    接收所述目标路段上的每个第一车辆的速度;
    根据所述监控视频,获得所述目标路段的道路全景图;
    根据所述道路全景图,获得所述每个第一车辆与其相邻车辆的相对距离;根据所述道路全景图和所述每个第一车辆的速度,获得所述每个第一车辆与其相邻车辆的相对速度;
    所述针对所述每个第一车辆,根据所述目标路段发生交通事故的历史信 息、所述每个第一车辆的车况信息和历史行为信息,以及所述环境状况信息,确定所述每个第一车辆发生交通事故的风险因子的步骤,包括:
    针对所述每个第一车辆,根据所述目标路段发生交通事故的历史信息、所述每个第一车辆的车况信息和历史行为信息,以及所述环境状况信息,确定所述每个第一车辆发生交通事故的初始的风险因子;
    针对每个第一车辆,根据该第一车辆与其相邻车辆的相对距离和相对速度,对该第一车辆发生交通事故的初始的风险因子中的交通事故发生概率值和/或发生交通事故的严重程度值进行修正,获得所述每个第一车辆发生交通事故的最终的风险因子。
  3. 根据权利要求2所述的方法,其特征在于,所述视频结构化信息还包括:所述每个第一车辆的当前司机行为信息;
    所述历史信息还包括:发生每种交通事故时,所述第二车辆的当前司机行为信息;
    针对所述每个第一车辆,根据所述目标路段发生交通事故的历史信息、所述每个第一车辆的车况信息和历史行为信息,以及所述环境状况信息,确定所述每个第一车辆发生交通事故的初始的风险因子的步骤,包括:
    针对所述每个第一车辆,根据所述目标路段发生交通事故的历史信息、所述每个第一车辆的车况信息和历史行为信息,以及所述环境状况信息和所述每个第一车辆的当前司机行为信息,确定所述每个第一车辆发生交通事故的初始的风险因子。
  4. 根据权利要求3所述的方法,其特征在于,所述针对所述每个第一车辆,根据所述目标路段发生交通事故的历史信息、所述每个第一车辆的车况信息和历史行为信息,以及所述环境状况信息和所述每个第一车辆的当前司机行为信息,确定所述每个第一车辆发生交通事故的初始的风险因子的步骤,包括:
    获取已存储的评分表,所述评分表是预先针对历史上发生的一种交通事故,对每个发生该种交通事故的第二车辆的所述历史信息进行评分的表;
    针对所述每个第一车辆,在所述评分表中查找该每个第一车辆的车况信息和历史行为信息,以及所述环境状况信息和所述每个第一车辆的当前司机行为信息的对应评分;
    根据所述对应评分,获得所述每个第一车辆的交通事故发生概率值和/或发生交通事故的严重程度作为该第一车辆发生交通事故的初始的风险因子。
  5. 根据权利要求3所述的方法,其特征在于,所述针对所述每个第一车辆,根据所述目标路段发生交通事故的历史信息、所述每个第一车辆的车况信息和历史行为信息,以及所述环境状况信息和所述每个第一车辆的当前司机行为信息,确定所述每个第一车辆发生交通事故的初始的风险因子的步骤,包括:
    针对所述每个第一车辆,将所述每个第一车辆的车况信息和历史行为信息,以及所述环境状况信息和所述每个第一车辆的当前司机行为信息,输入至多个预设的分类器中,每个分类器对应所述目标路段历史上发生的一种交通事故;所述每个分类器,是预先针对历史上发生的一种交通事故,对每个发生该种交通事故的第二车辆的车况信息和历史行为信息,以及发生该种交通事故时的环境状况信息和所述第二车辆的当前司机行为信息进行训练获得的;
    获得所述每个分类器输出的交通事故发生概率值和/或发生交通事故的严重程度值作为该第一车辆发生交通事故的初始的风险因子。
  6. 根据权利要求4或5所述方法,其特征在于,所述针对每个第一车辆,根据该第一车辆与其相邻车辆的相对距离和相对速度,对该第一车辆发生交通事故的初始的风险因子中的交通事故发生概率值和/或发生交通事故的严重程度值进行修正,获得所述每个第一车辆发生交通事故的最终的风险因子的步骤,包括:
    针对每个第一车辆发生交通事故的初始的风险因子中的每种交通事故发生概率值和/或发生交通事故的严重程度值,用如下公式分别进行计算,获得所述每个第一车辆发生交通事故的最终的风险因子:
    Figure PCTCN2017090874-appb-100001
    Figure PCTCN2017090874-appb-100002
    Figure PCTCN2017090874-appb-100003
    其中,
    Figure PCTCN2017090874-appb-100004
    为第i辆车发生第t种交通事故的最终的风险因子,
    Figure PCTCN2017090874-appb-100005
    为第i辆车发生第t种交通事故的初始的风险因子,Ni为与第i辆车相邻的车辆数量或者与第i辆车的距离小于第一预设值的车辆数目,d为所述道路全景图信息中的第i辆车和第j辆车之间的相对距离,v为所述道路全景图信息中第i辆车和第j辆车之间的相对速度,
    Figure PCTCN2017090874-appb-100006
    为第i辆车和第j辆车之间的相对距离影响因子,
    Figure PCTCN2017090874-appb-100007
    为第i辆车和第j辆车之间的相对速度影响因子,X为前车发生事故t类型导致后车发生τ事故类型或加剧后车事故严重程度的影响因子,
    Figure PCTCN2017090874-appb-100008
    为第j辆车发生第τ种交通事故的初始的风险因子,T为事故类型集合,t为前车发生事故类型,τ为后车发生事故类型,a、β、γ、η、X分别为预设常数,均为0~1之间的任一小数。
  7. 根据权利要求6所述的方法,其特征在于,该方法还包括:
    判断所述每个第一车辆发生交通事故的最终的风险因子中的每种交通事故发生概率值是否大于预设第一阈值;
    如果判断结果为是,针对大于预设第一阈值的每个第一车辆,生成预警信息输出;
    和/或
    判断所述每个第一车辆发生交通事故的最终的风险因子中的每种交通事 故的严重程度值是否大于预设第二阈值;
    如果判断结果为是,针对大于预设第二阈值的每个第一车辆,生成预警信息输出。
  8. 一种道路发生交通事故的风险预测方法,其特征在于,包括:
    获得目标路段上的监控视频;
    根据所述监控视频,获得视频结构化信息;所述视频结构化信息至少包括所述目标路段上每个第一车辆的车牌号码信息;
    根据所述车牌号码信息,在预设数据库中查询所述目标路段上每个第一车辆的车况信息和历史行为信息;
    获取目标路段在拍摄所述监控视频时的环境状况信息;
    针对所述每个第一车辆,根据所述目标路段发生交通事故的历史信息、所述每个第一车辆的车况信息和历史行为信息,以及所述环境状况信息,确定所述每个第一车辆发生交通事故的交通事故发生概率值;所述历史信息包括:所述目标路段历史上发生的一种交通事故的第二车辆的车况信息和历史行为信息,以及发生该种交通事故时的环境状况信息;
    对所述目标路段上的每个第一车辆的信息进行统计,获得所述目标路段上的所有车辆的平均车况信息、平均历史行为信息、发生交通事故的交通事故发生概率值的平均值、以及平均当前司机行为信息和所述目标路段上的道路状况信息;所述每个第一车辆的信息,包括:所述视频结构化信息、所述每个第一车辆的车况信息和历史行为信息、所述每个第一车辆发生交通事故的交通事故发生概率值;
    将获得的所述目标路段上的所有车辆的平均车况信息、平均历史行为信息、发生交通事故的交通事故发生概率值的平均值、以及平均当前司机行为信息和所述目标路段上的道路状况信息,输入至预先设置的预测模型中,所述预测模型是预先针对历史上发生的一种交通事故,对每个发生该种交通事故所述目标路段上的所有车辆的平均车况信息、平均历史行为信息、发生交通事故的交通事故发生概率值的平均值、以及平均当前司机行为信息和所述 目标路段上的道路状况信息,以及发生该种交通事故时的环境状况信息进行训练获得的;
    获得所述预测模型输出的道路发生交通事故的第一发生概率值;
    将所述第一发生概率值确定为所述目标路段的道路发生交通事故的风险预测结果。
  9. 根据权利要求8所述的方法,其特征在于,在所述针对所述每个第一车辆,根据所述目标路段发生交通事故的历史信息、所述每个第一车辆的车况信息和历史行为信息,以及所述环境状况信息,确定所述每个第一车辆发生交通事故的交通事故发生概率值的步骤前,该方法还包括:
    接收所述目标路段上的每个第一车辆的速度;
    根据所述监控视频,获得所述目标路段的道路全景图;
    根据所述道路全景图,获得所述每个第一车辆与其相邻车辆的相对速度;根据所述道路全景图和所述每个第一车辆的速度,获得所述每个第一车辆与其相邻车辆的相对速度;
    所述针对所述每个第一车辆,根据所述目标路段发生交通事故的历史信息、所述每个第一车辆的车况信息和历史行为信息,以及所述环境状况信息,确定所述每个第一车辆发生交通事故的交通事故发生概率值的步骤,包括:
    针对所述每个第一车辆,根据所述目标路段发生交通事故的历史信息、所述每个第一车辆的车况信息和历史行为信息,以及所述环境状况信息,确定所述每个第一车辆发生交通事故的初始的交通事故发生概率值;
    针对每个第一车辆,根据该第一车辆与其相邻车辆的相对距离和相对速度,对该第一车辆发生交通事故的初始的交通事故发生概率值进行修正,获得所述每个第一车辆发生交通事故的最终的交通事故发生概率值;
    将所述每个第一车辆发生交通事故的最终的交通事故发生概率值确定为所述每个第一车辆发生交通事故的交通事故发生概率值。
  10. 根据权利要求9所述的方法,其特征在于,包括:
    所述视频结构化信息还包括:所述每个第一车辆的当前司机行为信息;
    所述历史信息还包括:发生每种交通事故时,所述第二车辆的当前司机行为信息;
    针对所述每个第一车辆,根据所述目标路段发生交通事故的历史信息、所述每个第一车辆的车况信息和历史行为信息,以及所述环境状况信息,确定所述每个第一车辆发生交通事故的初始的交通事故发生概率值的步骤,包括:
    针对所述每个第一车辆,根据所述目标路段发生交通事故的历史信息、所述每个第一车辆的车况信息和历史行为信息,以及所述环境状况信息和所述每个第一车辆的当前司机行为信息,确定所述每个第一车辆发生交通事故的初始的交通事故发生概率值。
  11. 根据权利要求10所述的方法,其特征在于,所述针对所述每个第一车辆,根据所述目标路段发生交通事故的历史信息、所述每个第一车辆的车况信息和历史行为信息,以及所述环境状况信息和所述每个第一车辆的当前司机行为信息,确定所述每个第一车辆发生交通事故的初始的交通事故发生概率值,包括:
    获取已存储的评分表,所述评分表是预先针对历史上发生的一种交通事故,对每个发生该种交通事故的第二车辆的所述历史信息进行评分的表;
    针对所述每个第一车辆,在所述评分表中查找该每个第一车辆的车况信息和历史行为信息,以及所述环境状况信息和所述每个第一车辆的当前司机行为信息的对应评分;
    根据所述对应评分,获得所述每个第一车辆发生交通事故的交通事故发生概率值作为该第一车辆发生交通事故的初始的交通事故发生概率值。
  12. 根据权利要求10所述的方法,其特征在于,所述针对所述每个第一车辆,根据所述目标路段发生交通事故的历史信息、每个第一车辆的车况信息和历史行为信息,以及所述环境状况信息和所述每个第一车辆的当前司机行为信息,确定所述每个第一车辆发生交通事故的初始的交通事故发生概率值的步骤,包括:
    针对所述每个第一车辆,将所述每个第一车辆的车况信息和历史行为信息,以及所述环境状况信息和所述每个第一车辆的当前司机行为信息,输入至多个预设的分类器中,每个分类器对应所述目标路段历史上发生的一种交通事故;所述每个分类器,是预先针对历史上发生的一种交通事故,对每个发生该种交通事故的第二车辆的车况信息和历史行为信息,以及发生该种交通事故时的环境状况信息和所述第二车辆的当前司机行为信息进行训练获得的;
    获得所述每个分类器输出的交通事故发生概率值作为该每个第一车辆发生交通事故的初始的交通事故发生概率值。
  13. 根据权利要求11或12所述方法,其特征在于,所述针对每个第一车辆,根据该第一车辆与其相邻车辆的相对距离和速度关系,对该第一车辆发生交通事故的初始的交通事故发生概率值进行修正,获得所述每个第一车辆发生交通事故的最终的交通事故发生概率值的步骤,包括:
    针对每个第一车辆发生交通事故的初始的交通事故发生概率值,用如下公式分别进行计算,获得每个第一车辆发生交通事故的最终的交通事故发生概率值:
    Figure PCTCN2017090874-appb-100009
    Figure PCTCN2017090874-appb-100010
    Figure PCTCN2017090874-appb-100011
    其中,
    Figure PCTCN2017090874-appb-100012
    为第i辆车发生第t种交通事故的最终的交通事故发生概率值,
    Figure PCTCN2017090874-appb-100013
    为第i辆车发生第t种交通事故的初始的交通事故发生概率值,Ni为与第i辆车相邻的车辆数量或者与第i辆车距离小于第二预设值的车辆数目,d为所述道路全景图信息中的第i辆车和第j辆车之间的相对距离,v为第i辆车 和第j辆车之间的相对速度,
    Figure PCTCN2017090874-appb-100014
    为第i辆车和第j辆车之间的相对距离影响因子,
    Figure PCTCN2017090874-appb-100015
    为第i辆车和第j辆车之间的相对速度影响因子,X为前车发生事故t类型导致后车发生τ事故类型或加剧后车事故严重程度的影响因子,
    Figure PCTCN2017090874-appb-100016
    为第j辆车发生第τ种交通事故的初始的交通事故发生概率值,T为事故类型集合,t为前车发生事故类型,τ为后车发生事故类型,a、β、γ、η、X分别为预设常数,均为0~1之间的任一小数。
  14. 根据权利要求8所述的方法,其特征在于,所述方法还包括:
    判断所述预测模型输出的道路发生交通事故的第一发生概率值是否大于预设第三阈值;
    如果判断结果为是,生成预警信息输出。
  15. 一种车辆发生交通事故的风险预测装置,其特征在于,包括:
    第一获得单元,用于获得目标路段上的监控视频;
    处理单元,用于根据所述监控视频,获得视频结构化信息;所述视频结构化信息至少包括所述目标路段上每个第一车辆的车牌号码信息;
    查找单元,用于根据所述车牌号码信息,在预设数据库中查询所述目标路段上每个第一车辆的车况信息和历史行为信息;
    第二获得单元,用于获取目标路段在拍摄所述监控视频时的环境状况信息;
    第一确定单元,用于针对所述每个第一车辆,根据所述目标路段发生交通事故的历史信息、所述每个第一车辆的车况信息和历史行为信息,以及所述环境状况信息,确定所述每个第一车辆发生交通事故的风险因子;所述风险因子包括:交通事故发生概率值和/或发生交通事故的严重程度值,所述历史信息包括:所述目标路段历史上发生的一种交通事故的第二车辆的车况信息和历史行为信息,以及发生该种交通事故时的环境状况信息;
    第二确定单元,用于将所述风险因子确定为所述目标路段的车辆发生交 通事故的风险预测结果。
  16. 一种道路发生交通事故的风险预测装置,其特征在于,包括:
    第一获得单元,用于获得目标路段上的监控视频;
    处理单元,用于根据所述监控视频,获得视频结构化信息;所述视频结构化信息至少包括所述目标路段上每个第一车辆的车牌号码信息;
    查找单元,用于根据所述车牌号码信息,在预设数据库中查询所述目标路段上每个第一车辆的车况信息和历史行为信息;
    第二获得单元,用于获取目标路段在拍摄所述监控视频时的环境状况信息;
    第一确定单元,用于针对所述每个第一车辆,根据所述目标路段发生交通事故的历史信息、所述每个第一车辆的车况信息和历史行为信息,以及所述环境状况信息,确定每个第一车辆发生交通事故的交通事故发生概率值;所述历史信息包括:所述目标路段历史上发生的一种交通事故的第二车辆的车况信息和历史行为信息,以及发生该种交通事故时的环境状况信息;
    第三获得单元,用于对所述目标路段上的每个第一车辆的信息进行统计,获得所述目标路段上的所有车辆的平均车况信息、平均历史行为信息、发生交通事故的交通事故发生概率值的平均值、以及平均当前司机行为信息和所述目标路段上的道路信息;所述每个第一车辆的信息,包括:所述视频结构化信息、所述每个第一车辆的车况信息和历史行为信息、所述每个第一车辆发生交通事故的交通事故发生概率值;
    输入单元,用于将获得的所述目标路段上的所有车辆的平均车况信息、平均历史行为信息、发生交通事故的交通事故发生概率值的平均值、以及所有车辆的平均当前司机行为信息和所述目标路段上的道路状况信息,输入至预先设置的预测模型中,所述预测模型是预先针对历史上发生的一种交通事故,对每个发生该种交通事故所述目标路段上的所有车辆的平均车况信息、平均历史行为信息、发生交通事故的交通事故发生概率值的平均值、以及平均当前司机行为信息和所述目标路段上的道路状况信息,以及发生该种交通事故时的环境状况信息进行训练获得的;
    第四获得单元,用于获得所述预测模型输出的道路发生交通事故的第一发生概率值;
    第二确定单元,用于将所述第一发生概率值确定为所述目标路段的道路发生交通事故的风险预测结果。
  17. 一种车辆发生交通事故的风险预测系统,其特征在于,所述系统包括:
    视频监控设备,用于采集目标路段上的监控视频;
    车辆信息数据库,用于存储所述目标路段上每个第一车辆的车况信息和历史行为信息;
    智能分析服务器,用于获得所述视频监控设备采集的目标路段上的监控视频;根据所述监控视频,获得视频结构化信息;所述视频结构化信息至少包括所述目标路段上每个第一车辆的车牌号码信息;根据所述车牌号码信息,在所述车辆信息数据库中查询所述目标路段上每个第一车辆的车况信息和历史行为信息;在外部信息源系统中获取目标路段在拍摄所述监控视频时的环境状况信息;
    针对所述每个第一车辆,根据所述目标路段发生交通事故的历史信息、所述每个第一车辆的车况信息和历史行为信息,以及所述的环境状况信息,确定所述每个第一车辆发生交通事故的风险因子;所述风险因子包括:交通事故发生概率值和/或发生交通事故的严重程度值,所述历史信息包括:所述目标路段历史上发生的一种交通事故的第二车辆的车况信息和历史行为信息,以及发生该种交通事故时的环境状况信息;
    将所述风险因子确定为所述目标路段的车辆发生交通事故的风险预测结果。
  18. 根据权利要求17所述的系统,其特征在于,所述系统还包括:
    传感器,用于测量所述目标路段上的每个第一车辆的速度;
    所述视频结构化信息还包括:所述每个第一车辆的当前司机行为信息;
    所述历史信息还包括:发生每种交通事故时,所述第二车辆的当前司机 行为信息;
    所述智能分析服务器具体用于,接收所述传感器发送的所述目标路段上的每个第一车辆的速度;根据所述监控视频,获得所述目标路段的道路全景图;根据所述道路全景图,获得所述每个第一车辆与其相邻车辆的相对距离;根据所述道路全景图和所述每个第一车辆的速度,获得所述每个第一车辆与其相邻车辆的相对速度;
    获取已存储的评分表,所述评分表是预先针对历史上发生的一种交通事故,对每个发生该种交通事故的第二车辆的所述历史信息进行评分的表;针对所述每个第一车辆,在所述评分表中查找该每个第一车辆的车况信息和历史行为信息,以及所述的环境状况信息和所述第一车辆的当前司机行为信息的对应评分;根据所述对应评分,获得所述每个第一车辆的交通事故发生概率值和/或发生交通事故的严重程度作为该第一车辆发生交通事故的初始的风险因子;或者,
    针对所述每个第一车辆,将所述每个第一车辆的车况信息和历史行为信息,以及所述的环境状况信息和所述每个第一车辆的当前司机行为信息,输入至多个预设的分类器中,每个分类器对应所述目标路段历史上发生的一种交通事故;所述每个分类器,是预先针对历史上发生的一种交通事故,对每个发生该种交通事故的第二车辆的车况信息和历史行为信息,以及发生该种交通事故时的环境状况信息和所述第二车辆的当前司机行为信息进行训练获得的;获得所述每个分类器输出的交通事故发生概率值和/或发生交通事故的严重程度值作为该第一车辆发生交通事故的初始的风险因子;
    针对每个第一车辆发生交通事故的初始的风险因子中的每种交通事故发生概率值和/或发生交通事故的严重程度值,用如下公式分别进行计算,获得所述每个第一车辆发生交通事故的最终的风险因子:
    Figure PCTCN2017090874-appb-100017
    Figure PCTCN2017090874-appb-100018
    Figure PCTCN2017090874-appb-100019
    其中,
    Figure PCTCN2017090874-appb-100020
    为第i辆车发生第t种交通事故的最终的风险因子,
    Figure PCTCN2017090874-appb-100021
    为第i辆车发生第t种交通事故的初始的风险因子,Ni为与第i辆车相邻的车辆数量或者与第i辆车的距离小于第一预设值的车辆数目,d为所述道路全景图信息中的第i辆车和第j辆车之间的相对距离,v为所述道路全景图信息中第i辆车和第j辆车之间的相对速度,
    Figure PCTCN2017090874-appb-100022
    为第i辆车和第j辆车之间的相对距离影响因子,
    Figure PCTCN2017090874-appb-100023
    为第i辆车和第j辆车之间的相对速度影响因子,X为前车发生事故t类型导致后车发生τ事故类型或加剧后车事故严重程度的影响因子,
    Figure PCTCN2017090874-appb-100024
    为第j辆车发生第τ种交通事故的初始的风险因子,T为事故类型集合,t为前车发生事故类型,τ为后车发生事故类型,a、β、γ、η、X分别为预设常数,均为0~1之间的任一小数;
    判断所述每个第一车辆发生交通事故的最终的风险因子中的每种交通事故发生概率值是否大于预设第一阈值;如果判断结果为是,针对大于预设第一阈值的每个第一车辆,生成预警信息输出;和/或
    判断所述每个第一车辆发生交通事故的最终的风险因子中的每种交通事故的严重程度值是否大于预设第二阈值;如果判断结果为是,针对大于预设第二阈值的每个第一车辆,生成预警信息输出。
  19. 根据权利要求18所述的系统,其特征在于,所述系统还包括:
    预警设备,用于接收所述智能分析服务器输出的预警信息。
  20. 一种道路发生交通事故的风险预测系统,其特征在于,所述系统包括:
    视频监控设备,用于采集目标路段上的监控视频;
    车辆信息数据库,用于存储所述目标路段上每个第一车辆的车况信息和历史行为信息;
    智能分析服务器,用于获得所述视频监控设备采集的所述目标路段上的监控视频;根据所述监控视频,获得视频结构化信息;所述视频结构化信息至少包括所述目标路段上每个第一车辆的车牌号码信息;根据所述车牌号码信息,在所述车辆信息数据库中查询所述目标路段上每个第一车辆的车况信息和历史行为信息;在外部信息源系统中获取目标路段在拍摄所述监控视频时的环境状况信息;
    针对所述每个第一车辆,根据所述目标路段发生交通事故的历史信息、所述每个第一车辆的车况信息和历史行为信息,以及所述环境状况信息,确定所述每个第一车辆发生交通事故的交通事故发生概率值;所述历史信息包括:所述目标路段历史上发生的一种交通事故的第二车辆的车况信息和历史行为信息,以及发生该种交通事故时的环境状况信息;
    对所述目标路段上的每个第一车辆的信息进行统计,获得所述目标路段上的所有车辆的平均车况信息、平均历史行为信息、发生交通事故的交通事故发生概率值的平均值、以及平均当前司机行为信息和所述目标路段上的道路状况信息;所述每个第一车辆的信息,包括:所述视频结构化信息、所述每个第一车辆的车况信息、所述每个第一车辆的历史行为信息、所述每个第一车辆发生交通事故的交通事故发生概率值;
    将获得的所述目标路段上的所有车辆的平均车况信息、平均历史行为信息、发生交通事故的交通事故发生概率值的平均值、以及所有车辆的平均当前司机行为信息和道路状况信息,输入至预先设置的预测模型中,所述预测模型是预先针对历史上发生的一种交通事故,对每个发生该种交通事故所述目标路段上的所有车辆的平均车况信息、平均历史行为信息、发生交通事故的交通事故发生概率值的平均值、以及平均当前司机行为信息和所述目标路 段上的道路状况信息,以及发生该种交通事故时的环境状况信息进行训练获得的;
    获得所述预测模型输出的道路发生交通事故的第一发生概率值;将所述第一发生概率值确定为所述目标路段的道路发生交通事故的风险预测结果。
  21. 根据权利要求20所述的系统,其特征在于,所述系统还包括:
    传感器,用于测量所述目标路段上的每个第一车辆的速度;
    所述视频结构化信息还包括:所述每个第一车辆的当前司机行为信息;
    所述历史信息还包括:发生每种交通事故时,所述第二车辆的当前司机行为信息;
    所述智能分析服务器具体用于,接收所述传感器发送的所述目标路段上的每个第一车辆的速度;根据所述监控视频,获得所述目标路段的道路全景图;根据所述道路全景图,获得所述每个第一车辆与其相邻车辆的相对速度;根据所述道路全景图和所述每个第一车辆的速度,获得所述每个第一车辆与其相邻车辆的相对速度;
    获取已存储的评分表,所述评分表是预先针对历史上发生的一种交通事故,对每个发生该种交通事故的第二车辆的所述历史信息进行评分的表;针对所述每个第一车辆,在所述评分表中查找该每个第一车辆的车况信息和历史行为信息,以及所述环境状况信息和所述第一车辆的当前司机行为信息的对应评分;根据所述对应评分,获得所述每个第一车辆的交通事故发生概率值作为该第一车辆发生交通事故的初始的交通事故发生概率值;
    针对所述每个第一车辆,将所述每个第一车辆的车况信息和历史行为信息,以及所述环境状况信息和所述每个第一车辆的当前司机行为信息,输入至多个预设的分类器中,每个分类器对应所述目标路段历史上发生的一种交通事故;所述每个分类器,是预先针对历史上发生的一种交通事故,对每个发生该种交通事故的第二车辆的车况信息和历史行为信息,以及发生该种交通事故时的环境状况信息和所述第二车辆的当前司机行为信息进行训练获得的;
    获得所述每个分类器输出的交通事故发生概率值作为该每个第一车辆发生交通事故的初始的交通事故发生概率值;其中,所述历史信息包括:所述目标路段历史上发生的一种交通事故的第二车辆的车况信息和历史行为信息,以及发生该种交通事故时的环境状况信息和所述第二车辆的当前司机行为信息;
    针对每个第一车辆发生交通事故的初始的交通事故发生概率值,用如下公式分别进行计算,获得每个第一车辆发生交通事故的最终的交通事故发生概率值:
    Figure PCTCN2017090874-appb-100025
    Figure PCTCN2017090874-appb-100026
    Figure PCTCN2017090874-appb-100027
    其中,
    Figure PCTCN2017090874-appb-100028
    为第i辆车发生第t种交通事故的最终的交通事故发生概率值,
    Figure PCTCN2017090874-appb-100029
    为第i辆车发生第t种交通事故的初始的交通事故发生概率值,Ni为与第i辆车相邻的车辆数量或者与第i辆车距离小于第二预设值的车辆数目,d为所述道路全景图信息中的第i辆车和第j辆车之间的相对距离,v为第i辆车和第j辆车之间的相对速度,
    Figure PCTCN2017090874-appb-100030
    为第i辆车和第j辆车之间的相对距离影响因子,
    Figure PCTCN2017090874-appb-100031
    为第i辆车和第j辆车之间的相对速度影响因子,X为前车发生事故t类型导致后车发生τ事故类型或加剧后车事故严重程度的影响因子,
    Figure PCTCN2017090874-appb-100032
    为第j辆车发生第τ种交通事故的初始的交通事故发生概率值,T为事故类型集合,t为前车发生事故类型,τ为后车发生事故类型,a、β、γ、η、X分别为预设常数,均为0~1之间的任一小数;
    将所述每个第一车辆发生交通事故的最终的交通事故发生概率值确定为所述每个第一车辆发生交通事故的交通事故发生概率值;
    判断所述预测模型输出的道路发生交通事故的第一发生概率值是否大于预设第三阈值;如果判断结果为是,生成预警信息输出。
  22. 根据权利要求21所述的系统,其特征在于,所述系统还包括:
    预警设备,用于接收所述智能分析服务器输出的预警信息。
  23. 一种存储介质,其特征在于,所述存储介质用于存储可执行程序代码,所述可执行程序代码用于在运行时执行如权利要求1-7任一项所述的一种车辆发生交通事故的风险预测方及权利要求8-14任一项所述的一种道路发生交通事故的风险预测方法。
  24. 一种应用程序,其特征在于,所述应用程序用于在运行时执行如权利要求1-7任一项所述的一种车辆发生交通事故的风险预测方及权利要求8-14任一项所述的一种道路发生交通事故的风险预测方法。
  25. 一种电子设备,包括:
    处理器、存储器、通信接口和总线;
    所述处理器、所述存储器和所述通信接口通过所述总线连接并完成相互间的通信;
    所述存储器存储可执行程序代码;
    所述处理器通过读取所述存储器中存储的可执行程序代码来运行与所述可执行程序代码对应的程序,以用于执行如权利要求1-7任一项所述的一种车辆发生交通事故的风险预测方及权利要求8-14任一项所述的一种道路发生交通事故的风险预测方法。
PCT/CN2017/090874 2016-12-06 2017-06-29 发生交通事故的风险预测方法、装置及系统 WO2018103313A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201611111798.2 2016-12-06
CN201611111798.2A CN108154681B (zh) 2016-12-06 2016-12-06 发生交通事故的风险预测方法、装置及系统

Publications (1)

Publication Number Publication Date
WO2018103313A1 true WO2018103313A1 (zh) 2018-06-14

Family

ID=62468416

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2017/090874 WO2018103313A1 (zh) 2016-12-06 2017-06-29 发生交通事故的风险预测方法、装置及系统

Country Status (2)

Country Link
CN (1) CN108154681B (zh)
WO (1) WO2018103313A1 (zh)

Cited By (17)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2018233558A1 (en) * 2017-06-19 2018-12-27 Beijing Didi Infinity Technology And Development Co., Ltd. SYSTEM AND METHODS FOR TRANSPORTATION SERVICE SECURITY ASSESSMENT
CN109615852A (zh) * 2018-11-20 2019-04-12 上海海事大学 一种突发事故下路段交通流分配快速预测的方法
CN110766258A (zh) * 2018-07-25 2020-02-07 高德软件有限公司 一种道路风险的评估方法及装置
CN111179141A (zh) * 2019-12-04 2020-05-19 江苏大学 一种基于双阶段分类的事故多发路段识别方法
CN112150807A (zh) * 2020-09-18 2020-12-29 腾讯科技(深圳)有限公司 车辆预警方法和装置、存储介质及电子设备
CN112735124A (zh) * 2020-12-16 2021-04-30 北京百度网讯科技有限公司 交通数据的分析方法、装置、设备、车辆及存储介质
CN112802335A (zh) * 2020-12-31 2021-05-14 维特瑞交通科技有限公司 一种基于北斗导航系统的智能交通管理方法
CN112927497A (zh) * 2021-01-14 2021-06-08 阿里巴巴集团控股有限公司 一种浮动车识别方法、相关方法和装置
CN113066287A (zh) * 2021-03-24 2021-07-02 公安部交通管理科学研究所 一种高速公路交通事故现场风险主动防控方法及系统
CN113256997A (zh) * 2021-04-30 2021-08-13 贵州数据宝网络科技有限公司 一种交通车辆违规行为检测装置及方法
CN113744526A (zh) * 2021-08-25 2021-12-03 江苏大学 一种基于lstm和bf的高速公路风险预测方法
CN113888866A (zh) * 2021-09-30 2022-01-04 苏州百宁通智能科技有限公司 一种具有多级预警功能的道路车辆管理系统
CN114333317A (zh) * 2021-12-31 2022-04-12 杭州海康威视数字技术股份有限公司 一种交通事件的处理方法、装置、电子设备及存储介质
US20220180738A1 (en) * 2019-05-06 2022-06-09 3M Innovative Properties Company Risk assessment for temporary zones
CN115100846A (zh) * 2022-05-09 2022-09-23 山东金宇信息科技集团有限公司 一种隧道内的道路事故预测方法、设备及介质
CN115691130A (zh) * 2022-10-28 2023-02-03 田海艳 交通事故通畅时长预测系统
WO2023132800A3 (en) * 2022-01-06 2023-08-17 Xena Vision Yazilim Ve Savunma Anonim Sirketi A method for anticipation detection and prevention in a 3d environment

Families Citing this family (14)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108357496A (zh) * 2018-02-12 2018-08-03 北京小马智行科技有限公司 自动驾驶控制方法及装置
WO2019242286A1 (en) * 2018-06-19 2019-12-26 Beijing Didi Infinity Technology And Development Co., Ltd. Systems and methods for allocating service requests
CN109558969A (zh) * 2018-11-07 2019-04-02 南京邮电大学 一种基于AdaBoost-SO的VANETs车辆事故风险预测模型
CN110458244B (zh) * 2019-08-20 2021-03-30 合肥工业大学 一种应用于区域路网的交通事故严重度预测方法
CN111862628B (zh) * 2019-10-24 2022-11-08 联友智连科技有限公司 基于云服务器的车速定制系统及方法
CN110867076B (zh) * 2019-11-05 2021-02-26 武汉理工大学 一种交通事故预测方法及装置
CN111159478A (zh) * 2019-12-31 2020-05-15 上海依图网络科技有限公司 一种基于视频解析的事件预测方法、装置及其介质和系统
CN111354225B (zh) * 2020-03-03 2022-05-03 中交第一公路勘察设计研究院有限公司 一种高速公路车辆事故风险评估及预警干预方法
CN111339168B (zh) * 2020-03-06 2023-08-22 德联易控科技(北京)有限公司 数据处理方法、装置、系统、存储介质和处理器
CN114446042B (zh) * 2020-11-04 2023-11-24 腾讯科技(深圳)有限公司 预警交通事故的方法、装置、设备以及存储介质
CN113096405B (zh) * 2021-06-10 2021-09-03 天津所托瑞安汽车科技有限公司 预测模型的构建方法、车辆事故预测方法及装置
CN113536949B (zh) * 2021-06-21 2023-07-28 上汽通用五菱汽车股份有限公司 事故危险等级评估方法、装置及计算机可读存储介质
CN113749915B (zh) * 2021-10-13 2023-09-01 中国计量大学 一种场景复现的导盲方法与系统
CN117494589B (zh) * 2024-01-03 2024-04-09 北京中机车辆司法鉴定中心 基于车身颜色的车辆事故预测方法、设备和存储介质

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2010066895A (ja) * 2008-09-09 2010-03-25 Toshiba Corp 道路交通情報提供システム及び方法
CN103198711A (zh) * 2013-03-21 2013-07-10 东南大学 一种降低不同严重程度交通事故概率的车辆调控方法
CN103531042A (zh) * 2013-10-25 2014-01-22 吉林大学 基于驾驶人类型的车辆追尾预警方法
CN105389976A (zh) * 2014-08-29 2016-03-09 福特全球技术公司 用于道路风险指数产生的方法和设备
CN105719510A (zh) * 2016-04-15 2016-06-29 江苏大学 车联网环境下道路交通事故链阻断系统的效率评价方法

Family Cites Families (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP4578795B2 (ja) * 2003-03-26 2010-11-10 富士通テン株式会社 車両制御装置、車両制御方法および車両制御プログラム
KR101711025B1 (ko) * 2011-04-21 2017-02-28 한국전자통신연구원 우선 관제 대상 선정 장치와 방법 및 대상물 관제 장치
CN103198685B (zh) * 2013-03-15 2016-04-13 Tcl康钛汽车信息服务(深圳)有限公司 一种实现驾驶安全预警的方法、系统
CN103646534B (zh) * 2013-11-22 2015-12-02 江苏大学 一种道路实时交通事故风险控制方法
CN104867327B (zh) * 2014-02-21 2017-05-03 中国移动通信集团公司 一种驾驶安全监测方法及装置
CN104732075B (zh) * 2015-03-06 2017-07-07 中山大学 一种城市道路交通事故风险实时预测方法

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2010066895A (ja) * 2008-09-09 2010-03-25 Toshiba Corp 道路交通情報提供システム及び方法
CN103198711A (zh) * 2013-03-21 2013-07-10 东南大学 一种降低不同严重程度交通事故概率的车辆调控方法
CN103531042A (zh) * 2013-10-25 2014-01-22 吉林大学 基于驾驶人类型的车辆追尾预警方法
CN105389976A (zh) * 2014-08-29 2016-03-09 福特全球技术公司 用于道路风险指数产生的方法和设备
CN105719510A (zh) * 2016-04-15 2016-06-29 江苏大学 车联网环境下道路交通事故链阻断系统的效率评价方法

Cited By (28)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2018233558A1 (en) * 2017-06-19 2018-12-27 Beijing Didi Infinity Technology And Development Co., Ltd. SYSTEM AND METHODS FOR TRANSPORTATION SERVICE SECURITY ASSESSMENT
US10650618B2 (en) 2017-06-19 2020-05-12 Beijing Didi Infinity Technology And Development Co., Ltd. Systems and methods for transportation service safety assessment
US10970944B2 (en) 2017-06-19 2021-04-06 Beijing Didi Infinity Technology And Development Co., Ltd. Systems and methods for transportation service safety assessment
CN110766258A (zh) * 2018-07-25 2020-02-07 高德软件有限公司 一种道路风险的评估方法及装置
CN110766258B (zh) * 2018-07-25 2022-04-01 阿里巴巴(中国)有限公司 一种道路风险的评估方法及装置
CN109615852A (zh) * 2018-11-20 2019-04-12 上海海事大学 一种突发事故下路段交通流分配快速预测的方法
CN109615852B (zh) * 2018-11-20 2021-03-30 上海海事大学 一种突发事故下路段交通流分配快速预测的方法
US20220180738A1 (en) * 2019-05-06 2022-06-09 3M Innovative Properties Company Risk assessment for temporary zones
CN111179141A (zh) * 2019-12-04 2020-05-19 江苏大学 一种基于双阶段分类的事故多发路段识别方法
CN112150807A (zh) * 2020-09-18 2020-12-29 腾讯科技(深圳)有限公司 车辆预警方法和装置、存储介质及电子设备
CN112150807B (zh) * 2020-09-18 2024-01-09 腾讯科技(深圳)有限公司 车辆预警方法和装置、存储介质及电子设备
CN112735124A (zh) * 2020-12-16 2021-04-30 北京百度网讯科技有限公司 交通数据的分析方法、装置、设备、车辆及存储介质
US11626013B2 (en) 2020-12-16 2023-04-11 Apollo Intelligent Connectivity (Beijing) Technology Co., Ltd. Traffic data analysis method, electronic device, vehicle and storage medium
CN112802335A (zh) * 2020-12-31 2021-05-14 维特瑞交通科技有限公司 一种基于北斗导航系统的智能交通管理方法
CN112802335B (zh) * 2020-12-31 2023-04-07 维特瑞交通科技有限公司 一种基于北斗导航系统的智能交通管理方法
CN112927497A (zh) * 2021-01-14 2021-06-08 阿里巴巴集团控股有限公司 一种浮动车识别方法、相关方法和装置
CN113066287A (zh) * 2021-03-24 2021-07-02 公安部交通管理科学研究所 一种高速公路交通事故现场风险主动防控方法及系统
CN113066287B (zh) * 2021-03-24 2022-04-26 公安部交通管理科学研究所 一种高速公路交通事故现场风险主动防控方法及系统
CN113256997A (zh) * 2021-04-30 2021-08-13 贵州数据宝网络科技有限公司 一种交通车辆违规行为检测装置及方法
CN113744526A (zh) * 2021-08-25 2021-12-03 江苏大学 一种基于lstm和bf的高速公路风险预测方法
CN113744526B (zh) * 2021-08-25 2022-12-23 贵州黔通智联科技股份有限公司 一种基于lstm和bf的高速公路风险预测方法
CN113888866A (zh) * 2021-09-30 2022-01-04 苏州百宁通智能科技有限公司 一种具有多级预警功能的道路车辆管理系统
CN114333317B (zh) * 2021-12-31 2023-06-02 杭州海康威视数字技术股份有限公司 一种交通事件的处理方法、装置、电子设备及存储介质
CN114333317A (zh) * 2021-12-31 2022-04-12 杭州海康威视数字技术股份有限公司 一种交通事件的处理方法、装置、电子设备及存储介质
WO2023132800A3 (en) * 2022-01-06 2023-08-17 Xena Vision Yazilim Ve Savunma Anonim Sirketi A method for anticipation detection and prevention in a 3d environment
CN115100846A (zh) * 2022-05-09 2022-09-23 山东金宇信息科技集团有限公司 一种隧道内的道路事故预测方法、设备及介质
CN115691130A (zh) * 2022-10-28 2023-02-03 田海艳 交通事故通畅时长预测系统
CN115691130B (zh) * 2022-10-28 2023-08-25 易点无忧(北京)网络科技有限责任公司 交通事故通畅时长预测系统

Also Published As

Publication number Publication date
CN108154681A (zh) 2018-06-12
CN108154681B (zh) 2020-11-20

Similar Documents

Publication Publication Date Title
WO2018103313A1 (zh) 发生交通事故的风险预测方法、装置及系统
US11113961B2 (en) Driver behavior monitoring
US10933881B1 (en) System for adjusting autonomous vehicle driving behavior to mimic that of neighboring/surrounding vehicles
US10853882B1 (en) Method and system for analyzing liability after a vehicle crash using video taken from the scene of the crash
JP7355151B2 (ja) 情報処理装置、情報処理方法、プログラム
WO2020052436A1 (zh) 车辆超载报警方法、装置、电子设备及存储介质
US9008954B2 (en) Predicting impact of a traffic incident on a road network
CN110533909B (zh) 一种基于交通环境的驾驶行为分析方法及系统
US20170263128A1 (en) Methods for enabling safe tailgating by a vehicle and devices thereof
CN108932849B (zh) 一种记录多台机动车低速行驶违法行为的方法及装置
US11034293B2 (en) System for generating warnings for road users
Fu et al. Using microscopic video data measures for driver behavior analysis during adverse winter weather: opportunities and challenges
Park et al. Opportunities for preventing rear-end crashes: findings from the analysis of actual freeway crash data
JP6365107B2 (ja) 事故情報算出装置、及びプログラム
JP2022532941A (ja) 車両信号を処理して挙動危険性の測度を計算するための装置及び方法
US20230373384A1 (en) Systems and methods for displaying contextually-sensitive braking information
Ebner et al. Identifying and analyzing reference scenarios for the development and evaluation of active safety: application to preventive pedestrian safety
JP2022054296A (ja) 運転評価装置、運転評価システム、及び運転評価プログラム
US11414088B2 (en) Anomalous driver detection system
Wang et al. Development of requirement specifications for transit frontal collision warning system
Ebner et al. Methodology for the development and evaluation of active safety systems using reference scenarios: Application to preventive pedestrian safety
Sukegawa et al. Estimation of Drivers' Cognitive Load Through Foot Placement Analysis in a Car-Sharing Service
Ess et al. Estimating the potential of a warning system preventing road accidents at pedestrian crossings
CN115099984A (zh) 一种基于驾驶员经验评估的车险保费推荐方法及装置
JP2023012147A (ja) 運転評価装置、運転評価システム、及び運転評価プログラム

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 17878880

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 17878880

Country of ref document: EP

Kind code of ref document: A1