CN-121999443-A - Expressway pedestrian threat assessment method and system based on multidimensional risk fusion

CN121999443ACN 121999443 ACN121999443 ACN 121999443ACN-121999443-A

Abstract

The invention relates to the technical field of intelligent traffic safety, in particular to a highway pedestrian threat assessment method and system based on multidimensional risk fusion. The method comprises the steps of obtaining expressway pedestrian monitoring image data of a target area, conducting pedestrian target detection processing to generate pedestrian target detection data, conducting position risk quantification calculation and behavior risk quantification calculation on the basis of the pedestrian target detection data to generate position risk assessment data and behavior risk assessment data, obtaining environment perception data, conducting environment risk quantification calculation to generate environment risk assessment data, conducting weighting fusion calculation on the position risk assessment data, the behavior risk assessment data and the environment risk assessment data after normalization processing to generate comprehensive threat level assessment data, conducting grading comparison on the comprehensive threat level assessment data and a preset threat threshold value, and generating grading response strategy execution data matched with threat level. According to the invention, by establishing the three-dimensional independent quantitative risk model of the position risk, the behavior risk and the environmental risk and carrying out normalized weighted fusion evaluation, the accurate quantitative grading and differential grading response of the expressway pedestrian risk degree are realized, and the refinement level and the emergency response efficiency of the safety management decision are improved.

Inventors

ZHENG YUBIAO
ZHENG YOUHUA
Zheng Boqu

Assignees

国科星图(深圳)数字技术产业研发中心有限公司

Dates

Publication Date: 20260508
Application Date: 20260409

Claims (10)

1. The expressway pedestrian threat assessment method based on multidimensional risk fusion is characterized by comprising the following steps of: step S1, acquiring expressway pedestrian monitoring image data of a target area, and performing pedestrian target detection processing on the expressway pedestrian monitoring image data to generate pedestrian target detection data; S2, acquiring a spatial distance parameter of a pedestrian relative to the center line of a highway lane based on the pedestrian target detection data, and performing position risk quantification calculation on the spatial distance parameter to generate position risk assessment data; Step S3, acquiring pedestrian movement behavior parameter data based on the pedestrian target detection data, wherein the pedestrian movement behavior parameter data comprises pedestrian movement speed data, pedestrian movement direction angle data and pedestrian acceleration data; S4, obtaining environment perception data of a current expressway area, wherein the environment perception data comprises traffic flow density data, weather condition data and illumination intensity data; Step S5, respectively carrying out normalization processing on the position risk assessment data, the behavior risk assessment data and the environment risk assessment data to obtain normalized position risk data, normalized behavior risk data and normalized environment risk data; And S6, carrying out grading comparison on the comprehensive threat level evaluation data and a preset threat threshold value, and generating grading response strategy execution data matched with the threat level.
2. The expressway pedestrian threat assessment method based on multidimensional risk fusion according to claim 1, wherein step S1 comprises the steps of: Step S11, acquiring original expressway pedestrian monitoring image data of a target area through image acquisition equipment, and performing image preprocessing operation on the original expressway pedestrian monitoring image data, wherein the image preprocessing operation comprises image denoising processing, image enhancement processing and image registration processing, so as to generate preprocessed expressway pedestrian monitoring image data; And step S12, carrying out pedestrian target recognition and positioning on the preprocessed expressway pedestrian monitoring image data by utilizing a pre-trained pedestrian target detection model to generate pedestrian target detection data, wherein the pedestrian target detection data comprises pedestrian position information data and pedestrian detection confidence data in an image coordinate system.
3. The expressway pedestrian threat assessment method based on multidimensional risk fusion according to claim 1, wherein step S2 comprises the steps of: s21, acquiring position information data of pedestrians in an image coordinate system based on the pedestrian target detection data, and converting the position information data in the image coordinate system into pedestrian geographic coordinate data in a geographic coordinate system by utilizing a coordinate conversion algorithm; S22, obtaining geographic coordinate reference data of a lane center line of the expressway, and calculating a spatial distance parameter of a pedestrian and the lane center line according to the geographic coordinate reference data of the pedestrian and the geographic coordinate reference data of the lane center line; And S23, carrying out position risk quantification calculation according to a preset safety distance threshold and the space distance parameter to generate position risk assessment data, wherein the calculation method of the position risk assessment data comprises the steps that a position risk value is equal to a larger value between zero and minus the pedestrian distance divided by the safety distance threshold.
4. The expressway pedestrian threat assessment method based on multidimensional risk fusion according to claim 1, wherein step S3 comprises the steps of: Step S31, based on the continuous multiframe pedestrian target detection data, obtaining displacement data of pedestrians between adjacent frames, and calculating pedestrian moving speed data according to the displacement data and inter-frame time intervals; S32, acquiring expressway lane direction reference data, and calculating included angle data of a pedestrian moving direction and a lane direction according to the displacement direction between adjacent frames of the pedestrian and the lane direction reference data; step S33, calculating pedestrian acceleration data according to the moving speed variation quantity of the pedestrian in the continuous multi-frames; And step S34, respectively multiplying the pedestrian moving speed data, the included angle data between the pedestrian moving direction and the lane direction and the pedestrian acceleration data by corresponding weight coefficients, and then carrying out weighted summation calculation to generate behavior risk assessment data, wherein the speed weight coefficients represent the influence of the pedestrian moving speed on the available reaction time, the direction weight coefficients represent the moving components of the pedestrian crossing the lane direction through sine function values of the direction angles, and the acceleration weight coefficients represent the influence of the pedestrian moving trend change on the risk.
5. The expressway pedestrian threat assessment method based on multidimensional risk fusion according to claim 1, wherein step S4 comprises the steps of: S41, acquiring traffic flow density data of a current expressway area, and performing traffic risk quantitative calculation according to the traffic flow density data to generate traffic risk factor data; Step S42, weather condition data of a current expressway area are obtained, weather risk quantification calculation is carried out according to the weather condition data, and weather risk factor data are generated; Step S43, acquiring illumination intensity data of a current expressway area, and performing visual risk quantification calculation according to the illumination intensity data to generate visual risk factor data; and S44, carrying out weighted combination calculation on the traffic risk factor data, the weather risk factor data and the visual risk factor data to generate environmental risk assessment data.
6. The expressway pedestrian threat assessment method based on multidimensional risk fusion according to claim 1, wherein in step S5, the normalization process is to map each dimension risk assessment data to a value interval from zero to one by using a minimum-maximum normalization method, and in the weighted fusion calculation, a sum of a position risk weight, a behavior risk weight and an environmental risk weight is equal to one, and the position risk weight is greater than the behavior risk weight, and the behavior risk weight is greater than the environmental risk weight.
7. The expressway pedestrian threat assessment method based on multidimensional risk fusion according to claim 1, wherein step S6 comprises the steps of: Step S61, comparing the comprehensive threat level evaluation data with a first preset threshold value and a second preset threshold value, wherein the first preset threshold value is larger than the second preset threshold value; Step S62, when the comprehensive threat level evaluation data is greater than or equal to the first preset threshold value, judging that the comprehensive threat level evaluation data is high in risk level, and generating emergency response strategy execution data, wherein the emergency response strategy execution data comprises height control data for controlling an execution device to descend to a first preset height interval, acoustic alarm control data for starting a first preset decibel level, high-brightness strobe warning control data and emergency evacuation voice broadcasting control data; Step S63, when the comprehensive threat level evaluation data is greater than or equal to the second preset threshold value and smaller than the first preset threshold value, judging the comprehensive threat level evaluation data as a medium threat level, and generating conventional response strategy execution data, wherein the conventional response strategy execution data comprises height control data for controlling an execution device to keep a second preset height interval, acoustic alarm control data for starting a second preset decibel level, active medium brightness alarm control data and start safety reminding voice broadcasting control data; And S64, when the comprehensive threat level evaluation data is smaller than the second preset threshold value, judging that the comprehensive threat level evaluation data is in a low risk level, and generating early warning reminding strategy execution data, wherein the early warning reminding strategy execution data comprises height control data for controlling an executing device to keep a third preset height interval, prompting sound control data for starting a third preset decibel level and general safety reminding voice broadcasting control data.
8. The method for evaluating the pedestrian threat on the expressway based on the multidimensional risk fusion according to claim 1, further comprising a weight adaptive adjustment step of: The position risk weight, the behavior risk weight and the environment risk weight are dynamically adjusted according to weather condition data and illumination intensity data in the environment perception data, the environment risk weight is increased when the illumination intensity data is lower than a preset illumination threshold value or the weather condition data indicates bad weather, and the position risk weight is increased when the traffic flow density data is higher than a preset flow threshold value.
9. The expressway pedestrian threat assessment method based on multidimensional risk fusion of claim 1, further comprising a response effect assessment step of: The method comprises the steps of generating and executing the grading response strategy execution data, continuously obtaining pedestrian follow-up pedestrian target detection data of pedestrians, calculating variation data of pedestrian positions and behaviors based on the pedestrian follow-up target detection data, evaluating the execution effect of the response strategy according to the variation data, automatically improving the intensity level of the response strategy when the pedestrians are not far away from a dangerous area, and recording the comprehensive threat level evaluation data, the grading response strategy execution data and the execution effect data for optimizing threat evaluation model parameters.
10. A system for assessing highway pedestrian threat based on multi-dimensional risk fusion, for performing the method for assessing highway pedestrian threat based on multi-dimensional risk fusion according to claim 1, the system for assessing highway pedestrian threat based on multi-dimensional risk fusion comprising: the system comprises a pedestrian target detection module, a pedestrian detection module and a pedestrian detection module, wherein the pedestrian target detection module is used for acquiring expressway pedestrian monitoring image data of a target area; The position risk assessment module is used for acquiring the spatial distance parameter of the pedestrian relative to the center line of the expressway lane based on the pedestrian target detection data, carrying out position risk quantification calculation on the spatial distance parameter, and generating position risk assessment data; the behavior risk assessment module is used for acquiring pedestrian movement behavior parameter data based on the pedestrian target detection data, wherein the pedestrian movement behavior parameter data comprises pedestrian movement speed data, pedestrian movement direction angle data and pedestrian acceleration data; the system comprises an environment risk assessment module, a data processing module and a data processing module, wherein the environment risk assessment module is used for acquiring environment perception data of a current expressway area, and the environment perception data comprises traffic flow density data, weather condition data and illumination intensity data; The comprehensive threat assessment module is used for respectively carrying out normalization processing on the position risk assessment data, the behavior risk assessment data and the environment risk assessment data to obtain normalized position risk data, normalized behavior risk data and normalized environment risk data; And the hierarchical response strategy generation module is used for performing hierarchical comparison according to the comprehensive threat level evaluation data and a preset threat threshold value to generate hierarchical response strategy execution data matched with the threat level.

Description

Expressway pedestrian threat assessment method and system based on multidimensional risk fusion Technical Field The invention relates to the technical field of intelligent traffic safety, in particular to a highway pedestrian threat assessment method and system based on multidimensional risk fusion. Background The designed running speed of the expressway is usually between 80km/h and 120km/h, and once pedestrians enter the expressway by mistake or illegally, fatal accidents are extremely easy to cause. For a long time, the safety management of expressway pedestrians mainly relies on regular inspection by traffic police and visual observation of fixed sentry posts, and mechanical traffic signs and light reflecting facilities are assisted. Such means have limited ability to discover sudden events such as false entry of pedestrians, and are often perceived after an accident has occurred. Later closed-circuit television monitoring technology is gradually popularized, and a management department starts to deploy fixed cameras on a key road section, but the problems of low manual review efficiency, limited camera coverage range, monitoring blind areas and response lag are not completely solved. In recent years, target detection technologies such as convolutional neural networks and the like can automatically identify pedestrians and vehicles from video streams, millimeter wave radars and laser radars can detect moving targets all the time, and multi-source data fusion technologies are also continuously improving detection accuracy. Meanwhile, the scheme that the unmanned aerial vehicle is provided with the multispectral camera for highway inspection enters practical application in part of road sections, and viewing angle limitation of the fixed cameras is made up. However, in the prior art, when safety management is performed on expressway pedestrians, binary judgment logic is generally adopted, namely, whether pedestrians exist in a target area or not is judged, quantitative evaluation means for the risk degree of the pedestrians are lacking, meanwhile, spatial position states, movement behavior characteristics and current environmental conditions of the pedestrians cannot be integrated into a unified analysis framework, and further, the safety management decision lacks quantitative basis, and a differentiated grading response strategy cannot be generated according to the actual risk degree of the pedestrians. Based on this, it is necessary to provide a method and a system for evaluating pedestrian threat on the expressway based on multidimensional risk fusion, so as to solve at least one of the above technical problems. To achieve the above object, a highway pedestrian threat assessment method based on multidimensional risk fusion, the method comprising the steps of: step S1, acquiring expressway pedestrian monitoring image data of a target area, and performing pedestrian target detection processing on the expressway pedestrian monitoring image data to generate pedestrian target detection data; S2, acquiring a spatial distance parameter of a pedestrian relative to the center line of a highway lane based on the pedestrian target detection data, and performing position risk quantification calculation on the spatial distance parameter to generate position risk assessment data; Step S3, acquiring pedestrian movement behavior parameter data based on the pedestrian target detection data, wherein the pedestrian movement behavior parameter data comprises pedestrian movement speed data, pedestrian movement direction angle data and pedestrian acceleration data; S4, obtaining environment perception data of a current expressway area, wherein the environment perception data comprises traffic flow density data, weather condition data and illumination intensity data; Step S5, respectively carrying out normalization processing on the position risk assessment data, the behavior risk assessment data and the environment risk assessment data to obtain normalized position risk data, normalized behavior risk data and normalized environment risk data; And S6, carrying out grading comparison on the comprehensive threat level evaluation data and a preset threat threshold value, and generating grading response strategy execution data matched with the threat level. According to the invention, the pedestrian detection data is generated by acquiring the expressway pedestrian monitoring image data of the target area and carrying out pedestrian target detection processing on the expressway pedestrian monitoring image data, so that the automatic identification and positioning of the pedestrian target in the expressway area are realized, and basic data support is provided for subsequent multidimensional risk assessment. Based on the pedestrian target detection data, the spatial distance parameter of the pedestrian relative to the center line of the expressway lane is obtained, the spatial distance parameter is subjected to position risk quanti