Agglomerate Fog Monitoring And Early Warning Based On Video Image And GWR Model

With the launch of China’s strategy of a strong transportation country and rapid social and economic development,traffic safety has become an increasing focus of attention.According to the "National Road Traffic Accident Statistics Annual Report White Paper",traffic accidents caused by bad weather account for about 30% of the total number of accidents.By the end of 2020,the total mileage of roads in the country reached 5,198,100 km,of which 160,000 km of highways were open to traffic.Therefore,the monitoring and warning of agglomerate fog has become an important issue that needs to be addressed.At present,there are problems such as high hardware costs,scattered monitoring data,road conditions,low monitoring accuracy and difficulties in early warning of agglomerate fog,etc.This paper takes road video images as the research object,and conducts research on the problems in agglomerate fog monitoring and early warning.This paper focuses on the research of agglomerate fog monitoring algorithms based on video images,the analysis of agglomerate fog causation based on geographically weighted regression(GWR)and the development of a agglomerate fog monitoring and warning system.To address the problem of low accuracy of video image monitoring of dense fog,this paper reverses the use of Gaussian mixture model,eliminates the moving target and retains the background,establishes the time series of background image,analyses the change pattern of background image elements when sudden dense fog occurs,selects the image contrast,grey scale value and edge features,and constructs the visibility model of background image based on time series,and verifies that the monitoring accuracy can reach 95.8%.In order to enhance the early warning effect of the model and realise the prediction of visibility,a visibility prediction model based on neural network was constructed.Based on the high-precision visibility monitoring model constructed in this paper,a video image-based agglomerate fog monitoring and warning system is designed.Based on the analysis of the spatial and temporal characteristics of highway agglomerate fog in Shandong Province,this paper analyses the influence of geographical environmental factors on agglomerate fog on the basis of a geographically weighted regression model that takes into account local spatial differentiation characteristics.Combining the geographical environment data such as elevation(ASTER-GDEM-30M),river density and Normalized Difference Vegetation Index(NDVI)in Shandong Province,a GWR-based model of highway fog occurrence was constructed.The correlation between elevation difference and agglomerate fog occurrence is greater than that of river density.In order to address the problems of high hardware cost,scattered monitoring data and road conditions,this paper develops a agglomerate fog monitoring and early warning system based on the monitoring model constructed by this paper,which does not require additional expensive equipment,only monitoring cameras placed along the highway,and adopts distributed computing methods to achieve real-time monitoring and early warning of agglomerate fog.This paper conducts research on the current problems in the monitoring of agglomerate fog,constructs a visibility monitoring model based on Gaussian model and temporal column images,which improves the accuracy of monitoring;constructs a GWR-based model of multiple occurrences of agglomerate fog,which can solve the problems of camera deployment and agglomerate fog warning;develops an efficient agglomerate fog monitoring and warning system,which effectively reduces the cost and enhances the monitoring and warning of agglomerate fog;this research can effectively solve the problem of agglomerate fog.This study can effectively solve the problem of monitoring and early warning of the agglomerate fog and improve the level of traffic safety,which is important to reduce traffic accidents,protect the safety of people and property,and maintain social stability.