Construction Of Regional Visibility Inversion Model Based On Monitoring Images

Atmospheric visibility is an important surface meteorological observation project,which is very important for aviation,navigation and road traffic.It can provide services for air pollution,traffic safety and air quality.At present,the instruments used to measure atmospheric visibility have many problems,such as high price,complex operation,and often need on-site maintenance,etc.However,the visibility detection method based on image processing provides the possibility to measure atmospheric visibility timely,accurately and cheaply.With the layout of a large number of monitoring facilities,the acquisition of monitoring images is easier and the cost is lower.Moreover,compared with other images,monitoring images have the characteristics of wide coverage and all-weather real-time acquisition.Therefore,regional atmospheric visibility detection based on monitoring images has a wide application space.However,the current atmospheric visibility detection methods based on monitoring images mostly need to calibrate the camera or place the target object in advance,which is not practical.Therefore,the regional visibility inversion model based on monitoring images is constructed and optimized in this paper to provide technical support for the detection of atmospheric visibility,which has important scientific value and application significance.This paper studies the atmospheric visibility detection method and inversion model based on the monitoring image.By exploring the relationship between the corresponding features in the monitoring image and the atmospheric visibility,the atmospheric visibility level and the atmospheric visibility value are calculated,and the spatio-temporal interpolation of regional visibility is realized by the combination of temporal interpolation and spatial interpolation.The main research contents are as follows:(1)Based on image processing and analysis of atmospheric visibility classification model construction.In this paper,a visibility estimation model is established by constructing the corresponding relationship between the global average gradient and dark channel ratio of the image and the visibility level.According to the threshold of atmospheric visibility level in meteorology,the monitoring image is divided into different atmospheric visibility levels.In order to improve the accuracy of the algorithm,histogram test module and blue sky operator test module are added to modify the visibility level.The experimental results show that the model can effectively identify the atmospheric visibility level.(2)An inversion model for estimating atmospheric visibility based on the transmittance of monitoring image.Aiming at the problems of rough data and blocky effect in the method of estimating atmospheric visibility based on the dark channel prior theory,this paper studies how to refine the transmittance of monitoring image and reduce the transmittance error of image.By introducing guided filtering algorithm to eliminate the blocky effect in the transmittance image,and improving its parameters to reduce the error of the image transmittance estimation,the fine atmospheric transmittance of the image can be calculated,and then the atmospheric visibility value can be estimated by combining with the distance information in the image.Through the experimental verification,the conclusion shows that: the atmospheric visibility of the monitoring image calculated by the atmospheric visibility inversion model in this paper meets the atmospheric visibility detection standard of WMO with the relative error less than 20%,which has good practical application value.(3)Spatial and temporal interpolation model of regional atmospheric visibility.Based on the idea of time-space interpolation,this paper studies the regional interpolation method of atmospheric visibility based on the above inversion model.From atmospheric visibility of the two attributes of time and space,according to the actual atmospheric visibility observations,estimate time interpolation and spatial interpolation estimate,calculate time and space of the time and space interpolation of weight coefficient,and with the interpolation time and with the location of the interpolation of atmospheric visibility data,realize regional atmospheric visibility interpolation of time and space.Experimental results show that the improved spatiotemporal interpolation method is more accurate than other spatiotemporal interpolation methods.