Road Damage Extraction And Capacity Assessment Based On High Resolution Imagery

After an earthquake occurs,roads,as the “line-line” for evacuation and the transportation of relief materials,once get damaged,it will seriously affect the progress of relief and rescue works,which tends to cause a greater casualties and economic losses.Therefore,it is crucial for post-quake rescue commander to acquire the road damage information and to determine the road accessibility situation as soon as possible.In order to obtain the damage situation,the traditional on-site investigation needs a lot of manpower and resources and cannot guarantee the coverage of the earthquake-stricken areas.However,by using satellite or UAV remote sensing,rapid investigation can be applied.It is of great significance to use high resolution imagery to imply the assessment the road damage in the earthquake areas.Taking the remote sensing images of Jiuzhaigou earthquake in 2017 and the Lushan earthquake in 2013 as experimental data,this thesis studies the method of extracting and quantitative assessment of road damage information.The identification marks of road damage images are set up,at the same time,the visual interpretation ability of road earthquake damage in different resolution RS images are summarized and analyzed,and the method of road damage information extraction assisted with vector data is forwarded.After that,a model of quantitative evaluation of road capacity is established.The research will play a certain role in earthquake relief command decision-making and improve the practicality of remote sensing for disaster information acquisition and analysis.The main contents are as follows:(1)The research of interpretation mark and identification ability of remote sensing images is carried out.The road damages are divided into four categories: buried road surface,blocked by rolling stones,collapsed subgrade and cracking road surface,meanwhile,the damage degree contains three levels: basically intact,partially damaged and completely destroyed.In the thesis,the interpretation mark library of all kinds and levels of road damage is established,and the ability of damage recognition of images is summarized.Images with the resolutions of 0.2-0.5m,1-2m and 2-4m are discussed about their difficulty for identification of damage and the lowest space resolution need for object recognition of different damage degree are proposed.It pro-vides support for the selection of remote sensing data sources and the rapid interpretation of road damage in stricken areas.(2)The method of road damage information extraction assisted by vector data is designed: by creating the road buffer to select the possible road region and then object-oriented classification is applied to carry out multi-scale segmentation,to eliminate non-road objects.In this thesis,a new process flow of damage object extraction is forwarded.Based on the concept of feature statistics,the brightness and GLCM Contrast are chosen as the main feature to be used in the rule set.The experiment is implied aiming at the three grades of road: provincial,county and township road.The optimal segmentation scale of road at all levels are studied.By using the boundary of features' boxplot,the rule is set and road damage information is extracted.The method is tested by data of Lushan earthquake,which shows a good.Comparing and analyzing the results of road earthquake damage extraction with different resolution images,the accuracy of extraction from different data sources is obtained.After acquiring the imagery data,this study can be used to quickly determine the applicable road level and provide a feasible scheme for extracting the road destruction information by computer.(3)This thesis uses UAV images of Jiuzhaigou earthquake as an example.It puts forward two assessment parameters,width and area,to evaluate the degrees of road damage.Based on these damage parameters,the quantitative road capacity assessment model is established.According to different needs and sizes of vehicles,the relationship between damage degree and road capacity is founded,and according to various road widths and vehicle types,the calculation method of theoretical threshold is obtained.Through this relationship method,the county road Z120 after Jiuzhaigou earthquake is evaluated and its capacity can be determined.The total accuracy is above 80%.The experiment shows the width parameter assessment model has a higher consistence.The proposed method solves the problem that traditional road damage assessment method cannot determine the traffic capacity and can provide more specific information for remote sensing emergency work and decision-making after an earthquake.