Research On Building Seismic Damage Extraction Of Object-oriented Based On UAV Images

The remote sensing image of the disaster area acquired immediately after the earthquake can provide reliable data support for emergency rescue,command decision and disaster assessment.Low-altitude Unmanned Aerial Vehicle(UAV)remote sensing system makes up the shortages of the remote sensing of satellite,like the weather,timeliness and others.At the same time,UAV,light,flexible and less affected by the weather,can fly to the disaster area to acquire the images where rescuers can’t enter,and is playing an increasingly important role in the earthquake emergency.Damage of the buildings in the event of an earthquake is an important factor to casualties and economic losses,therefore precise assessment of building damage is of great importance to reduction of disaster losses.Low-altitude UAV remote sensing system can quickly obtain the high-resolution ortho images of the disaster area.Through the high-resolution ortho-images,rescuers can be the first time to understand the disaster situation in each region,so as to command more scientifically and rationally emergency rescue,material delivery and so on.However,ortho-imaging can only obtain the top information of the buildings,and it is difficult to obtain the bottom or wall damage information of the damaged building,which limits the accuracy of the disaster loss assessment.In recent years the rapid development of UAV oblique photography technology is a good solution to this problem.Oblique photography technology can be done from multiple angles to capture the building images,not only the top information of the building,but also the wall cracks or stair collapse and other three-dimensional seismic damage characteristics.In the automatic extraction of building damage information,the traditional data source for the disaster area is the ortho-images,as only the 2D building features used to extract the building seismic damage,while the use of oblique photography technology can generate 3D characteristics of the buildings.So it can improve the extraction precision of building damage.In this thesis,in order to study the seismic damage features of building on the images by oblique photography technology and the application of oblique photography technology in seismic damage automatic extraction,the high-resolution ortho and oblique images of the study area-Qushan Town,the old city of Beichuan,as a ruins caused by Wenchuan earthquake in 2008,are acquired by the low-altitude UAV remote sensing system.The 3D model technology is used to generate the high-resolution ground 3D model,3D dense point clouds and DOM.Based on the above data,the study is divided into the following two parts:(1)Based on the damage characteristics of the study area,the fine classification system of building damage based on the oblique image feature is constructed,and the type of building damage in the area is divided into 11 categories,and the 3D seismic damage characteristics were compared with the ground truth by on-site survey.The results show that the high-resolution 3D images of the building can be more fine distinction of many seismic damage types,very similar with on-site survey,except some minor wall cracks difficult to identify The results demonstrate that oblique photography technology realizes the important transit of seismic damage assessment from the conventional ortho-images remote sensing to the on-site survey,and has the capability of remote sensing based assessments of light degree seismic damage and low seismic intensity.(2)In the building seismic damage automatic extraction,combined with the building three-dimensional characteristic information can effectively improve its extraction accuracy.In this paper,we use the UAV image reconstruction point data to extract the 3D characteristics of the building,combined with the DOM for building seismic damage automatic extraction.Firstly,the reconstructed point cloud is processed by using the height lattice calculation method.Then,we use the object-oriented method to study the DOM,DOM + nDSM,slope characteristics + nDSM.The experimental results,compared with the visual interpretation results,show that higher classification accuracies have been obtained by use both the ortho-images combined with nDSM and the slope characteristics images then the single ortho-images.It demonstrates from the research results of the thesis,the low-altitude UAV remote sensing technology provides rapidly and reliable data support for earthquake emergency rescue and disaster loss assessment.At the same time,the oblique photography technology provides an effective way to accurately assess the seismic damage of buildings,and achieved the transition of the earthquake damage interpretation of ortho-images remote sensing to the actual survey.The 3D seismic damage characteristics of the building provided by the oblique photography technique can be used as auxiliary data to improve the precision of building damage information extraction.Therefore,the low-altitude UAV remote sensing technology will have important application earthquake emergency rescue,intensity and loss assessment.