Research Of Building Damage Extraction Based On Three Dimensional Images From Oblique Photography

Low-altitude UAV remote sensing system with a low cost, flexible, convenient, et al, can quickly obtain high-resolution images in disaster areas, especially which are severely affected and are not accessible, and be playing an increasingly important role in the aviation photography. With the traditional two-dimensional orthophoto maps used for analysis of damage only from the vertical perspective, we cannot get the height of the building, wall structures, facades cracks and other information. However, accurate building damage assessment is critical in earthquake emergency, disaster relief, economic loss assessment. So the development of building damage analysis based on the two-dimensional remote sensing image will be restricted. Combined with traditional aerial photography and digital terrestrial acquisition technology, airborne multi-perspectives oblique photography emerges at the right moment, by carrying one or more sensors on the same flight platform, obtains both roof and facades of buildings simultaneously from multiple perspectives, model-generated high precision three-dimensional images address the limitations of traditional two-dimensional images used to analyze damage only from the vertical perspective.In this paper, we collect images of the M6.5 Ludian Earthquake occurred on August 3,2014 acquired by using the UAV oblique photography collection techniques and study the highly resolution three-dimensional model images and Digital Surface Model(DSM) generated by modeling technology. By studying the oblique photography and the building damage extraction methods which the traditional two-dimensional remote sensing image are used, we take dimensionality reduction means by exporting texture images that are corresponding to three-dimensional models. Combined with object-oriented analysis methods, we obtain concrete facades and structural cracks, then according to 《Classification of earthquake damage to buildings and special structures(GB/T 24335-2009)》, the results are analyzed to determine the degree of building damage.By studying the characteristics of DSM images data generated by oblique photography and the traditional means to extract building based on DSM images, we try to convert the DSM image into the DSM depth image and normalize the DSM image into the nDSM image, by setting thresholds to extract buildings of the same area, then evaluate the accuracy. The research of buildings extraction based on DSM images shows that the normalized DSM building extraction has advantages in undulating terrain of the area than the DSM depth image. Therefore, we extract intact buildings by setting different height threshold based on the normalized DSM image, then analysis the collapse rate of the region based on the collapse rate formula.Our country suffers seriously the earthquake disasters. Building damage assessments take influence on delineation of correction of intensity, vulnerability of buildings and casualties after earthquakes. Unmanned Aerial Vehicle(UAV) oblique photography system can capture images from multiple perspectives, and generate three-dimensional images and DSM with high resolution by modeling techniques. Extracting building damage and the collapse rate after the earthquake based on three-dimensional explore the research of oblique photography to some extent, while extracting building damage is of great significance in seismic intensity determination, on-site emergency and the assessment of seismic damage, casualties, economic loss.