Study On Building Damage Extraction Factors Using Post-Earthquake Airborne LiDAR Point Cloud Data

Timely post-earthquake information acquisition has practical significance to emergency rescue and disaster assessment, as earthquake is one of the worst natural disasters for human beings. Remote sensing plays a more and more important role and has been full applied in many earthquake events because of its advantages of quickly, large area, and full-day observation, and so on.Earthquake-induced building collapse is one of the main causes of casualties and economic losses. Airborne Li DAR system has advantages of little affected by weather, strong penetrability, work fast, high resolution, rich data products. The high resolution 3D point cloud acquired by airborne Li DAR system can provide important data source for building damage research.According to the different Li DAR data source, earthquake damage extraction method can be divided into three aspects: using post-earthquake Li DAR point cloud data, multitemporal Li DAR data and Li DAR point cloud data fusion with other data. The concept of building damage extraction factors were introduced in this paper. Under the analysis of different damage characteristic, this paper established the secondary factors of 3D building point cloud and analyzed the quantitative of these factors. Using the Haiti earthquake Li DAR data to determine the effective building damage extraction factors. The main contents include:1. According to the characteristic of post-earthquake Li DAR data, the paper initially establish the post-earthquake airborne LIDAR data preprocessing scheme. After de-noising and filtering, the paper use the planar density parameter to remove ground points and retain the non-ground points.2. Based on analyze the different damage characteristic of non-damage and damage building, we calculate the value of Intensity, Height, Normal information, Volume and Slope of non-collapse and collapse building and analysis the change of these value, then establish the earthquake-induced building damage extraction factors: variation of standard deviation of Intensity(SSDI), variation of standard deviation of Height(SSDH), variation of standard deviation of Angle(SSDA) between zenith and normal vector of building roof points, variation of standard deviation of Volume(SSDV), variation of standard deviation of Slope(SSDS). Then analyze the quantitative of these factors and determine the threshold of every factor.3. Based on the building damage extraction factors(SSDI, SSDH, SSDA, SSDV, and SSDS), single factor detection and Support Vector Machine detection method are used to detect building damage. The result shows SSDH、SSDA,SSDV and SSDS can be the effective building damage extraction factors, the effect of the combination of different factors is better than single factor, SVM method can realize rapid building damage detection and the overall accuracy has reached 90%...