Key Technology And Application Of UAV Laser Radar In Torrential Disaster Investigation

In recent years, due to special geographical environment, extreme weather disasters and human activities and many other factors, flash flood disasters has become the main hazard which cased economic losses and casualties of our country, and is also the current focus and difficulty of the study of the flood control and disaster mitigation work.Mountain flood investigation work can provide important basic information for flash flood prone area, which has great significance for disaster prevention and mitigation.Traditional method of mountain flood survey has the problems of low operating efficiency,less available data especially for special areas and difficult to acquire data in real time.Therefore, this paper presents applying the UAV laser radar technology to the torrent disaster investigation and studies the key technologies involved in the application, aimed at quickly access the result data from laser point data such as river section, outline and center height of households and so on. The main work and innovation of this paper is as follows:(1) Relying on "national torrential disaster investigation and assessment of UAV remote sensing technology application and demonstration" project of China Institute of Water Resources and Hydropower Research, in combination with Luanchuan county unmanned aerial vehicle laser radar data, studies the UAV radar data acquisition process,the laser radar data processing methods and processes, sums up a complete process from a set of original laser point cloud to the river section, outline and center height of households and other information of mountain flood prevention focus area.(2) In terms of massive point cloud data, this paper proposed a grid-based elevation iterative filtering method to quickly classified the ground and non-ground data, and through qualitative and quantitative analysis, finally validated the method feasibility.(3) For filtered ground point cloud data, this paper gives the river section automatic extraction process and the detailed extraction steps at the same time, through using TPS interpolation method to interpolate river section height, finally achieved the river section ofthe batch. In addition, compared and analyzed river section extraction results based on RTK measurement, aerial decoding and point cloud data, the results show that the river section extracted from point cloud data can well reflect the topography, and meet the accuracy requirements of mountain flood investigation.(4) In terms of the presented data of building point cloud acquired by UAV laser radar,this paper uses the vector-based fuzzy c-means clustering method to extract building point cloud, and through setting the initial parameters and the range of normal vector finally acquired good extraction results. Then based on the extraction results through network configuration and threshold settings obtained the outline and center height of households.(5) In terms of the river section, center height of households and historical flood marks, the settlement elevation was corrected along the river, combined with the population, finally acquired the settlement height and population distribution which is along the river.(6) Based on the processes and methods proposed by this paper conducted the design and development of feature extraction software used in mountain flood investigation work,carried on the project application and demonstration which takes the point cloud data of Tao Wan small watershed as an example, acquired the feature extraction result, and in addition, showed the basic data results required by the aerial photography.