The Method Of Airborne Laser Scanning Data Filtering

Airborne LIDAR technology provides users with a convenient, reliable and economical data collection method. It has been used in many industries, while the research on data postprocessing algorithms is still underway. As the most fundamental aspect of the LIDAR data postprocessing, filtering results directly affect the quality of generated DEM. How to extract topographic data from-LIDAR data quickly and accurately is a problem which we are faced with and in dire need of solving now. The main contents of this thesis include:1. LIDAR system hardware components, laser surveying principle, characteristics and advantages are introduced. And the main application fields of airborne LIDAR technology is presented. Three dimensional coordinates formulas of laser points are derived from the triangular relationship of solid geometry, while further the long-term outlook for development of the technology and future trends is concluded.2. The characteristics and organization styles of LIDAR data are analyzed. LIDAR data not only contains three dimensional coordinates of points, but also includes intensity, echo, scan angle, color values and other informations. The common data exchange standard format of LIDAR data—LAS format is recommended.3. LIDAR data filtering concept and principle is discussed and several domestic and overseas filteing algorithms are studies. Further, the advantages and disadvantages are also covered, the methods which be used for generating raster grid from discrete laser points and interpolation experiments are sprobed.4. Progressive densification of adaptive TIN filtering algorithm is studied in depth, aiming at its problems, an improved method is proposed. First removing error points before filtering to reduce the impact on follow-up processing, and establishing self-adaptive threshold functions under the triangle slope is proposed. In addition, the moving surface fitting filtering algorithm is improved, whether the seed points are in one line is determined and the nearest point to the fitting surface is added each time. The experiments demonstrate the feasibility of improved algorithms, and the processing results are analysed both qualitatively and quantitatively. The main factors which affect the quality of filtered data are summarized, and it is discussed that whether the data resolution has an effect on filtering results.The experimental results show that improved algorithms can deal with LIDAR data effectively and remove error points, perigees and derive digital elevation model. The Type I errors, Total errors is relatively less and the filtering results' accuracy is improved.