Research On Filtering And Classification Of Airborne LiDAR Data

Airborne LiDAR(Light Detection and Ranging)system provides a new technical method for 3D spatial data acquisition.Which has the characteristics of quick and accurate acquisition of 3D information.And Airborne LiDAR system plays an important role in many application,such as terrain mapping,environmental monitoring,forest protection and urban 3D modeling.Nowadays the hardware part of Airborne LiDAR has been well improved,however,the development of point cloud data post processing is relatively lagging behind.For the data post processing,filtering and classification are the most important part.However,due to the complexity of terrain environment and features,existing algorithms have the disadvantages of low automation,poor efficiency,and numerous parameters.And the main subject of this thesis is airborne LiDAR data post-processing.In this thesis,we mainly focus on the content of airborne LiDAR point cloud data filtering and classification.In order to effectively improve the computational efficiency and overcome the data loss problem,the organization data of the virtual grid is used to reduce the neighborhood search time and maintain the accuracy of the original data.And several typical LiDAR point cloud data filtering and classification algorithms are analyzed in detail,and the problems that need to be solved are summarized.By analyzing and summarizing the advantages and disadvantages of the typical filtering methods of LiDAR point cloud data,we observe that the computationally burden of triangulation iterative encryption algorithm is ver y large,and the threshold adaptiveness of the surface fitting algorithm is poor.Moreover,the efficiency of operation and the adaptability of threshold parameter between the point and the surface can be improved by the point positioning and local optimization of the triangulation process.And we have preprocessed the original data by using the proposed method,and this improved method have been experimented by the standard test data.In addition,the efficiency and the filtering error are qualitatively and quantitatively analyzed,and the validity and applicability of the method are proved.According to the insufficiency of the classification algorithm and the basis of filtering,the possibility of classifying ground points and feature points are analyzed by using the number of echoes and echo intensity information.A point cloud data classification method based on elevation texture,the number of echoes and echo intensity is integrated.The experiments validation of the filter data and qualitative analysis of the classification results is also performed.We can see that the method achieves certain effects in the classification process.