A Comparative Study Of Ground Object Classification Methods Based On The Intensity Of The Airborne LIDAR Point Cloud Return Light

Airborne laser radar is an integrated active remote sensing technology.The airborne laser LIDAR system is composed of three parts: global positioning system,laser scanning distance measurement system and dynamic carrier attitude measurement system.The airborne LIDAR can quickly acquire spatial three-dimensional coordinate information of the target feature,and the pulsed laser beam emitted by the airborne LIDAR system has a good strong penetration,and can penetrate part of the forest to the ground.In recent years,due to its high efficiency and high precision data,the airborne LIDAR technology has been widely used in many fields such as terrain mapping and 3D modeling with its unique advantages.Among the data collected by the onboard LIDAR used,the most important is the 3D point cloud data.How to use the airborne LIDAR technology to extract spatial information and how to carry out fast and efficient processing and classification based on the airborne LIDAR point cloud data is a difficult point that researchers must overcome.The main research content of this article is as follows:1.A detailed description of the hardware structure and characteristics of the airborne LIDAR system,the working methods of LIDAR,data characteristics,and error sources is presented.The process flow of airborne LIDAR point cloud data is summarized in detail.2.According to the characteristics of airborne LIDAR point cloud data,based on the previous classification method of airborne LIDAR point cloud,a classification method based on corrected light intensity was proposed.This paper proposes a new correction model that considers the influence of the incident angle on the light intensity of the point cloud.It can effectively reduce the influence of the incident angle generated by the airborne LIDAR scanning the target object on the information of the intensity of the returned light.3.The KNN algorithm and BP neural network algorithm are used to classify the point cloud data corrected by the light intensity information.Firstly,a lot of experiments were carried out to determine the optimality of the threshold in thealgorithm,and two algorithms were used to classify the massive point cloud data to verify the feasibility of the method.4.Experiments verify the feasibility of airborne LIDAR point cloud correction model.Two kinds of algorithms are used to classify point cloud data of the same condition,and the classification effect is compared and analyzed.Experiments show that in the fine classification,because of the active learning ability of BP neural network,its classification effect is more ideal.