Research And Implementation On Calibration Method Of Unmanned Helicopter LiDAR System

Airborne laser scanning mapping(LiDAR) system is the most advanced mapping system to get the three-dimensional information of the earth’s surface currently, and it has been widely used in topographic mapping, environmental emergency monitoring, island coastal protection, etc. The LiDAR system which is based on small scale unmanned helicopter can fulfill the task of low altitude, fast, fine scanning, so it can obtain more detailed feature of the earth’s surface and a higher level of data accuracy is required. However, the errors of LiDAR system have a wide range of source and the errors are very difficult to be calibrated out directly, so it is necessary to establish effective error calibration models for systematic calibration to ensure the accuracy of point cloud before real scanning. In this dissertation, this problem is discussed and studied, and the related scanning experiments are carried out.This dissertation first introduces the basic principle and components of airborne LiDAR system. After that, the point cloud generation equation is derived from raw data. Then, based on the analysis of internal systematic error sources and their impact, two different calibration methods are designed, which are as follows: 1) Model calibration method based on strip adjustment. At first, from the difference of corresponding points between parallel overlapping flight strip, it establishs a difference analysis model under the restricted restrain of scanning data; Then the point cloud registration of two parallel overlapping strip is implemented by optimizing the traditional iterative closest point(ICP) algorithm, thus the adjustment parameters between two flight strips can be collected; At last, six original system parameter errors can be obtained by applying least squares method to the combination of the adjustment parameters and variance analysis model. 2) Model calibration method based on the feature points. At first, by centering all the systematic errors on three placement angle errors, it establishes a placement angle errors calibration model based on the feature points; Then in order to get corresponding points, the feature points are extracted and matched by interpolating in regular grid; At last, the placement angle errors are calculated by least squares method.This dissertation mainly focuses on the impacts of system errors on positional deviation of the overlap strip corresponding points, and establishes two error calibration models. The test under real coordinate of feature points is not necessary for this two types of methods and thus avoids many restrictions in traditional manual calibration methods which require real coordinates of feature points or standard calibration field. Finally, the effectiveness of the two kinds of calibration methods is verified through simulation flight experiment and field flight experiment respectively. The laser point cloud data can be obtained with higher precision, which establishes the foundation for subsequent data processing.