基于多元辅助信息的机载LIDAR点云数据滤波分类研究

Filtering and Classifying Airborne LIDAR Cloud Data Based on Multiple Subsidiary Information

借助多元辅助信息对黑龙江省凉水国家级自然保护区内las标准格式的原始激光雷达点云数据进行处理。首先根据研究区内地形起伏分布和地物的不同裁剪分割点云数据,去除掉极高点与极低点;然后采用渐进式不规则三角网算法和带权的线性迭代预测算法,通过设置不同的参数组分别对具有代表性特征的分块数据进行滤波试验,对比分析了两种算法在不同地形区域的滤波效果,进而得到各自的适用范围。再结合这两种算法对所有分块进行滤波处理,并充分利用las格式点云携带的强度与回波及同步获取的影像等多元信息,帮助分类。最后评价多元信息滤波分类的效果。结果表明:结合两种算法和多元辅助信息对凉水地区进行滤波分类能很好地保留地形与地物的局部细节信息,从而为准确构建研究区域的数字高程模型和数字表面模型提供了技术依据,同时验证了机载LIDAR技术在东北林区应用推广的可行性与优势性。

Assisted by multiple subsidiary information the standard las format LIDAR data in Liangshui nature reserve,Heilongjiang are used in cloud data processing.Firstly,subdivide the cloud data according to different distribution of the topography and feature.Secondly,remove the noisy points such as the very high and low points.Thirdly,the TIN filtering algorithm and the weighted linear iterative filtering algorithm are used to filter the data,and set the different parameter groups respectively to process the representative block of data.Fourthly,compare and analyze the results of two methods,then got their suitable zones respectively.Fifthly,combine two algorithms to process all LIDAR data in Liangshui nature reserve.What＇s more,making full use of information such as the multiple echoes、intensity and these images which obtained synchronously with LIDAR cloud data to help the classification.At last,make a quality assessment of this method.Results show that combining two filtering algorithms and multiple information can well kept local details of the original feature and targets,which provided a technical basis for constructing the accurate digital elevation model and digital surface model;besides,this research also verified the feasibility and superiority of LIDAR technology in Northeast China region.