Point Cloud Data Filtering Processing And Feature Extraction

Light Detection And Ranging (LIDAR) as a kind of effective and rapid Geo-spatial data acquisition method, it has been the hot research of present Geo-spatial information processing. The point cloud data processing is one important aspect of the Geo-spatial information processing. Filtering and classifying the point cloud and extracting feature from it are one of most important part of point cloud data process, and they are also the base of production of subsequent digital product.The paper have researched the method of the LIDAR point cloud data filtering, analyze the LIDAR point cloud data the formation and characteristics deeply, introduce a standard format of LIDAR point cloud data storage. Discuss about the difficulty of the point cloud data filtering and the key factors of this process. The main work and innovation point:(1) In base on the analysis of method which is about LIDAR point cloud data filtering algorithm, made some improvement for the iterative least square filtering algorithm. Add some method like eliminate low-out point and subdividing point cloud data into several patches with grid,especially set a "buffer threshold" to reduce the total error of the filtering resulto Use the least square method fitting surface experiment and achieve a point cloud data filtering algorithm which consider "buffer threshold".(2) Introduce the processing methods of remote sensing image into the classification of the point cloud data process, use supervised classification methods combine both of geometry information and other attribute information dose a preliminary research on feature extraction of colored point cloud data.(3)Combining the research of point cloud filtering and feature extraction,develop a LIDAR point cloud data processing system software, achieve the basic point cloud data processing functions.