Research On Building Rapid Extraction Method Based On LiDAR Data

The airborne laser scanning and ranging system is an advanced remote sensing sy stem for obtaining three-dimensional spatial information of terrain surface and other im age information in real time.It is one of the important data sources in the field of re mote sensing surveying and mapping.Accurate extraction of LiDAR data building is a prerequisite for the application of Li DAR data in dynamic monitoring of urban change s,post-disaster investigation of buildings,target determination of fighter bombing tasks and it is an important part of LiDAR data processing.With the gradual deepening of ai rborne laser scanning building classification research,new precise extraction methods of buildi ngs have been proposed,and their accuracy is even as high as 90% or more.In practical a pplications,due to the large time consumption of point cloud filtering iteration process and building edge discrimination rules,the building extraction speed cannot meet the rapid extraction requirements of applications such as damage assessment,real-time mo nitoring of dynamic changes,and rapid positioning of illegal construction,which severel y restricts the LIDAR technology development and application.This article aims to stu dy how to effectively improve the building extraction speed while ensuring the accura cy of building extraction.The contents are as follows:(1)Research on the characteristics of LiDAR data and surface features:on the basis of i n-depth analysis of the physical meaning of various types of Li DAR data features,a study w as conducted on its distinguish-ability in all kinds of data considering the features of buildings :From the aspects of data acquisition,point cloud filtering and extraction me thods,factors affecting the speed and accuracy of building extraction are analyzed,an d corresponding solutions are proposed.(2)Discussion on several methods of LiDAR date points cloud filtering,including mathematical morphology filtering,linear prediction filtering,progressive encryption fi ltering and cluster segmentation filtering.The key factors affecting accuracy and speed wh en applying skewness balance to point cloud filtering are analyzed: the increase of calculati ons is caused by point-by-point calculations during the iterative process and the decrease of a ccuracy is mainly due to the misclassification at data voids.According to the characterist ics of Li DAR data and multi-spectral data,the method of multi-spectral-guided bias p oint cloud filtering was designed.The validity of the method was verified by case ana lysis and comparison.(3)Research on the method of building extraction based on Hough lines detection : c ombining the characteristics of differences between vegetation and buildings,analyze th e factors that affect the extraction time of buildings,construct change detection operat ors and initially extract building areas,filter out the building extraction interference fa ctors of Li DAR data.Analyze the characteristics of the building's edges to construct t he rule for recognizing the straight line segments of the building's edge,then obtain t he results of building edge extraction through decision rules and verify the validity of the method.