Building Outline Extraction Based On Airborne Lidar Point Cloud Under Tree Obscured

Buildings are an important part of the city and an indispensable part of the"Digital City".Extracting buildings from airborne Lidar point cloud is one of the key issues in urban modeling,and the outline of the buildings is the key message to express the buildings.At present,the study of building outline extraction is mostly directed at the complete building point cloud.In view of the fact that the buildings are obscured by the adjacent tall trees and so on,the building point cloud is partially missing and the building outline is not complete.At present,there is no effective outline extraction method for this situation.In order to accurately extract the outline of the obscured buildings,two main problems are studied in this research:1)Filtering,separating the ground and non-ground points in the original Lidar point cloud;2)Building outline extraction,extraction of building points from the non-ground point cloud,and extract its outline.Firstly,Skewness balancing filtering algorithm based on height difference(SKF-HD)is proposed by introducing the local height difference into Skewness balancing filtering(SKF)algorithm instead of height.This algorithm not only preserves the extraction effect of the taller objects,but also improves the extraction effect of lower objects,and significantly improved the adaptability of terrain undulating areas.The experimental results of three different terrain and different regions show that the new algorithm is not only can better extract lower objects but also applicable to varying degrees of undulating terrain.Compared with the SKF algorithm,the F-measure of the SKF-HD algorithm is increased by4.8%,5.1%and 13.3%,in the flat areas,slight undulating terrain and significant undulating terrain areas,respectively.Under different terrain conditions,especially in the areas of significant undulating terrain,the filtering accuracy of the new algorithm is improved significantly.Secondly,A Multi-level Minimum Bounding Rectangle(MMBR)algorithm is proposed based on the MBR(Minimum Bounding Rectangle)algorithm for the obscured regular polygon buildings to ensure accurate extraction of the outline of the buildings.The algorithm can not only extract the accurate outline of the building where is not obscured,but also extract the accurate outline of the obscured area.Experimental results of three groups of buildings with different obscure and different shapes are presented:the algorithm can accurately extract outline of the building's obscured area.Compared with the Hierarchical Least Squares Solution With Perpendicularity Constraints(HLSPC)algorithm,the proposed algorithm can accurately extract the outline of buildings in regular rectangular buildings,L-shaped complex polygonal buildings and complex polygonal buildings,and the Vd values were reduced by 31.8%?14.3%and 12.5%,respectively.The MMBR algorithm has higher accuracy in different obscure cases.