Homogenization Research Of Waveform Sampling Lidar Point Cloud Data

Light Detection and Ranging(Li DAR)is a new type of technology for acquiring three-dimensional geographic information with advantages of active laser detection,high precision and high efficiency in surveying and mapping.It is widely used in the field of remote sensing,reconnoiter and so on.The Streak Tube Imaging Li DAR(STIL)is a kind of Li DAR with waveform sampling ability.And the Li DAR systems applying streak tube have attracted great attention due to their special properties about detection sensitivity and distance resolution.This kind of Li DAR generates massive point cloud data with high efficiency.However,the distribution of laser foot points is found usually irregular due to the scanning mode of this kind of Li DAR.There are places where the footprints are redundant,sparse or even missing.Therefore,the accuracy in extracting terrain information is usually unsatisfactory through the original point cloud data.Homogenization research of the point cloud data obtained by STIL is necessary.The homogenization of point cloud is mainly based on interpolation process.Based on building a uniform space grid on the target region,this paper focuses on the homogenization methods to realize uniform distribution of the point cloud using interpolation techniques,including nearest neighbor,arithmetic mean and inverse distance weighted interpolation.Specifically,according to the evaluation of the results of different interpolation algorithms,we propose a new homogenization method called optimized inverse distance weighted in which the inverse distance weighted interpolation method is improved.The new algorithm combines the advantages of the nearest neighbor interpolation and has a good adaptability in the homogenization of different terrains.In terms of processing effect,these four interpolation methods are all effective for the filling of the value-free regions,but the specific results are different.When dealing with flat areas,the root-mean-square error value of nearest neighbor,arithmetic mean,inverse distance weighted and the improved inverse distance weighted interpolation is 0.26 m,0.22 m,0.21 m,0.21 m,respectively.It's obvious that the root-mean-square error of inverse distance weighted interpolation is the least with the best interpolation effect and the improved algorithm still maintains the processing advantages of the planar area.When the target area undulates,the nearest neighbor method makes the elevation values of the edge show a step change.This method is found to possess satisfactory effect for dealing with the targets with step edges.The arithmetic mean interpolation has an over-homogenization effect of the target point cloud with an unsatisfactory result..The inverse distance weighted interpolation can better show the gradient of the elevation of the whole point cloud data,and it is ideal for the continuous change of the topography,but it can't retain the details of the edges.The improved inverse distance weighted interpolation combines the advantages of the nearest neighbor interpolation algorithm.By judging the elevation feature in the neighborhood of the target point,the terrain is classified,and then the appropriate algorithm is selected for processing,which increases the adaptability of the algorithm.Moreover,the efficiency of the original inverse distance weighted method is also improved.In this paper,the improved inverse distance weighted interpolation is used to homogenize buildings,vegetation,and hills,and satisfactory interpolation results are obtained.