Research On Filtering Method Of UAV Image-dense Matching Point Cloud

In recent years,aerial photogrammetry technology can obtain high-precision scene information at low cost and over a wide range,and has become one of the important means for acquiring data in the construction of “digital city” in the new era.The UAV platform is equipped with GPS,IMU and non-measurement cameras to collect high-precision position and ground image data and attitude data,and can generate digital orthophotomaps,digital surface models,and dense point cloud data through steps such as automatic aerial triangulation and image matching.However,the dense point cloud data contains information such as buildings and vegetation on the surface,and it cannot realistically simulate the actual surface morphology.It is necessary to separate the ground points and non-ground points in the point cloud.The process of dividing ground points and non-ground points in point cloud data is called point cloud filtering.At present,all kinds of classical point cloud filtering algorithms need to set more complex parameters in the algorithm,requiring professionals to have a deep understanding of the work area in order to achieve a certain degree of filtering effect.This paper adopts cloth simulation filtering algorithm with simple setting parameters,and implements this algorithm and applies it to dense matching point cloud data.The experiment analyzed the filtering effect and provided the reference value of the algorithm on the threshold setting.It has important research significance in hydrological analysis,geomorphological analysis,flood prevention and disaster reduction,and digital mapping.The research contents of this paper are as follows:(1)The key technologies for the acquisition of UAV image matching point cloud are described from the principles of camera calibration,POS-assisted aerial triangulation,and multi-view image intensive matching,and specifically describe the method flow of acquiring UAV image data and generating point cloud by image processing.This article uses PIX4 Dmapper software to process the digital orthophotomaps and dense point cloud data in the study area to provide theoretical and data support for subsequent experimental analysis of filtering algorithms.(2)In-depth study of several classical point cloud filtering algorithms,detailed analysis of the inconvenience of the parameters set in the corresponding algorithm,describes the advantages and disadvantages of point cloud filtering.Aiming at the complexity of the threshold selection in the filtering process in the classical algorithm,cloth simulation filtering algorithm using numerical parameters and Boolean parameters is adopted.According to several easily set parameters in the algorithm,the image matching point cloud data were tested,and the filtering effect was evaluated qualitatively and quantitatively.(3)Combine the obtained digital orthophoto map and filtered ground point data to make a digital line graphic.First,thinning the data of the dense point cloud data to generate the digital elevation model and contour line data;then combining the generated contour line data with the vectorized partial features in the digital orthophoto map,make the digital line graphic after local modification;finally select checkpoints for accuracy analysis.