Research On ICESat-2 Point Cloud Data Assisted Satellite Image DEM Extraction Method

The laser height measurement satellite ICESat-2 is a satellite that integrates lidar,global positioning system and inertial navigation system.It can quickly obtain high-precision three-dimensional information on the surface of the earth.It has the characteristics of fast speed and wide detection range.Digital elevation model(Digital Elevation Model)is one of the main terrain products.The use of mapping satellite stereo image pairs to extract DEM has the advantages of low data acquisition cost and fast data collection speed.It can shorten the DEM production cycle and improve production efficiency.It is currently applied.A wide range of surveying and mapping methods.Satellite image regional network adjustment is an important technique for uncontrolled satellite imagery mapping.The accuracy of regional network adjustment directly determines the accuracy of DEM extraction,while the positioning accuracy of uncontrolled regional network adjustment mainly depends on the quality of satellite image data.Current technology Under the conditions,uncontrolled regional network adjustments are generally difficult to meet the requirements of high-precision DEM generation.At the same time,it is extremely difficult to obtain control points for surveying areas that are difficult to reach overseas and the field.Therefore,obtaining high-precision three-dimensional control point-assisted satellite stereo pairs to extract high-precision DEM is the focus of this paper.In view of the high-precision three-dimensional coordinates of the altitude measurement satellite ICESat-2 point cloud data and not limited by the region,this paper focuses on the automatic extraction of control points by combining ICESat-2 point cloud data and optical images,and using high-precision control points to assist the three images The regional network adjustment finally generates a DEM that meets the accuracy requirements for research.The main work and research results are as follows:1.In view of the characteristics of point cloud data such as large background noise and two-dimensional narrow band distribution,the point cloud denoising and boundary feature point detection algorithm based on the point cloud spatial density isintroduced,and the coding is implemented.Experiments show that the algorithm has excellent denoising effect,and the denoising accuracy is better than 98%;the boundary feature point detection is applied to the denoised point cloud data,so as to realize the extraction of ICESat-2 point cloud feature points.2.For image corner detection,there are a lot of corner problems that do not meet the conditional response.This paper improves it,using LSD(a Line Segment Dectctor)fast line detection to generate feature lines as constraints,and culling the original corner points.The results show that this method retains the geometric corner information of the image well.The nearest neighbor search method is used to register the point cloud with the image,so as to realize automatic extraction of control point results with high-precision three-dimensional coordinates.3.Introduce the principle and method of extracting the DEM of the stereo pair,and focus on the adjustment of the three satellite image regional network adjustment of the auxiliary point of the control point based on the automatic extraction of ICESat-2 point cloud data and its DEM automatic extraction method and process steps.By analyzing the accuracy of the regional network adjustment of the stereo image in the test area,the experiment shows that the elevation accuracy of the laser point cloud assisted regional network adjustment has been significantly improved,which provides support for the generation of DEM results that meet the accuracy requirements.