Research On Classification And Mapping Application Of Dense Point Cloud Of Multi-view Image Matching

With the development of computer vision and photogrammetry technology,the multi-view oblique photography with multi-view stereo(MVS)algorithm and the multi-view stereo(MVS)algorithm developed rapidly.The multi-view oblique photogrammetry can capture the ground features from multiple angles,and can obtain the side texture information of the object besides the vertical direction,so that the texture information of the object is more abundant and can comprehensively reflect the texture attribute of the object.The SFM and MVS algorithms are used to match the multi-view image to obtain the sparse point cloud,dense point cloud,and the orthophoto image is produced by using vertical image.Respectively,using Griding Mathematical Morphology and Iterative Triangulation Interpolation on the dense point cloud data classification processing,the ground point cloud after the classification generate DEM,and then generate contours which combine with the the digital map of orthographic image to make DLG.The methods and results of this paper are as follows:(1)To study the method of multi-image matching.After using the SFM algorithm to get the sparse point cloud data object.But the object for visualization of the sparse point cloud is low,cannot reflect the characteristics of the object.therefore,using PMVS algorithm to process intensive matching.Based on the result to carry the following point cloud classification and mapping application.(2)The dense point cloud data obtained by the multi-image matching is the discrete point,Reference to LIDAR Point Cloud Discrete Point Classification Processing Algorithm,focusing on the grid mathematical morphology method and iterative triangulation method to deal with dense matching point cloud data.Through the experimental study,comparingthe results processed by two algorithms,based on the qualitative and quantitative evaluation methods,the results of each algorithm are analyzed,and the results show that the iterative triangulation method is more suitable for the classification of the point cloud in the dense image matching,and the data of the ground surface and the non ground points are obtained.(3)Finally,the digital line graph(DLG)is produced using the ground point data and the obtained orthophoto data.Due to the density of dense point cloud data and the large amount of data,it is difficult to deal with it.First of all,we dilute the point cloud,then use point clouddiluted to make the digital elevation model(Digital Elevation Model,DEM),generate contour lines and use the orthophoto map as the map to make the plan of the digital line drawing,finally,add the contour line to form the complete digital line drawing.