Research On True Orthophoto Production Methods With The Lidar Data-aided

The traditional Orthophoto Map(DOM)is a digital differential correction of the image and digital elevation model(Digital Elevation Model DEM).Since DEM does not contain information such as buildings and vegetation,the generated DOM will appear in the urban area as ghosting,tilting,and pulling flowers,which cannot achieve the desired effect.True orthophoto emerges in this context.True orthophoto is based on the digital surface model(DSM)to digitally correct the image and detect and compensate the occlusion area.True orthophoto based on DSM not only eliminates the projection error caused by the terrain fluctuation,but also moves the misplaced building to the correct position,solves the problem of unnatural splicing,and has important significance for the construction of digital three-dimensional city.LiDAR technology can acquire high-precision 3D point cloud data and provide data source for obtaining high-precision DSM.However,due to its lack of texture information,it is difficult to directly discriminate the ground object,and aerial image data has rich texture information.Therefore,this paper does the research on true orthophoto production methods with the LiDAR Data-aided to solve the problem of inaccurate production of traditional orthophoto in urban areas.This paper takes the campus of S?o Paulo State University and its surrounding area as the experimental area.Firstly,the high-precision DSM obtained by LiDAR data is used,and then the orthophoto correction and occlusion area detection and texture repair are performed on the aerial image based on DSM,and the real image of the area is finally obtained.The main research work and results of this paper are as follows:(1)LiDAR point cloud and aerial image registration: the image matching technology is used to obtain the dense point cloud data of the experimental area.Based on the optimized Iterated Closest Points(ICP)algorithm,the LiDAR point cloud and the aerial image intensive matching point are matched.The clouds are registered such that the LiDAR point cloud is consistent with the densely matched point cloud coordinate system,attenuating or eliminating coordinate offsets between the aerial image and the LiDAR data.The experimental results show that the ICP algorithm used in this paper is superior to the traditional ICP algorithm in both efficiency and precision.(2)Obtaining DSM: Firstly,the ground point and non-ground point of the experimental area are separated by cloth filtering,and the high-precision DEM is constructed by using the filtered ground point.Then,according to the characteristics of the building and vegetation on the elevation texture,the original DSM and DEM's elevation grayscale image are compared to obtain a normalized digital surface model(nDSM).Edge detection is performed based on Canny operator,and the buildings in the experimental area are acquired to construct a high-precision digital building model(Digital Building Model,DBM),finally using the filtered ground point and building point cloud to reconstruct the DSM.The experimental results show that the three types of errors in the cloth filtering algorithm are smaller than the irregular triangular mesh filtering.The cloth filtering algorithm in urban area is simple,the parameters are less,and the precision is high.The Kappa coefficient is used to verify the high extraction accuracy of the building.The Kappa coefficient is 0.8471.(3)True orthophoto production: Firstly,the orthophoto correction is performed on the aerial image based on DSM,then the occlusion detection is performed on true orthophoto by using the height-based occlusion detection method,and the texture is repaired by using the adjacent image for the detected occlusion area.After the production,true orthophoto is completed.And by calculating the error in the plane position of the true orthophoto,it proves that it meets the actual production requirements.