Research On Generation Method Of UAV Aerial Image Seam Line Based On Dijkstra Algorithm

With the rapid development of photogrammetry technology,digital orthophotos in 4D products have many advantages such as high precision,simple recognition,readable geographic information data,easy to zoom in and out.These advantages have played a great important part in national economy and social development.The steps of producing digital orthophotos,including image preprocessing,image matching,image positioning,image mosaic and image output.Image mosaic is a process in which aerial image data of an overlapping area is taken by a drone and processed into a digital orthophoto image.As far as the current development is concerned,the automatic production of digital orthophotos by seamless stitching is still a difficult problem in the development of photogrammetry and remote sensing.For example,in the digital orthophotos produced by stitching seams,the buildings and the independent objects are misplaced.As a result,the digital orthophoto can not achieve a seamless effect.The above problems are also the bottlenecks that restrict the production of high quality digital orthophoto products.Digital orthophoto buildings and independent objects are misplaced because the seam lines pass through areas with large height differences such as buildings and independent objects,resulting in digital orthophotos generated by stitching of seam lines are misplaced.Therefore,it is of great significance to study the generation method of UAV aerial image seam line.Against the problem that the seam line existing in the process of producing digital orthophoto images of UAV aerial orthophotos traverses large areas with different heights such as buildings and independent features.This article has done some research on the following aspects:(1)This paper expounds the automatic generation technology of seam line in the process of splicing of single orthophoto,and analyzes the research status of automatic generation technology of seam line at home and abroad,and research on the seam lines existing in the automatic generation technology of aerial image seam lines through buildings and independent objects.(2)In order to ensure the natural transition of the integrity of the ground object on the digital orthophoto,solve the problem of misalignment of buildings and independent objects on the generated digital orthophotos.In this article,the morphological filtering technique,the Delaunary algorithm and the Dijkstra algorithm are used to automatically generate the orthophoto seam line.Firstly,the morphological filtering technique is used to filter out buildings and independent objects with large elevations on the DSM(Digital Surface Model),and the candidate points are extracted by using the filtered DSM.Then,using the Delaunary algorithm,the extracted candidate points are formed into a triangulation network,and the constructed triangular mesh is optimized according to the proposed triangular surface rasterization optimization and edge average iterative optimization method,and weight optimization of the edges in the triangulation.Finally,constrained search for triangulation based on Dijkstra algorithm,and the search results are tracked to generate the seam line polygon.The resulting of seam line polygon does not cross the area with a large difference in elevation such as the building and the independent object,so that the building and the independent object on the digital orthophoto generated by the splicing will not be misaligned,it will improve the quality of digital orthophoto image.(3)By comparing and analyzing the seam line generation method and the voronoi diagram(Tyson polygon)algorithm to generate the seam line,and the subjective visual evaluation and objective data evaluation were carried out.Through the analysis of subjective visual evaluation,it is proved that the quality of the digital orthophoto image produced by the method is better than that of the digital orthophoto image produced by the voronoi diagram algorithm.The objective data evaluation verifies the results of the method by calculating the average value of the pixel gray scale,the standard deviation of the pixel and the mean value,and the average gradient in the digital orthophoto image generated by the seam line.At the same time,the objective data proves that there is no misalignment in buildings and independent objects in the digital orthophoto images,and there is no obvious splicing marks,and the digital orthophoto images have clear texture and rich information.The time efficiency of generating the seam line is superior to the seam line generated by the voronoi diagram algorithm.