Research On 3D Reconstruction Technology Based On UAV Aerial Photography

Construction of "smart city" is the development direction of central cities.It is an important part using three-dimensional reconstruction technology to rebuilt urban architecture,land and roads into a three-dimensional model,which involves two aspects: information collection and 3D reconstruction.Using UAV platform to collect image for Urban and geographical,instead of satellite remote sensing,large unmanned aerial vehicles and other traditional methods,Not only in cost and efficiency have a great advantages,the datas have a higher precision and greater information.The 3D reconstruction algorithm based on patch has a high degree of automation and the three dimensional model is rich in detail,but the reconstruction time is long.In this paper,the large scene is decomposed into multiple regions by cluster classification principle.Each region is parallel reconstructed to reduce the overall reconstruction time.We propose a method of planar region detection at low resolution,and then reduce the number of point clouds in the complex area.Increasing the number of point clouds and decreasing the overall reconstruction time.This paper mainly research the three-dimensional reconstruction of large data sets,The main work and achievements are as follows:1.Based on the reconstruction effect and the requirement of automation,this paper chooses three-dimensional reconstruction algorithm based on Stereo Vision from a variety of algorithms.2.According to the characteristics of the image data,the range of the point cloud is added into the algorithm to improve the accuracy of the point cloud and reduce the additional time consumption.3.In view of large amounts of image sets of three-dimensional reconstruction,this paper introduces the image clustering algorithm to decompose the large scenes into several small areas to reduce the three-dimensional reconstruction time.4.According to the similarity principle of the scene in the PMVS algorithm,Scene plane area is extracted in low resolution image layer,and in this plane the number of dense cloud points is improved by the improved patch diffusion method.The improved algorithm is analyzed by the actual collected data.The results show that the improved algorithm has a great improvement in point cloud density and in reconstruction speed.