ERS-TD: An Efficient Reconstruction and Sampling Method for TDOM Based on 3D Point Cloud Geometry in Image Cloud Computing

Abstract

Digital Orthophoto Map (DOM), as a crucial surveying and mapping product, often encounters issues such as occlusion, ghosting, and edge blur in practical applications, which undoubtedly have a negative impact on the accuracy of image analysis. The introduction of True Digital Orthophoto Map (TDOM) aims to effectively address these challenges, with its core objective being to eliminate interferences caused by terrain and ground target projection distortions. However, traditional TDOM production methods rely on digital surface models for orthorectification, a process that often results in the occlusion of high-rise buildings in the imagery due to the displacement of image points. In view of this, this paper proposes an Efficient Reconstruction and Sampling Method for TDOM Based on 3D Point Cloud Geometry in Image Cloud Computing (ERS-TD). This method innovatively integrates key technical components such as point cloud registration, absolute orientation, equal interval sampling, vertical projection, and texture mapping. Among these, the design of the equal interval sampling method is particularly ingenious, as it aims to efficiently and accurately extract surface point cloud data from the model. By flexibly adjusting the grid size of the sampling interval and comprehensively traversing the entire matrix, this method can rapidly and precisely obtain the surface point cloud information of the entire model, significantly improving efficiency compared to traditional methods. Furthermore, the vertical projection method projects the surface point cloud onto a plane, effectively eliminating the influence of elevation values and clearly revealing the outlines of buildings. In addition, to meet the demand for 3D map construction, this study further explores how to utilize the acquired 3D point cloud data and texture information to construct more refined and realistic 3D maps on the basis of generating True Digital Orthophoto Maps. By optimizing the processing flow of point cloud data and enhancing the accuracy of texture mapping, this study successfully achieves the transition from 2D imagery to 3D maps, providing new ideas and methods for 3D map construction.