ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences
Download
Publications Copernicus
Download
Citation
Articles | Volume IV-4
https://doi.org/10.5194/isprs-annals-IV-4-119-2018
https://doi.org/10.5194/isprs-annals-IV-4-119-2018
19 Sep 2018
 | 19 Sep 2018

EFFICIENT AND ACCURATE FUSION OF MASSIVE VECTOR DATA ON 3D TERRAIN

Z. Liu, C. Li, Z. Zhao, D. Zhang, F. Wang, and Y. Wang

Keywords: vector data, 3D terrain, seamless fusion, multi thread, level of detail

Abstract. This paper presents a viewpoint-related fusion method of massive vector data and 3D terrain, in order to superpose the massive 2D vector data onto the undulating multi-resolution 3D terrain precisely and efficiently. First, the method establishes an adaptive hierarchical grid spatial index for vector data. It will determine the geographic spatial relationship between vector data and the tiles of 3D terrain in the visible area; secondly, this method will use the improved sub-pixel graphics engine AggExt to generate textures for vector data that has been bound to terrain tiles in real time; Finally, considering that a large amount of vector data will generate a lot of 2D textures in the computer memory, the method should release the “expired” vector textures. In this paper, in order to take into account the real-time convergence and the smooth interactivity of 3D scenes, this method will adopt a multi-threading strategy. The experimental results show that this method can realize the real-time and seamless fusion of massive vector objects on the 3D terrain, and has a high rendering frame rate. It can also reduce the aliasing produced by traditional texture-based methods and improve the quality of vector data fusion.