A Dynamic Tiles Loading and Scheduling Strategy for Massive Oblique Photogrammetry Models

With the development of oblique photogrammetry technology and automatic 3D model reconstruction technique, massive urban 3D models can be quickly and efficiently generated. The demand is urgent for 3D display capability on a Web-based virtual Globe for massive oblique photogrammetry models. In order to achieve the visualization of the data, in the process of production, the origin data should be partitioned into tiles firstly and then HLOD algorithm is applied on each tile, so that the output data has a chunk-based multiple resolution data structure. However, there is no standard spatial data structure for 3D data segmentation. Therefore, for some tiled data., there may be tiles with large data size, so the problems such as memory overflow and browser crash may occur due to excessive data loaded in the process of visual interaction. In order to guarantee the visual interaction on different terminal devices for different oblique photogrammetry models, this paper firstly designs a 3D tile model suitable for network transmission and application, and then analyzes the scheduling of Cesium 3D Tiles and its existing problems. A 3D tiles dynamic loading and scheduling strategy is proposed based on terminal memory size and screen space error. This strategy monitors the load of data in real time during the visual interaction process. When the data loaded exceeds the preset threshold, some high resolution tiles will be unloaded, using slight lower resolution tiles to take the replace. Experiments prove that adopting this strategy can ensure the stability of the loaded data amount and better visual effects during the visual interaction, and the browser can always operate normally.