GPU-based Global Multi-resolution Terrain Data Visualization In Realtime

The core issues for global terrain visualization is how to efficiently organize & manage massive multi-resolution global terrain data, and use the limited processing power of computer hardware to achieve better real-time rendering. This paper started with the discrete global grid system(DGGs), conbined with the latest graphics hardware features and drawbacks of existing systems, systematically studied the GPU-based global multi-resolution terrain data vasualization in realtime. The main research contents and achievements include:1) Research on massive multi-resolution global terrain data organization and management methods. First, by summarizing the characteristics and drawbacks of existings DGGs, conbined with the new graphics hardware features and the efficient real-time rendering techniques requirements of DGGs, we selected HEALPix DGGs model as the basis for this study. Then, according to application needs, we improved the storage requirements and indexing of the original model, the maximum resolution that can be stored in the DGGs has increased from 0.012 meters to 7.07×10-13 meters, the storage requirement can be reduced to at least 1/3 of the original. Finally, we designed a scheme to organize and manage massive multi-resolution global terrain data.2) Research on GPU-based real-time global terrain data rendering algorithm. Combined with the improved HEALPix model and existing terrain rendering algorithms, this paper proposed a improved HEALPix based global multi-resolution data rendering in real-time scheme. This scheme stores terrain elevation data in the GPU memory as texture data, and generate the terrain mesh by GPU hardware algorithm at real-time rendering, and then complete the 3D terrain rendering in the GPU shader programs. In this scheme, the rendering process is done almost entirely in the GPU, full use of the GPU hardware subdivision capabilities and GPU highly parallel characteristics, which significantly improves the scene rendering speed and reduces the use of memory and CPU resources. Compare existing schemes, this solution solves the visual distortion near the poles, improves the real-time rendering speed and effects, simplifies the complexity of the rendering engine design.3) Design and implementation of the prototype system. In order to validate the proposed scheme, we design and implement global terrain data management systems and global terrain rendering system, using the former to establish a global terrain elevation database and global topography image database, using the latter to establish a database for real-time rendering and model test, the results are preferable.