基于提升小波的大地形累进压缩及实时渲染

Lifting Wavelet-Based Progressive Compression and Real-Time Rendering to Large Terrain

为了减轻实时交互的大地形可视化系统面临的海量数据存储、传输及渲染压力,提出一种基于提升小波实现数据累进压缩与实时粗粒度LOD渲染结合的算法.首先建立地形块的提升小波变换模型,将其简化和精化操作映射为相应的小波变换;然后基于以地形块为单位的小波变换建立块的四叉树层次结构,将构建过程中产生的小波系数采用基于间接查表的方式累进压缩存储,以形成压缩的累进地形块层次结构.基于该结构,在实时递进构造视点相关的激活地形块的同时完成数据的累进载入与实时解压,实现大地形渲染,其中地形块数据的实时解压与小波变换基于GPU实现.实验结果表明,该算法实现了有效的数据压缩,具有很高的实时解压效率以及优化的实时渲染性能.

To alleviate the burden for massive-data storage, transmission and rendering in real-time large terrain visualization systems, based on lifting wavelet, an algorithm is proposed for integrating terrain dataset progressive compression and real-time aggregated LOD rendering. By the algorithm, the lifting wavelet transform model for terrain chunk is built to map the chunk simplification and refinement to corresponding wavelet transforms. Then the chunk-quadtree is constructed using the chunk-based wavelet transform model. The wavelet coefficients generated during the hierarchy construction are progressively compressed by an indirect look-up table technique. Based on the chunk-hierarchy compressed, the active chunks for a viewport are decided, then the datasets needed are incrementally loaded and uncompressed, with the adaptive terrain mesh produced at the same time, to perform terrain rendering. During rendering, real-time chunk uncompress and wavelet transforms are processed on GPU. Test results show that the algorithm can be used to compress datasets efficiently and its real-time uncompress efficiency and rendering performance are both excellent.