Abstract:

The transmission, storage and processing of large-scale terrain data limit the real-time rendering speed of terrain, this paper presents an algorithm based on fast layered Discrete Cosine Transform（DCT） embedded zerotree coding to achieve terrain data compression, and then completes terrain rendering with view-based dynamic Levels of Detail（LOD）. Firstly, the fast DCT is used to replace the traditional DCT to transform the terrain data, and the different frequency sub-bands are reconstructed according to the spatial position, the low frequency is used as the input of the lower layer after the inverse DCT, repeat the above steps until the requirements are met. Then zero-tree coding is applied to the obtained multi-resolution frequency band. Finally, the terrain is re-rendered with dynamic LOD. The experiments show that the algorithm is relatively simple, compression rate and frame rate are improved effectively, which is more in line with human visual perception.

Key words: large terrain, fast DCT（Discrete Cosine Transform）, embedded zerotree coding, data compression

摘要：

为了解决大规模地形实时渲染中地形数据的传输、存储和处理效率问题，提出了基于快速层式DCT的嵌入式零树编码对地形数据进行压缩，然后结合基于视点的动态LOD完成地形绘制。用快速DCT代替传统DCT对地形数据进行变换，按空间位置重组不同频率子带，对低频做逆DCT后作为下层输入，重复上述步骤直到满足要求；在获得的多分辨率频带上执行零树编码；结合动态LOD重新渲染地形。实验结果表明，该算法计算相对简单，压缩率与帧速率得到了有效的提升，绘制效果更加契合人眼的视觉感受。

关键词: 大规模地形, 快速离散余弦变换, 嵌入式零树编码, 数据压缩