快速EZC-DCT地形压缩算法的并行优化  

Parallel optimization of fast EZC-DCT terrain compression algorithm

作　　者：高保禄[1] 窦明亮 张冰[1] 

机构地区：[1]太原理工大学计算机科学与技术学院,山西太原030024

出　　处：《计算机工程与设计》2018年第1期189-193,共5页Computer Engineering and Design

基　　金：虚拟现实技术与系统国家重点实验室开放基金项目(BUAA-VR-15KF-17)

摘　　要：为能够在大规模地形实时渲染中提高渲染及数据压缩的速率,提出一种利用GPU并行优化的快速EZC-DCT地形压缩算法。采用二维快速DCT变换代替EZC-DCT算法中的DCT变换,在利用GPU对算法进行并行加速的基础之上,对算法的并行方案进行优化改进,更加有效地利用GPU强大的并行计算能力,分担CPU的负荷,快速完成相关计算。实验结果表明,该算法帧速率比原EZC-DCT方法提升约10个百分点,满足地形渲染的实时性要求。To improve the speed of rendering and data compression in real-time rendering of large algorithm based on GPU parallel optimization was proposed. Two-dimensional fatt DCT transform was used instead of DCT transform in EZC-DCT algorithm. On the basis of using GPU to speed up the algorithm, the parallel algorithm was optimized and improved. The powerful parallel computing ability of GPU was used, which not only shared the load of CPU, but alsocompleted the correlation calculation quickly. Experimental results show that the proposed algorithm can by 10% compared with original EZC-DCT algorithm, and meet the requirements of real-time terrain rendering.

关 键 词：EZC-DCT算法 GPU并行计算 DCT变换 并行优化 地形渲染 

分 类 号：TP391.9[自动化与计算机技术—计算机应用技术]