Integer algorithm of lapped biorthogonal transform compression method with GPU
-
摘要: 地形数据的压缩/解压是大规模地形实时绘制方法的关键步骤,与绘制效率密切相关.通过对压缩/解压方法核心重叠双正交变换的分析,采用重叠双正交变换的整数提升方法将变换中的浮点数操作转换为整数操作及移位操作.使用支持图形处理单元(GPU,Graphic Processing Unit)通用计算的CUDA(Compute Unified Device Architecture)对变换过程及编码过程进行加速.针对数据超出显存容量的情况,采取数据分块的方法将数据分别载入显存进行变换与编码以完成对整体数据的处理.实验结果表明,基于GPU加速的重叠双正交变换整数提升方法的压缩算法有效提高了地形数据处理的效率,并加快了大规模地形绘制速度.
-
关键词:
- 重叠双正交变换整数提升 /
- 压缩/解压 /
- 大规模地形实时绘制 /
- 图形处理单元通用计算
Abstract: Compression/Decompression of huge terrain data is an important component of GC(Geometry Clipmap) method which has significant impact on rendering speed. Integral lapped biorthogonal transform(binLBT) method was used to reduce the complexity of computation by converting float point operations which spent lots of time in LBT method to integer operations and shift operations. In order to avoid the condition the data bigger than memory in graphic card, the method divided the data into some pitches which fit the memory and load one pitch each time, and then used CUDA(compute unified device architecture) to accomplish binLBT step which supports the parallel features of GPGPU(general purpose computing on graphics processing units). Experiment results show that this method based on binLBT by GPU can improve the efficiency of compression/decompression of huge terrain data and accelerate the rendering speed of GC. -
[1] Gobbetti E,Marton F,Cignoni P,et al.C-BDAM-compressed batched dynamic adaptive meshes for terrain rendering[J].Computer Graphics Forum,2006,25(3):333-342 [2] Losasso F,Hoppe H.Geometry clipmaps:Terrain rendering using nested regular grids //ACM Transactions on Graphics.California:ACM Press,2004,23(4):769-776 [3] Asirvatham A,Hoppe H.Terrain rendering using GPU-based geometry clipmaps,GPU Gems 2[M].New York:Addison-Wesley,2005:19-33 [4] Malvar H S.Fast progressive image coding without wavelets //Storer J A.Proceedings of the Conference on Data Compression. Washington:IEEE Computer Society,2000:243-253 [5] Malvar H S. Biorthogonal and nonuniform lapped transforms for transform coding with reduced blocking and ringing artifacts[J].IEEE Trans Signal Processing,1998,46:1043-1053 [6] 钟广军,成礼智,陈火旺.双正交重叠变换的整数实现算法与图像压缩[J].电子学报,2001,29(11):1475-1477 Zhong Guangjun,Cheng Lizhi,Chen Huowang.Integer algorithm of lapped biorthogonal transform and image compression[J].Acta Electronica Sinica,2001,29(11):1475-1477(in Chinese) [7] Ryoo S,Rodrigues C,Baghsorkhi S,et al.Optimization principles and application performance evaluation of a multithreaded GPU using CUDA //Chatterjee S.Proceedings of the 13th ACM SIGPLAN Symposium on Principles and Practice of Parallel Programming. Salt Lake City:ACM Press,2008:73-82 -
点击查看大图
计量
- 文章访问数: 3941
- HTML全文浏览量: 243
- PDF下载量: 3087
- 被引次数: 0
下载:
