摘要
真实感绘制中使用大量高分辨率纹理图像将受到存储空间及带宽限制,一般压缩图像由于解压过程复杂,难以满足绘制的实时性要求。针对该问题,提出可增长的自组织处理(SGSOP)遥感纹理压缩及对压缩纹理的并行硬件解压方法。SGSOP利用自增长和侧向联想处理方式构建压缩纹理码书,降低了计算复杂性,提高了码书的质量。基于硬件的解压重构能充分发挥图形硬件的SIMD并行处理特性、提高解压速度。
The storage space and transmission bandwidth becomes the bottleneck while using a large number of high resolution texture pictures in realistic rendering. If compressed picture are used, the real-time requirement by the rendering is hard to satisfy. Aiming at solving the problem, the paper presents a Self-Growing Self-Organizing Process (SGSOP) remote sensing texture compression method, and the way to parallel decompress the compressed file using GPU. SGSOP utilize the way of self-growing and lateral association process to make the compressed texture sample set. The computational complexity is decreased compared to VQ. As SGSOP maintains the random access feature of the compressed data, the data can be parallel reconstructed based on the SIMD parallel process feature of GPU.
出处
《系统仿真学报》
CAS
CSCD
北大核心
2009年第S1期42-45,50,共5页
Journal of System Simulation
基金
国家"973"计划资助(2009CB723803)
国家"863"项目资助(2006AA01Z309)
