GPU-Based Large Seismic Data Parallel Compression

Abstract

The motivation of compressing seismic data is to reduce enormous storage space and transmission bandwidth. In this paper, a parallel compression technique is presented for large seismic data compression based on graphics processing unit (GPU). We take advantage of combinations of GPU-based parallel processing and 3D set partitioning in hierarchical tree (3D SPIHT) to accelerate the whole compression process. Experimental results show that our method achieve fast compression for very large seismic data (2.56GB) on standard PC hardware.