Multiparallel decompression simultaneously using multicore central processing unit and graphic processing unit

The discrete wavelet transform (DWT)-based compression algorithm is widely used in many image compression systems. The time-consuming computation of the 9/7 discrete wavelet decomposition and the bit-plane decoding is usually the bottleneck of these systems. In order to perform real-time decompression on a massive bit stream of compressed images continuously down-linked from the satellite, we propose a different graphic processing unit (GPU)-accelerated decoding system. In this system, the GPU and multiple central processing unit (CPU) threads are run in parallel. To obtain the maximum throughput via a different pipeline structure for processing continuous satellite images, an additional balancing algorithm workload has been implemented to distribute the jobs to both CPU and GPU parts to have approximately the same processing speed. Through the pipelined CPU and GPU heterogeneous computing, the entire decoding system approaches a speedup of 15x as compared to its single-threaded CPU counterpart. The proposed channel and source decoding system is able to decompress 1024 x 1024 satellite images at a speed of 20 frames/s. (C) 2013 Society of Photo-Optical Instrumentation Engineers (SPIE)