International Journal of Computer Applications |
Foundation of Computer Science (FCS), NY, USA |
Volume 69 - Number 14 |
Year of Publication: 2013 |
Authors: Munesh Singh Chauhan, Sharmi S, Abeer Marhoon Al-sideiri |
10.5120/11910-8015 |
Munesh Singh Chauhan, Sharmi S, Abeer Marhoon Al-sideiri . On-board Implementation of Fractal Compression of Satellite Images using Distributed Networked GPUs. International Journal of Computer Applications. 69, 14 ( May 2013), 17-20. DOI=10.5120/11910-8015
On-board image compression has been a growing trend in most recent satellite missions. Since majority of satellite applications deal with imagery; compression of images due to limited on-board data storage mediums has become a necessity. The idea of treating satellite imageries as fractals and then encoding them provides an efficient way of conserving bandwidth and per-bit storage costs. Fractal encoding is characterized by slow encoding times which somehow had hindered its popularity in spite of its impressive compression ratio scaling many orders as compared to JPEG. In order to circumvent this handicap, Fractal compression is implemented using powerful GPUs (Graphical Processor Units) that are capable of reaching astronomical computing speeds of around 900 GFLOPS (Quadro Graphic Cards from NvidiaTM) with internal memory bandwidth ranging to 100 GB/s. This astounding parallel capability is probed to be used on board systems, providing much needed boost for image compression. As the decoding part of compressed fractal images is almost instantaneous, this part can be handled without any specific hardware at the ground station level. Further, the issue of on-board data storage mechanisms is discussed with emphasis on use of HDD instead of SSD and flash memories. In sum, the prime aim is to provide a seamless image compression mechanism coupled with de-compression at ground station level thus providing real-time streaming of satellite images from satellite to the ground.