An image compression system for LEO satellites

Kriegler, Eduard (2003-12)

Thesis (MScEng)--Stellenbosch University, 2003.

Thesis

ENGLISH ABSTRACT: Data volumes produced by the next generation of earth observation sensors have increased greatly in recent years. Sensors are generating more data than can be easily stored onboard satellites and transmitted to the ground-stations. There are two strategies for solving this problem. The first is to process all images onboard the satellite, and only extract the useful or valuable information. The second is to use a compression algorithm to reduce the data volume. This thesis looks at both strategies and then focusses on an evaluation of the Embedded Zerotree Wavelet (EZW) algorithm, a wavelet-based lossy image compression algorithm, as a solution to reduce the data volumes. Possible hardware implementation strategies for this algorithm are also explored. Finally, a suggested implementation of the EZW algorithm is compared with the FlexWave-II system and with JPEG2000.

AFRIKAANSE OPSOMMING: Die data volumes wat deur die nuwe generasie van aardobservasiesensore geproduseer word, het dramaties vergroot in die laaste paar jaar. Daar word nou meer data geproduseer as wat aanboord van die satelliet gestoor kan word en meer as wat in die beperkte kommunikasietyd aan die grondstasie gestuur kan word. Daar is twee strategiee om hierdie probleem aan te spreek. Eerstens kan beelde aanboord die satelliet verwerk word om die belangrikste of waardevolste inligting uit te haal en die res van die data word dan geskrap. Die alternatief is om 'n beeldkompressie-algoritme te gebruik om die data te verminder. Hierdie tesis ondersoek hierdie strategieë en fokus dan op 'n evaluasie van die "Embedded Zerotree Wavelet" -algoritme. Die EZW-algoritme is 'n verlieserige, golfie-gebaseerde beeldkompressie-algoritme. Moontlike hardeware-implementeringsopsies word ondersoek en die resultate van een voorgestelde opsie word vergelyk met die FlexWave-II stelsel asook die nuwe JPEG2000-standaard.

Please refer to this item in SUNScholar by using the following persistent URL: http://hdl.handle.net/10019.1/53264
This item appears in the following collections: