This paper addresses the problem of multispectral-image compression and proposes a new encoding scheme, based on the classification of the multispectral image into regions of homogeneous land cover. Separate regions resulting from classification are efficiently encoded by means of conventional transform coding techniques, whereas the classification information is compacted resorting to spatial prediction and Ziv-Lempel coding. Simulation results show that the proposed algorithm assures both a high compression ratio and a good reproduction quality, with a reasonable computational complexity.
Multipectral-image coding by spectral classification / Finelli, M.; Gelli, Giacinto; Poggi, Giovanni. - (1996), pp. 605-608. (Intervento presentato al convegno 1996 International Conference on Image Processing (ICIP-1996) tenutosi a Lausanne (Svizzera) nel Settembre) [10.1109/ICIP.1996.560935].
Multipectral-image coding by spectral classification
GELLI, GIACINTO;POGGI, GIOVANNI
1996
Abstract
This paper addresses the problem of multispectral-image compression and proposes a new encoding scheme, based on the classification of the multispectral image into regions of homogeneous land cover. Separate regions resulting from classification are efficiently encoded by means of conventional transform coding techniques, whereas the classification information is compacted resorting to spatial prediction and Ziv-Lempel coding. Simulation results show that the proposed algorithm assures both a high compression ratio and a good reproduction quality, with a reasonable computational complexity.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.