Title :
A new AR-based technique to exploit SAR image correlation properties for rain forest classification
Author :
Lombardo, P. ; Pellizzeri, T. Macrì ; Pastina, A. ; Libranti, A.
Author_Institution :
INFOCOM Dept., Rome Univ., Italy
Abstract :
In this paper we develop a detector that exploits both variance and correlation properties of SAR images for the discrimination between regions with different characteristics. This detector is based on the generalized maximum likelihood approach applied to an autoregressive (AR) correlation model for the pixels of a homogeneous region. The application of the newly derived detector to C-band SAREX images of the Brazilian rain forest shows improved performance in the discrimination of clearings from forest
Keywords :
autoregressive processes; correlation theory; edge detection; forestry; image classification; maximum likelihood estimation; radar imaging; remote sensing by radar; synthetic aperture radar; AR-based technique; Brazilian; C-band SAREX images; SAR image correlation properties; autoregressive model homogeneous region; clearings; forest; generalized maximum likelihood; rain forest classification; variance properties; Covariance matrix; Data mining; Detectors; Electronic mail; Matrix decomposition; Maximum likelihood detection; Maximum likelihood estimation; Rain; Remote sensing; White noise;
Conference_Titel :
Geoscience and Remote Sensing Symposium, 2001. IGARSS '01. IEEE 2001 International
Conference_Location :
Sydney, NSW
Print_ISBN :
0-7803-7031-7
DOI :
10.1109/IGARSS.2001.978208
