Despite the numerous applications that have been developed for synthetic aperture radar (SAR) images, their degradation due to the presence of noise makes them very difficult to use. Kalman filtering techniques have been investigated for various image models. Following a model suggested by Azimi-Sadjadi and Bannour (see IEEE Transactions on Geoscience and Remote Sensing, vol.29, no.5, p.742-753, 1991) we propose a modified Kalman technique for SAR imagery. The large variation of the statistics of the noise is compensated for by a variance to mean ratio. The Kalman filter gain is, therefore, adapted to the statistics of the local area. Comparisons with the method mentioned above are given for real ERS-1 SAR data collected from Victoria, BC.<<ETX>>
[1]
E. Nezry,et al.
Adaptive speckle filters and scene heterogeneity
,
1990
.
[2]
Mahmood R. Azimi-Sadjadi,et al.
Two-dimensional adaptive block Kalman filtering of SAR imagery
,
1991,
IEEE Trans. Geosci. Remote. Sens..
[3]
JOHN w. WOODS,et al.
Kalman filtering in two dimensions
,
1977,
IEEE Trans. Inf. Theory.
[4]
M. Azimi-Sadjadi,et al.
Two-dimensional block processors - Structures and implementations
,
1986
.
[5]
J. Goodman.
Some fundamental properties of speckle
,
1976
.
[6]
R. Roesser.
A discrete state-space model for linear image processing
,
1975
.