Title :
Analysis of multiplicative speckle models for template-based SAR ATR
Author :
Kaplan, Lance M.
Author_Institution :
Dept. of Eng., Clark Atlanta Univ., GA, USA
fDate :
10/1/2001 12:00:00 AM
Abstract :
It has been noted that signal aperture radar (SAR) imagery of clutter exhibits Rayleigh multiplicative noise due to speckle. We use a database of MSTAR target chips to verify that the noise is multiplicative rather than additive for all regions in the chip. Then, by examining histograms corresponding to the noise residuals, we show that a Weibull distribution that is almost Rayleigh best fits the data. However, when we restrict the analysis over the target region, the log-normal and Rayleigh models fit the noise equally as well. This can be attributed to scattering mechanisms that are unstable over five degrees of aspect angle
Keywords :
Weibull distribution; maximum likelihood estimation; mean square error methods; radar clutter; radar computing; radar imaging; radar target recognition; speckle; synthetic aperture radar; MSTAR target chips; Rayleigh models; Rayleigh multiplicative noise; Weibull distribution; approximate Rayleigh parameters; clutter; deterministic structure; histograms; log-normal models; maximum likelihood estimation; multiplicative speckle models; noise residuals; radar imagery; template-based SAR ATR; unstable scattering mechanisms; Additive noise; Histograms; Image databases; Radar clutter; Radar imaging; Radar scattering; Rayleigh scattering; Speckle; Synthetic aperture radar; Weibull distribution;
Journal_Title :
Aerospace and Electronic Systems, IEEE Transactions on
