|
Abstract
|
SAR radars are coherent radars which generate high resolution images. SAR systems are
capable of surface acquisition under all weather conditions. SAR images are corrupted by a
multiplicative noise called speckle. Speckle noise arises from image formation under coherent
radiation. The presence of speckle noise in SAR images is undesirable, since it makes scene
analysis and understanding very difficult.
It is not suitable for analyzing the non-stationary signal using Fourier Transform. Fourier
Transform has sinusoid as its basis function. This function does not have limited time duration.
When a small change occurs in the signal in the time domain, it will affect all the components
in the frequency domain. An other advantage of using wavelet bases instead of Fourier bases
is due to approximation power of wavelet series in signal with singularities since it would take
a larger number of Fourier coefficients than wavelet coefficients to represent a signal with discontinuities.
Typically, one constructs a twodimensional (2-D) wavelet by taking the tensor
product of one-dimensional (1-D) wavelets. This 2-D wavelet is still effective at approximating
point singularities (e.g., points in an image) but not for line singularities(e.g., edges in an
image). Multiresolution processing, either in separable wavelet domains or in nonseparable
domains, has proved a powerful tool in despeckling applications concerning SAR images. The
Contourlet Transform (CT), both in its decimated and in its nonsubsampled (NSCT) version,
is a powerful and versatile tool that allows a multiresolution and directional representation to
be achieved.
NSCT coefficients of SAR images typically exhibit strong non-Gaussian statistics, Chang
model wavelet(equally NSCT) coefficients with a generalized Gaussian distribution (GGD),
which matches well histograms of typical SAR images. However, GGD is not analytically
easy to work with due to its complex structure. Among alternative methods, a mixture density
of
|