Almost translation invariant wavelet transformations for speckle reduction of SAR images

Two wavelet transformations are use for speckle reduction and enhancement of synthetic aperture radar (SAR) images. First, a discrete wavelet transformation (DWT) based on oversampled filter banks is-used. The oversampled DWT. is called a double-density DWT (DD-DWT) and is based on a single-scaling function (lowpass) and two distinct wavelet functions (highpass). Second, a discrete wavelet transformation based on two dual real wavelet trees is applied. Each tree produces a set of real DWTs, which together form the complex wavelet transformation (CWT), i.e., a transformation with both real and imaginary parts. Both of these DWTs are almost translation invariant and are useful for speckle reduction through their subband images, and the speckle reduction is obtained by thresholding the subband image coefficients of the digitized SAR images. A thresholding method based on the use of nonlinear functions, which are adapted for each selected subband, is used. The nonlinear functions are based on sigmoid functions. The denoising method presented shows great promise for speckle removal and, hence, can provide good detection performance for SAR-based recognition.