A speckle reduction algorithm by soft-thresholding based on wavelet filters for SAR images
Jun Zhang, Cheng Xueguang, Liu Jian · 2002
This paper presents a speckle reduction algorithm by soft-thresholding for SAR images in multiscale and multiorientation using nonlinear wavelet analysis. Firstly, a base is chosen according to SAR image features, which is used to make a SAR image 2D wavelet transformation. Furthermore, we apply Donoho's (1994) soft-thresholding theory and extend it to 2D SAR image processing, then select a suitable soft-threshold aimed at SAR image features. After, we remove speckles by 2D wavelet filtering. Lastly, we invert the pyramid filtering, recovering the SAR image. The experimental results of removal of speckle on SAR images are as expected.