A despeckling algorithm combining curvelet and wavelet transforms of high resolution SAR images

SAR images despeckling is one of the most important research subjects in SAR images processing. Following discussion on the advantage and disadvantage of curvelet and wavelet transforms based SAR images despeckling, an algorithm which combined curvelet and wavelet transforms was proposed for despeckling of high resolution SAR images. Firstly, the curvelet and wavelet were used to denoise the original SAR image, then the related residual image between their results was acquired, which was then denoised through the curvelet transform. Finally the denoised residual image was added to the curvelet denoised image. Experimental results of TerraSAR-X high resolution images indicated that, compared with the Enhanced Lee filter, the Enhanced Frost filter, the wavelet transform based and curvelet transform based methods, the algorithm presented here had well despeckling effect, as well as fine edge-preserving ability.