A SAR Image De-Noising Method Based on Stationary Wavelet Transform Directional Energy Threshold Filtering
Zhang Yuye, Wang Yingying, Wang Yiheng · 2022
Because the speckle noise of SAR image is non-Gaussian distribution, it cannot be directly processed by the denoising technology of the optical imaging system. At present, there is no real ideal algorithm widely used in SAR image denoising. Based on the analysis of the advantages and disadvantages of the current popular denoising methods, this paper proposed a wavelet domain SAR image denoising method. In order to overcome the shortcomings of the discrete wavelet transform, such as lack of translation invariance and limited directional selectivity, this method uses a stationary wavelet transform to decompose the image into a low-frequency approximation signal and high-frequency detail signal; for the approximation signal with less noise information, the enhanced e-lee filter is used to denoise; for the detail signal, adaptive threshold filtering is carried out according to the directional energy characteristics of speckle noise in wavelet domain Finally, the denoised SAR image is reconstructed. Experimental results showed that the method had a strong de-noising ability and good edge preservation.