结合全变差和小波变换的图像去噪算法及在SAR图像上的应用

In Synthetic Aperture Radar (SAR) images, a common Speckle noise interference seriously affects the subsequent processing and analysis of SAR images, and also interferes with the visual effect. In order to solve this problem, this essay introduces a hybrid denoising method that combines wavelet transform, integer-order total variation (TV) and variable exponential fractional total variation (FTV). Wavelet transform can focus on subtle changes in the image and can be decomposed and reconstructed in a short time. First, the SAR image is decomposed into a low-frequency image and three high-frequency images by wavelet decomposition. Then use integer-order total variation and variable exponential fractional total variation to process high and low frequency images respectively. Finally, the denoised image is reconstructed by inverse wavelet transform.This method can not only remove speckle noise effectively, but also preserve edge and texture details very well. Many experiments on real SAR images have verified that this method is effective and has a certain practical value.