International Journal of Applied Mathematics and Machine LearningAuthor's: Yanling An, Jiaxin Li, Qi Zhang, Xiaole Ma, Jiao Pang and Shuai Cong
Pages: [89] - [99]
Received Date: June 25, 2019
Submitted by: Jianqiang Gao.
DOI: http://dx.doi.org/10.18642/ijamml_7100122077
Synthetic aperture radar (SAR) is widely used in military and civil fields. The main task of SAR image denoising is to suppress speckle. Combining the advantages of non-subsample shearlet transform (NSST) and weighted nuclear norm minimization (WNNM), we propose a SAR image denoising algorithm based on multi-scale weighted nuclear norm minimization. Firstly, speckle is transformed into additive noise by logarithmic transform. Secondly, NSST is used to decompose the image to low and high sub-bands coefficients, and we can denoise NSST coefficients by WNNM. Finally, NSST is used to reconstruct denoised images. The experimental results show that the proposed algorithm retains the local structure of the image better and achieves a good visual effect.
SAR image denoising, non-subsample shearlet transform, weighted nuclear norm minimization.
