Speckle Noise Removal for SAR Image Based on G0 Distribution Combining Total Variation and Total Curvature
Yunping Mu, Baoxiang Huang, Zhenkuan Pan, Huan Yang, Yajing Li · 2019
In order to remove speckle noise of synthetic aperture radar (SAR) images, we proposed a high order variational model based on G0distribution. Specifically, the model combines the total variation(TV) and total curvature(TC) regularizations. Besides, considering the terrain backscatter, we derived a new data fidelity term. Thus the new variational model can reduce the staircase effect, meanwhile effectively preserve the image features such as the edge, corner and fine details. Since the model has the characteristics of nonlinear, non-convex and non-smooth, we transformed it into an alternating optimization problem by importing auxiliary variables. Then, we designed a fast numerical approximation iterative scheme for proposed model. Qualitative and quantitative experiments on both synthetic and real SAR images were implemented to indicate the advantages of the proposed model and the high computation efficiency of the designed algorithm.