Edge-enhanced speckle suppression using curvelet transform with an optimal soft thresholding
Hongqiao Wang, Fuchun Sun, Yanning Cai, Zong-Tao Zhao · 2007
Aiming at the problem that common Curvelet may bring fuzzy edge when denoising, this paper proposes a novel edge-enhanced speckle suppression method using Curvelet transform with an optimal soft thresholding. The method combines the original images’ edge detection with the filtered image of each subband. In the edge area, first the edge information from the remnant image is extracted, and then the edge information is added to the filtered image. In the non-edge area, to increase the contrast with the edge and have a clean background, a mean filter is used to smooth the area. As a result, the edge enhanced image is gained. In order to reduce the affection of disadvantages from both hard thresholding and soft thresholding filtering to the Curvelet coefficients, the method also adopts an optimal soft shresholding for filtering. The simulation results show that the de-noising method can significantly reduce the speckles while enhancing the edge and the structure of the original SAR images with more smoothness in background areas. don transform