SAR Image Processing based on Fast Discrete Curvelet Transform

Curvelet transform is a new kind of multiscale analysis algorithm which is more suitable for image processing, as compared with Wavelet it can better analysis the line and curve edge characteristics, and it has better approximation precision and sparsity description, also has good directivity. This paper introduces that remote sensing image speckle reduction based on Curvelet transform. Synthetic Aperture Radar (SAR) image is easily polluted by speckle noise, which can affect further processing of SAR image. This paper put forward method of SAR image speckle deduction based on Fast Discrete Curvelet Transform (FDCT). This method firstly transform R image to Curvelet domain by using FDCT, and get the Curvelet coefficient, then estimate the Curvelet coefficient threshold of different scale and direction by using adaptive threshold method, treatment on Curvelet coefficient with hard threshold and soft threshold respectively, and the last recovery the original image by using IFDCT. This paper uses this method to single-look SAR image and compare with Wavelet de-noising method, the result shows that the effect of Curvelet flitting is better than Wavelet flitting, and the soft threshold is better than hard threshold