Speckle is an important characteristic of SAR image that obstructs the efficient interpretation and thus its applications. So

the image filtering is indispensable in the SAR related applications.In the SAR filtering

there is a persistent contradiction:constraining noises VS.keeping image details.In other words

the optimal goal of image filtering is to constrain the noises while keeping the image details at the same time. The traditional filters are mostly those in the space domains.Though developments of filters in space domain have gradually shortened the span of the two sides

there still are some difficulties in overcoming such contradiction.So

the filters in frequency domain have attracted more and more attention. In wavelet frequency domain

the image frequency can be efficiently decomposed and then the noise can be restricted.In the paper

the de-noise filters in frequency are analyzed and compared.Then based on the concepts of threshold-optimizing and edge-detecting in wavelet frequency domain

an new wavelet de-noise method is presented.The experiment shows that the new filter is efficient at both noise constraining and image detail keeping.