Speckle Reducing And Understanding Spaceborne Synthetic Aperture Radar Image

Spaceborne Synthetic aperture radar (SAR) is useful for military reconnaissance and civil activity, so it has practical meaning and application prospect to study feature extraction and object recognition method for spaceborne SAR images. In this paper, we study the spaceborne SAR image filter and understanding methods which are the indispensable steps in the process of SAR image, including the spaceborne SAR image filter,spaceborne SAR image classification,spaceborne SAR image edge detecting, and validate them by experimental results.A good image filter approach is to achieve despeckling effect with reducing the information losing. Many filters have been developed for speckle reduction. In this dissertation, the most well-known filters are analyzed and a novel adaptive despeckling approach is presented. The method bases on the adaptive wavelet transforms and the Wiener filter. Finally, we make a comparison of the traditional statistics filter with this wavelet transform filter in the process of spaceborne SAR image despeckling.Two key problems at spaceborne SAR image classification are proposed: how to choose classifying characters and how to select classifying method. A classification method for spaceborne SAR image based on wavelet energy distribution is proposed. After analysis the characters of the SAR image, the spaceborne SAR image is decomposed with stationary wavelet transforms. Each of these parts is composed as the energy-distribution-ratio characters. At last, the characters are unsupervised classified by the FCM and then achieved the result image. The experiments compare this method with the other classification method on spaceborne SAR images.Due to the presence of speckle, which can be modeled as a strong, multiplicative noise, edge detection in spaceborne SAR images is extremely difficult. Meanwhile the edge detectors developed for optical images are inefficient at this situation. A new edge detecting method for SAR images is proposed. The image is firstly decomposed with multi-scale stationary wavelets, and the transformed components corresponding to the same pixel are used to form a feature vector. The image is segmented with FCM method according to the feature vectors, and then the edge pixels are checked out on the segmented image. Experiments with normal images and images contaminated with different intensive noise have demonstrated that this algorithm is effective and robust.A SAR image processing platform is designed with MATLAB. This platform mainly contains three components: SAR image speckle reducing,classification and edge detecting. It realizes some classic SAR image processing methods and the methods which put forward in this paper.In the end, conclusions are made and the possible future work is explained.