Study About Improvement Of SAR Image Quality

Synthetic Aperture Radar (SAR) has an all-weather and day/night capability togenerate a two dimensional high-resolution image of the targets over a long range, so itis widely used in both civilian and military fields. With the further investigation inimaging radar, people demand to obtain better qualities of SAR images, includinghigher resolution, easy readability and clear visual effect. According to this, thedissertation mainly investigates how to improve the quality of SAR images by themeans of post imaging processing methods from the theoretical perspectives.Meanwhile some real data are provided to test our algorithms'effect.The first chapter is mainly associated with the theories of SAR imaging based onimaging model, then reviews the development history of SAR post imaging process,introduces the existed research at present, and at last concludes the whole contents ofthe dissertation.The second part discusses the autofocus technique about one kind of moving targetsin SAR images. Based on the parameters estimation of SAR scene's scatters, a newautofocus algorithm is proposed for moving targets, which have no range alignment butnearly two orders of polynomial phase errors exist in azimuth. Through alternateiteration, this approach can compensate the exact errors with less range cells.The third chapter studies some super-resolution algorithms in modern spectrumestimation. APES and CAPON algorithms are analyzed and used to make experimentswith SAR images. Up to now, no general comments and standards are given to explainthe concepts of super-resolution of radar imaging, so these methods'values in practicalapplications are still worth further discussions.The fourth part discusses the speckle suppression in SAR images. The mechanism ofspeckle produced is studied and some existed filtering approaches are discussed. Then anew adaptive filtering algorithm is proposed to remove speckle in SAR images based onthe inter-scale correlation in wavelet domain. The real data prove that this approach hassuperior performance than other conventional methods.The last chapter concludes all the contents of this dissertation and points out theproblems left to be further discussed.