A Study Of Speckle Reduction Algorithms Of Synthetic Aperture Radar Image

Synthetic aperture radar based on coherent imaging principle is a high-resolution imaging system. Owing to the unique advantage, it is widely used in military and civilian areas. Because of the inherent defects in the coherent imaging theory, the serious speckle noise inevitably exists in the synthetic aperture radar image, which very much affects the following image processing. Consequently, the research with the technology of speckle reduction is becoming a hot topic on the application of the synthetic aperture radar. The objective is removing speckle noises with great efforts and at the same time preserving the edge texture of image to the utmost extent.The primary coverage of this passage is first analyzes the feature of radar images and the formation mechanisms of speckle, then gives a systematical summary of current famous noise reduction methods of spatial filters such as LEE filter, KUAN filter, frost filter, GAMMA MAP filter, improved K-mean filter, and then does experiment simulations for parts of spacial filters. Experiments shows there are two problems exist in it, one is in the same window so long as pixel has the equidistance with central pixel, then it will takes the same effect on output central pixel which results in fuzzy edge details; And another is this filter has a common shortcomings in preserving edge details and its filter performance depends much on the size of window.Afterwards,this passage gives an evaluating indicator which determine quality of the filters. For the Frost filter's inherent defect, an improved method is proposed, which mainly combines the statistical information of the filtering window and the statistical information of each pixel of the window to adaptively compute the filter parameters. While compared with the Frost filtering of the different filtering parameters, this modified method has a better performance in terms of speckle noise suppressing and fine details preserving. In light of inherent defects of spacial filters, a modified method is raised, that is combining adaptive sliding window and detail preserving to reduce noises. In order to gain a due end, two actions are done:(1) gives a noises-reduction algorithm which combines adaptive sliding window and edge detection mechanisms;(2)poses a noises-reduction algorithm which joins adaptive sliding window with edge preserving optimization mechanisms. These two algorithms combine fixed window and adaptive window, square window and feasible window, noises suppression and edge preserving, experimental results show that this method provides a significant improvement on other existing filters. At the end of this paper, a preview of probable research areas on speckle noises reduction for SAR images in the foreseeable future is outlined.