Study Of Software-Methods Of Filtering SAR Speckle Noise

Synthetic Aperture Radar (SAR) is one of the most successful microwave sensors in 1950s.And it is most rapidly developed. SAR has been get broad recognition and application in manydomains such as surveying and mapping. The study of filtering the speckle noise of SARimage is one of important tasks in SAR image processing.This paper analyzes at length the mechanism giving rise to the speckle noise and introducesthe mathematical model of the speckle noise. On the base of the multiplicative noise model, itgets some statistical characteristics. Then it summarizes some representative filters in recentyears, such as the multi-look processing, the mean filter, the median filter, Lee filter, Kuanfilter, Frost filter, Sigma filter, Gamma MAP filter. It introduces the methods and characteristicof these filters at length.This paper analyzes and summarizes two models of adaptive filters. It proposes the starts forbuilding new adaptive filters. Then it proposes three filters-sub-window filter, improvedSigma filter, logarithmic mean filter.This paper also proposes a set of quantitative evaluation index to evaluate the filtered effects.It can be used to direct to filter speckle noise and to build new adaptive filters. Then it carriesout the contrast tests using the ERS single-look space-borne SAR data, Radarsat multi-lookspace-borne SAR data, high resolution airborne SAR data. And it does quantitative evaluationand analysis.At last, this paper discusses for the phase noise in InSAR. In particular, it proposes sine &cosine conversion to resolve the problem of the phase jump. And it creates two new filters forfiltering the phase noise-sine&cosine mean filter and sine&cosine median filter based on thesine & cosine conversion. Then it carries out contrast test. And it is proved that the twoproposed filters have good effect for the phase noise in fringe.