首页>
外国专利>
method for filtering noise of sar images speckle and relative device
method for filtering noise of sar images speckle and relative device
展开▼
机译:sar图像斑点噪声的滤波方法及相关装置
展开▼
页面导航
摘要
著录项
相似文献
摘要
A method of filtering SAR images obtained with a Synthetic Aperture Radar (SAR) from the speckle noise is described, said method comprising: - a first step of modeling the speckle noise of at least one set of pixels (x) of the SAR image by means of a gamma statistic (20) with radiometric resolution given by R r = 10 �¢ log 10 �¢ 1 + à ¼ , where R r is the radiometric resolution and µ and à are the average and standard deviations of the gamma statistic, respectively, - a second step of transforming at least one plurality of pixels (x) of the at least one set of pixels of the SAR image into a plurality of filtered pixels ( x ') by means of a data transfer function given by x '=G ±,² (1- x )=(1-(1- x ) ± ) ² , where x ' is the filtered pixel value, x is the pixel value of the SAR image, ± and ² are real parameters, - setting the invariance of the average µ of the gamma statistic for the filtered image, comprising said plurality of filtered pixels x ', so as µ'=(1-(1- µ) ± )² ‰¡ µ, where µ' is the value of the average of the gamma statistic (30) related to the filtered image; such a setting fixes the value of the parameter ² according to the parameter ± by the formula ² ± = log �¢ ¼ log ( 1 - 1 - ¼ ± , - setting the standard deviation Ã' of the gamma statistic related to the filtered image as ù = à / N , with N being the Equivalent Number of Looks which is intended to be obtained at the end of the filtering process, - setting Ã'=÷¦, where ¦ is the decrease factor of the initial deviation à due to the filtering process, - setting said factor ¦ equal to the transfer function derivative G ±,² about the point 1-µ, thus obtaining ¦ ± = ± �¢ ¼ ‹ log ¼ ‹ 1 - ¼ ± - 1 log �¢ 1 - 1 - ¼ ± ( 1 - 1 - ¼ ± ‰¡ 1 N , from which the value of the parameter ± is determined, - constructing the filtered image with the intensity values of the plurality of filtered pixels x ', ( Fig. 1 )
展开▼