Synthetic Aperture Radar Image Speckle Denoising Based On Independent Component Analysis

The polarimetric synthetic aperture radar (SAR) system is an important research in thefield of remote sensing. Polarimetric SAR is a set of coherent imaging radar, which receivesthe total scattering echo is not entirely dependent on the scattering coefficient of the target,but will appear larger random changes in the vicinity of the target scattering coefficient largerrandom changes.This random variation called speckle noise in the SAR images.The imagesignal-to-noise ratio is decreased with the increasing of the speckle noise. Increased specklenoise will lead to the edge of the SAR image blur, and even lead to the loss of polarized targetcharacteristics. Speckle noise filtering is the first step in post-processing polarimetric SARimaging.Noise suppression method has been a subject worthy of study the polarimetric SARsignal processing field.There are a lot of scholars speckle filtering methods, to some extent, the speckledenoising. However, in the field of noise reduction, edge preservation and adaptive parametercomputing, these methods have some flaws and difficulties.This needs to be furtherimproved.The independent component analysis (ICA) as a blind signal separation is the mostcommon and most effective treatment technology,and has been used in many signalprocessing field. Through statistically independent thinking, ICA use the sensor observationdata to isolate several independent components.The main work of this paper is to use multi-dimensional observation data as input signalsof ICA, which is in polarimetric SAR image on quantitative. ICA is a separation methodbased on higher order statistics and the basic model. It is able to isolate the image of thespeckle noise, and to achieve polarimetric SAR image speckle denoising.This richpolarimetric SAR image enhancement method. The specific work of the paper is as follows:First, the paper summarize the research value of the polarimetric SAR image, specklenoise model and denoising principle, and analysis existing denoising methods. Elaborated thethe ICA algorithm statistical model and the solution process.By the objective function to buildthe ICA an adaptive algorithm. Second, we will analysis whether the polarimetric SAR imagecan be processed by ICA algorithm, derived the ICA approach speckle denoising process.Simulation analysis were used the traditional spots Denoising and ICA algorithm to deal withthe polarimetric SAR images. Summed ICA algorithm in image denoising and edge detailpreservation is superior to the traditional algorithm. Finally, Through the using of theimproved ICA gradient learning algorithms to deal with three sets of polarimetric SAR images, We analyze this improved method is superior to the typical ICA algorithm in imageenhancement and edge details remain.We study the theory of polarimetric SAR and ICA algorithm. Through simulationexperiments, we prove that the the ICA optimization algorithms can effectively suppress noisein the processing of polarimetric SAR image speckle. This improved ICA learning algorithmhas unique advantages in terms of polarization information is saved and edge detail, whichcan better enhance polarimetric SAR images.