Research On SAR Image Edge Detection And Speckle Reduction

SAR(Synthetic Aperture Radar)images are made of a synthetic aperture radar imaging system to obtain high-resolution image,it plays an irreplaceable role,especially the optical sensor imaging difficult areas.SAR Image Edge Detection as a basic research tool for SAR image processing and computer vision,which is widely used in the field of feature detection and texture analysis.However,the presence of SAR image speckle generated a lot of negative impact on the detection and feature extraction in image processing.So the SAR image speckle reduction and SAR image edge detection algorithm based on speckle reduction have a very important practical significance.Currently,Research on SAR image speckle reduction and edge detection algorithms was expanded from three aspects which are the spatial domain and transform domain and anisotropic diffusion.This paper proposed SAR image edge detection algorithm from two aspects of airspace domain and transformation domain got four edge detection algorithms and a SAR image speckle reduction from partial Differential anisotropic diffusion.Four kinds of SAR image edge detection algorithm were proposed in this paper,the algorithm ? proposed adaptive parallel window SAR image edge detection algorithm based on level window of constrained anisotropic Gaussian kernel,it is composed of two horizontal windows anisotropic Gaussian kernel function with constraint function,and by constraining function to make the window adaptively select a different window types to edges or homogeneous area in SAR image.The algorithm ? makes false edges in the edge region and homogeneous areas are effectively reduced.The introduction of information entropy as a composite statistic in the algorithm II.When using information entropy image segmentation can retain more characteristics of image information,to ensure that the SAR images weak edge information is not missed in the process.It finally makes wavelet packet decomposition algorithm combined with the information entropy so as to remove speckle component,simultaneously edge details effectively preserved.Algorithm ? is SAR image edge detection algorithm based on the edge enhancement,which uses wavelet packet decomposition for image de-noising,and then use a combination of a number of scale to make low-resolution image got good reinforcing effect,and then remove speckle component and edge details effectively preserved simultaneously.It improved the weak edges can not easily be detected phenomenon.Algorithm ? is SAR image edge detection algorithm based on NSCT transformation.The algorithm used the correlation in scale which is from extending HMM of the background wavelet domain to NSCT transform domain to improve the correlation coefficient of NSCT.According to the corresponding scale between adjacent layers,it contains two direction coefficient of adjacent sub-bands of the fine scale layer and the coarse-scale which is fine-scale layer adjacent scales in the same sub-band.The difference between the directional coefficients,as a determination condition to distinguish the edge and noise signals in the scale.Finally,algorithm ? obtained better edge detection.Algorithms ? is a suppression speckle algorithm based on improved Frost filtering in the anisotropic diffusion.The algorithm is improved by using improved Frost filter coefficients to the estimated spread function of DPAD algorithm,lead to the degree of diffusion homogeneous region is increased and the degree of diffusion in the edge region is reduced.Finally,the speckle is sufficiently suppressed in homogeneous region,image smoothness and filtering effect is good at edge region by comparing each speckle suppression algorithm.This paper first describes the background and imaging principle of SAR images and proposed four SAR image edge detection algorithms in two aspects and proposes a SAR image speckle reduction algorithm.Finally,the paper summarizes the work and prospects.