Research On Speckle Suppression Algorithm By Wavelet Diffusion

Due to the imaging mechanism and equipment or the surrounding environmentalfactors，when Medical ultrasound image and SAR (Synthetic Aperture Radar, SyntheticAperture Radar) imaging, a kind of special noise are exist, we called it speckle noise.Ultrasound imaging is a common diagnostic medical imaging path, for its fast imaging andlow price Ultrasound imaging used as a clinical diagnostic tool widely. SAR is a surfaceimaging system, at present, developed rapidly in terms of geological exploration, militarytarget detection and so on. However, the presence of speckle affect the quality of these twoimages seriously, it is considered a contaminating factor that severely degrades image quality,moreover, have a bad impact on the further diagnostic or target detection.Speckle is modeled as spatial correlated multiplicative noise, increases the difficult todistinguish signal from noise. But the speckle noise can be rendered as the additive Gaussiannoise in the logarithmic domain. So speckle suppression in logarithmic domain becomes a hotresearch field.Wavelet diffusion is designed to utilize the favorable denoising properties of twofrequently used techniques: the sparsity and multiresolution properties of the wavelet, and theiterative edge enhancement feature of nonlinear diffusion. Recently years, wavelet diffusionwidely used in the speckle suppression. This paper bases on the existing wavelet literatures,proposes a new wavelet diffusion speckle suppression algorithm. Firstly, we studies thecharacteristics of the wavelet basis and its’ impact on the speckle suppression, than selects theappropriate wavelet basis; Secondly, adds the EM(expectation-maximization) algorithm toget the two new edge detection operators, and auto-segmentation the homogeneity region;Finally, bases on the homogeneous region, we estimates the new diffusion threshold to definea new PDE diffusion functions. Compared with other types of classic and efficient methods,the paper achieves better visual effects and quantitative analysis. Not only achieves bettereffective noise suppression, and preserves more edge features.