A New Algorithm About Speckle Noise For Denoising And Enhancement

Due to its specific imaging mechanisms and detection of tissue inhomogeneity, the noiseof Medical ultrasound image named speckle noise is different from a general image noise. Itbelongs to the multiplicative noise, and a large proportion is in the high frequency part.Unlike general image denoising, medical image de-noising require not only to remove thenoise as much as possible to keep details of the image, but also require the effect of thelesions edges more stringent which need to divide in the next study. So image edgeenhancement is needed at the same time when it is denoised. In response to these needs, ahybrid model is built in this paper which suits to remove speckle noise. Firstly, analyzing thetype of ultrasound image noise, and then exploring a denosing algorithm which is suitable toremove the speckle noise; Secondly, a hybrid model which is combined anisotropic diffusionmodel containing edge detection function with the shock filter containing the variable isconstructed; Finally, using Matlab procedures completes the ultrasound image denoising andenhancement, confirming the algorithm feasibility. Research of the dissertation isconcentrated on the following aspects:1、The advantages and disadvantages of anisotropic diffusion model is analyzed in thepaper, in order to overcome the fact that anisotropic diffusion model relies on detecting thegradient values to control the diffusion process but the value of the gradient is influenced bynoise at relatively large degree，an edge indicator function which is low-sensitivity to noiseis introduced, which can achieve the purpose of identification of the pseudo-edge quickly,thereby it can denoise fully.2、Describing the impact of the shock filter model, on the basis of shock filter, thefunction containing the variable function with respect to the number of iterations for edgejudgment is introduced, which can timely control the size of the shock amplitude.3、 When combine the two models together, on the one hand, it can overcome thedrawbacks of the diffusion equation which can not sharpen the edge, on the other hand, it can make up drawbacks when only a shock filter processes image caused instability.4、In the experiment, the Lena image which have been joined speckle noise is selected asthe experimental image,through a series of comparative experiments to validate the algorithmsuperiority in removing speckle noise;then continue to use thyroid ultrasound image whichcontains speckle noise by itself as the experimental image to validate the algorithm inremoving speckle noise but also can achieve the effect of edge enhancement,which can meetthe special medical ultrasound image pre-processing requirements.Experiments show that the new algorithm is superior to traditional filtering algorithmand anisotropic diffusion algorithm, which not only can be a good Speckle removal whileprotecting the edges, and can achieve the effect of sharpening and edge enhancement.