Research On Speckle Noise Removal Model Based On Energy Function Minimization

With the development of medical imaging technology,medical ultrasound images have received more and more attention in the medical field.Due to image transmission,compression process,and imaging equipment,medical ultrasound images usually have speckle noise,which affects the doctor's discrimination of medical ultrasound images and is not conducive to treating patients.Therefore,eliminating the noise in medical ultrasound images becomes a medical image a research hotspot for the researchers to deal with.Many scholars have proposed some effective models and methods for the removal of speckle noise,such as methods based on filters,wavelet transform,locally adaptive statistics,and partial differential equations.Although these methods have achieved good results in removing speckle noise,there are still some problems that have not been completely solved,such as the step projection effect of the denoised result and insufficient edge protection.For these problems in speckle noise removal,the following researches have been carried out in this academic paper:1.To deal with the problem of suppressing the step projection effect of images,a new model of adaptive diffuse speckle noise removal is proposed.The new model is based on the classical total variation(TV)model,and combines the degradation model of the speckle noise of the traditional Chinese medicine images proposed by Jin Zheng Ming(JIN'S model for short).Starting from local coordinates,a new regular term is selected according to the diffusion form of energy in different regions,and a new model of adaptive diffusion is constructed in combination with the degradation model of speckle noise.The regularization term of the model smooth the noise through the gradient of the image to achieve the effect of denoising,while the purpose of the data fidelity term is to ensure that the result cannot be too smooth and to make the image close to the original image.At the same time,there is a parameter betweenthe regular term and the fidelity term to balance their weight ratio.In addition,a parameter is added to the model to control the speed of energy diffusion,and the optimal parameter is selected through experiments.The finite difference method and the multiplier alternating direction method(ADMM)are used to solve the model numerically.Finally,the experimental results are compared with the traditional model.The numerical experiments of the two solutions show that the proposed new adaptive diffusion model has a better effect than the traditional model in removing speckle noise,and can significantly reduce the step projection of the image effect.Using the multiplier alternating direction method to solve the model is slightly better than the finite difference method.2.In terms of the problem of insufficient edge protection of traditional model images,a new model of speckle noise removal that diffuses forward and backward is proposed.The new model starts from the regularization term and combines the two regularization terms of forward diffusion and backward diffusion,so that the energy function of the new model can both diffuse forward and backward.In the new model,the regular term that diffuses forward can achieve the effect of removing noise,while the regular term that diffuses backward protects the edges of the image.The ADMM algorithm is used to solve the new model numerically.Numerical experiments show that the proposed new model of forward and backward diffusion can reduce the step projection effect of the image,and also have a better denoising effect than the traditional model,and can protect the edge of the image.