Research On Adaptive Denoising Method For Ultrasound Image

Ultrasound medical imaging is a common method in the diagnosis process,but the presence of noise in the image affects the diagnosis process.Noise removal is an important step in image preprocessing.Effective denoising algorithm can restore the image better,facilitate doctors to diagnose the disease more accurately,and other image processing operations,such as edge detection and image segmentation.The filtering algorithm can blur the useful information such as edge details in the image,so it needs a more accurate and effective algorithm.This paper introduces the principle of ultrasonic imaging,the causes of speckle noise formation,the mathematical model and statistical characteristics of noise,summarizes the classical image denoising algorithms,and proposes three improved ultrasonic image denoising algorithms for the shortcomings of the algorithm.Based on the IMRG denoising algorithm,the local windows are divided into two categories by using the IMRG super-pixel generation algorithm,and new local windows are obtained,which are applied in the edge part of the image,and the filtering results are clear and distinct.Based on the denoising method of ratio distance,aiming at the multiplicative model of speckle noise and Rayleigh distribution,a new similarity distance between pixels was obtained,which was used to stretch the central pixel in eight directions to obtain an accurate homogenous region window,which could be adaptive to expand the window and filter along the edge direction.The above two methods are the improvement of the local filtering algorithm.For the non-local filtering algorithm,an improved LPG-PCA algorithm is proposed.The three-dimensional image blocks of additive and multiplicative noise are constructed respectively.The PCA transform is used to filter the additive and multiplicative noise separately.The filtering results of the additive and multiplicative noise are taken into account as the final results,which can effectively remove the noise in the ultrasound image.Through the experiments of simulated ultrasound image and clinical ultrasound image,the three improved algorithms proposed in this paper are all superior to other classical algorithms in terms of each index.Compared with other algorithms,the three improved algorithms can effectively remove the noise in ultrasonic image and retain the useful information of edge details.