Ultrasound image denoising method based on optimal Bayesian non-local mean
In order to overcome the speckle noise that may be generated in ultrasound imaging,an optimized Bayesian non-local mean algorithm is proposed.This paper discusses the improvement of the optimized al-gorithm from the following two aspects:(1)Aiming at the problem of high computational complexity of the non-local means(NLM)algorithm,the pixel pre-selection algorithm is applied optimize the denoising time of the algorithm;(2)Aiming at the problem that the NLM algorithm is only applicable to additive Gaussian noise,the Gamma distribution noise model is used to derive a non-local mean algorithm suitable for ultra-sonic speckle noise through the Bayesian formula,and the Pearson distance is derived to calculate the distance between image blocks;and through multiple experiments,the ratio relationship between the noise standard deviation and the filtering parameters is obtained,and the parameters are self-adapted;finally,the algorithm is realized through the Matlab platform.In order to evaluate the denoising ability of the optimal algorithm,experiments were carried out on simulated ultrasound images and real ultrasound images.The results show that the optimal algorithm has a better denoising effect and a stronger ability to preserve image edges and details;and in terms of operating efficiency,compared with the original non-local mean algorithm,it has a greater improvement.
Ultrasound imageImage denoisingNon-local mean algorithmPerson distance
万方数据
维普
