低场核磁共振信号降噪方法研究进展

The Research Progress of De-Noising Methods in Low-Field NMR Signal

目的:低场核磁共振技术是近年来迅速发展起来的一种新技术,被广泛应用在食品安全、油气勘探、纺织化工、生物医药等各种领域。但由于低场核磁共振背景磁场强度低,信号微弱,信噪比较低。本文目的在于抑制低场核磁共振中的噪声,提高信噪比。方法:目前用于低场核磁共振的降噪方法还不是很多,本文主要介绍了三种降噪方法,双边滤波、基于偏微分方程的各向异性扩散方法和基于平稳小波变换的阈值滤波。结果:双边滤波既考虑像素点的空间相近度又顾虑到亮度相似度,达到了平滑滤波和边缘保持的效果;偏微分方程各向异性扩散方法在图像平滑区域增大扩散系数,在边缘区域减小扩散系数,对边缘保留起到了很好的效果;平稳小波变换具有平移不变性,去除了伪Gibbs现象,边沿对齐和改进的阈值函数都提高了降噪效果。结论:本文的三种降噪算法都是在以前学者提出的方法基础上作出了改进之后用于低场核磁共振系统,并且都取得了很好的效果。本文又综述了评价降噪效果的客观指标PSNR,RMSE和SSIM。

Objective： Low-field nuclear magnetic resonance （NMR） technology is a kind of new technology rapidly developed in the recent years, and widely used in the food safety, oil and gas exploration, textile and chemical industry, biological medicine and other fields. But due to low field NMR background magnetic field intensity is low,the weak signal ,the poor sig- nal-to-noise ratio （SNR）.The purpose of this paper is to suppress noise in low field NMR, improve signal-to-noise ratio.meth- ods：Currently, methods of noise reduction in low field nuclear magnetic resonance （NMR） are few, this paper mainly intro- duces three methods of noise reduction, bilateral filter, anisotropic diffusion method based on partial differential equation and threshold filter based on stationary wavelet transform.results： bilateral filter considers both geometric closeness and radiometric similarity of neighboring pixels,achieving to effect of smoothing and keeping edge;anisotropic diffusion based on Partial differ- ential equation in the image smooth areas increases diffusion coefficient and decrease the diffusion coefficient on the edge area, playing a very good role on the edge retention;Stationary wavelet transform has the translation invariance,so it wipes out the pseudo Gibbs phenomenon, edge alignment and improved threshold function both contributing to improve the noise reduction effect.Conclusions： in this paper, three kinds of noise reduction algorithms are based on the methods proposed by scholars be- fore and would be used in low field nuclear magnetic resonance （NMR） system after improvement, then have achieved good re- suits.This paper also reviews objective indicators of evaluating noise reduction effects, PSNR, RMSE and SSIM.