基于改进非局部均值的CPMG信号去噪算法

采用低场核磁共振技术进行检测时，接收到的回波信号微弱且信噪比低，真实的信号容易淹没在背景噪声中，严重影响到后续的反演等操作的准确性．针对这一问题，提出利用非局部均值滤波算法对CPMG(Carr Purcell Meiboom Gill)回波信号进行降噪的方法．首先，对算法中至关重要的参数选择的方法进行分析，提出了利用Stein无偏风险估计的自适应参数选取方法；然后，根据回波信号的特性对算法进行改进，即利用信号点数据方差的不同，自适应地求取各点进行非局部均值滤波时的相似窗宽度；最后，求取利用最优参数进行降噪后的CPMG回波信号．对仿真数据和真实数据的反演结果对比分析表明，该改进的非局部均值滤波算法能够取得更好的滤波效果，能够获得较优的反演谱．

An Improved Nonlocal Means Method for CPMG Signal Denoising

Echo signals obtained on low-field NMR spectrometer are often weak, and have low signal-to-noise ratio, such that the signals are easily to be buried in background noise. Aiming at solving this problem, an improved non-local means (NLM) algorithm for filtering CPMG echo signals was proposed. First, based on analysis of parameter selection methods, an adaptive damping parameter selection method with Stein unbiased risk estimation was selected for the NLM algorithm. According to the characteristics of the echo signal, an improved method using different signal point data variance was employed to find the width of the neighborhood window for the NLM algorithm. Lastly, the NLM algorithm was implemented with optimized parameters. The results on simulated and experimental data sets were reported. Compared with the existing NLM algorithm, the improved NLM algorithm was shown to be able to produce better results concerning both the filtered signals and inversed spectra.