分析化学窄带锐峰信号滤波新技术

A New Technique on De-noising for Narrowband Signals in Analytical Chemistry

窄带锐峰信号是指峰形狭窄、峰周围数据点极少的信号 ,采用常规滤波方法难以处理。本文提出了适合于这类信号的滤波新技术———MexicanHat小波最小二乘法 (MWLS)。它将原信号中的尖锐峰分离出来 ,以MexicanHat小波构造拟合函数 ,根据最小二乘原理对各数据段进行拟合 ,再重构信号 ,以去除噪声。大量处理实践证明 :该方法除噪能力强 ,滤波结果可靠 ,尖锐信号峰中的有用信息不会损失。即使对信噪比为 1.5的高噪声锐峰信号仍能获得满意结果。

In analytical chemistry, the peaks of narrowband signals are so sharp that those signals can not be denoised by the usual methods. A de-noising technique named Mexican Hat wavelet least square (MWLS) which is effective for narrowband signals is presented. MWLS extracts narrow peaks from original signals and fits data segments with a fitting function constructed by Mexican Hat wavelet. It has been proven to be powerful and reliable for narrowband signals, the useful information contained in peaks will not lose. Satisfactory results can be get even when the signals with very high noise of S/N = 1.5.