摘要
提出了一种采用中位数的滑动窗口滤波方法。所设置的窗口宽度为2m+1个数据点,使宽度大于m个数据点的峰被视为信号而无失真地保留,使宽度小于或等于m个数据点的峰被视为噪声而完全滤除。中位数法较之常用的加权和法具有滤波效果好、信号畸变小等优点。此法能保证谱峰高度和谱峰位置这两个重要参数不发生变化,故非常适宜用于仪器化学分析信号的滤波。
A new high fidelity filtering algorithm(HFFA) using medians in moving window is proposed. In case the window width is set to include 2m +1 data points, peaks with widths greater than m points are reserved without distortion, while peaks with widths smaller than m points are completely removed. Compared with other conventional filtering methods, in particular the widely used ones based upon weighted sums, this algorithm has advantages of better filtering effect and less distortion. Ensuring peak height and peak position, which are two major parameters, to be least changed, this new algorithm is especially feasible for filtering physico-chemical spectrum from noises.
出处
《高等学校化学学报》
SCIE
EI
CAS
CSCD
北大核心
1995年第7期1020-1023,共4页
Chemical Journal of Chinese Universities
基金
国家自然科学基金
