How to avoid losing small width peak data in smoothing functions?

3 次查看（过去 30 天）

Marmi Afrin 2012-10-9

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/50309-how-to-avoid-losing-small-width-peak-data-in-smoothing-functions

I don't have the curve fitting toolbox and I am trying to smooth out a very noisy data (not a time series). What are some of the best/well-tested smoothing functions out here? It'd be nice if I can get hold of the smooth.m function from the curve fitting box but I don't see how I can go on about obtaining that alone.

I tried to write my own using triangular smoothing but I am working in micron scales and I have a few legitimate peaks (not noise) of really high amplitude (I am measuring the surface of a substance and this is a scratch on it) and smoothing functions take that out. How would one go about avoiding that? Is there way a to ask a smoothing function to bypass that or even avoid smoothing if the slope of the data exceeds a certain value?

Thanks!