Find Peaks, how to ignore peaks that are not so sharp?

Question

Luciano Junior 2017-3-15

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/330068-find-peaks-how-to-ignore-peaks-that-are-not-so-sharp

评论： Adam 2017-3-16

I need to find peaks in a first derivative curve. Peaks that are no so sharp doest not fit to the purpose of my work. I have tried "findpeaks", "peakseek" and "peakdet" with no success. Setting a trheshold also does not help, once the undesireble peaks can be above or bellow zero, or higher than other important peaks. In the example bellow the red circle shows in red, the peak I want to ignore, and the other circles in green are ok. Is there a way to select a peak based on how sharp it is?

Thank you

0 个评论
显示 -2更早的评论隐藏 -2更早的评论

请先登录，再进行评论。

请先登录，再回答此问题。

Answer 1

Star Strider 2017-3-15

1
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/330068-find-peaks-how-to-ignore-peaks-that-are-not-so-sharp#answer_258900

You did not post the findpeaks code you used, so I am not certain what options you have used.

Use the findpeaks function with appropriate name-value pair arguments. The 'MinPeakProminence' and 'MinPeakHeight' options would be my first choice. Experiment to get the result you want.

3 个评论
显示 1更早的评论隐藏 1更早的评论

Luciano Junior 2017-3-16

'MinPeakProminence' seems to be more apropriate, but it gives me an error: "??? Error using ==> uddpvparse at 119 Invalid Parameter/Value pairs.

Error in ==> findpeaks at 46 hopts = uddpvparse('dspopts.findpeaks',varargin{:});"

is it because im using an old Matlab version??

Thank you for the help

Adam 2017-3-16

在 MATLAB Online 中打开

doc findpeaks

should tell you if it is an option in whichever version you have. I don't know which version it was added in, it is hard to tell without trawling through old docs or release notes.

请先登录，再进行评论。

Answer 2

Jan 2017-3-15

0
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/330068-find-peaks-how-to-ignore-peaks-that-are-not-so-sharp#answer_258898

Determine the peaks and the gradient() at the peaks. Then exclude the peaks with the low gardient.

It might be useful not to use the gradient between the direct neighbors, because this might be dominated by noise. Perhaps a Savitzky-Golay filter for the surrounding 100 points helps.