Remove peaks less than a threshold

Hello guys, I'm dealing with analysis of a spectra and I found that there are many unwanted small peaks in it(see the image). I want to remove the peaks that has the intensity (abscissa) less than say 15.0. Is there any function in matlab that does this. Can someone get me a script for that.

Im attaching my file here

[Adam]

What do you mean by 'remove them'? Set only the peak values to 0? Doing only this would obviously result in new small peaks in most places where the newly zeroed point would cause small peaks either side of it.

You can use various settings in the findpeaks function if you have the signal processing toolbox in order to not include the small peaks, but by the sounds of it you want to get rid of them from your spectrum rather than just find the peaks that do matter.

Answer 1

[Image Analyst]

To set values smaller than some threshold to zero, do this:

threshold = 15; % Whatever....
elementsToSetToZero = y < threshold;
y(elementsToSetToZero ) = 0;

I think Adam's comments about small side peaks getting created was if you zeroed out tall spikes, not short ones. Or else he was zeroing out just the single element at the very peak instead of everything (multiple elements) less than some threshold value. My code above will not create any additional peaks.

[Adam]

Yes, that works well. I got too fixated on the question of just removing peaks (i.e. the single identified peak samples) rather than just zeroing everything below the given value, whether peak or not.

Although if it is desired to retain the slope up to the higher peaks the effect of this may be undesirable as it will also trim them down to 0 at their wide base.

[Image Analyst]

Correct. We need to know the intent. Does he just need to know where the center of the tall peaks are? Or does he need to know the total area? If the total area is needed, he might have to start with the very peak and then "fall down" each side of it until the values turn around or at least level off, and do that with the original data, not the zeroed out data.

[SAHEER VC]

Thank you for the answer. Adam's code works but as pointed our by Image Analyst it zeros everything below the given number. Removing a smaller peak should not affect the larger peaks. See the two images below for example. From the first image I removed two peaks whose intensity is below 100 (removed the x and y data representing the peak). I did it manually, but it a tedious process as there are hundreds of such smaller peaks.

[Image Analyst]

What are you going to do? Why is it so important that the really low tails on the larger peaks be kept intact?