wavelet detrend

[battler]

Hi there,

I need to know more about the Advanced Signal Processing VIs, in particular those relating to Wavelets.  I'm interested, well I'm more than interested, I need to know, how the Detrend and the Denoise VIs work and more about the wavelet families available in LabVIEW.  How do I match the right wavelet family to my data?  There seems to be very little documentation on these VIs, I need to know heaps more.

Thanks very much

[Kalin_T]

Dear battler,

I am not sure how much you have read on Wavelets so far, so here are several useful links:
Wavelet Analysis Tools User Manual
https://www.ni.com/docs/en-US/bundle/labview-advanced-signal-processing-toolkit-api-ref/page/lvasptc...
NI LabVIEW Advanced Signal Processing Toolkit (click on Resources, and then on Manuals)
https://www.ni.com/en/shop/labview/select-edition.html

There are many tutorials and application notes available on our website. Simply type "wavelet" in the search engine of ni.com 

Also if you right-click on those VIs after you place them on the block diagram, you should be able to select "Help" and see the full description of the VIs as well as related examples.

Hope that helps to get you started!

[battler]

Hi Kalin,

Thanks for your response.  I have read the Advanced Signal Processing Toolkit and the Wavelet Analysis Tools user manuals.  I must point out that they're very much lacking in many areas.  I have also read all of the little material there is on wavelets on your website.  NI really must address this.

I'm using the wavelet detrend VI.  There are various wavelet families to choose from.  It states in the documentation that one must choose a wavelet family which matches the data, it does not say how to do it.  I understand that the detrend and perhaps the denoise VIs split the wavelet representation into two parts: the detail and the approximation.  Is this correct?  How can I learn more?

I must understand how these VIs work, though I do not have time to read an entire book on the subject.  I need to be able to match the mother wavelet to the data to receive optimum results.  I must be able to justify my choices and explain all this to my colleagues.

I very much appreciate your help.

[Kalin_T]

Dear buttler,

What version of the Advanced Signal Processing Toolkit and LabVIEW are you using?

[Kalin_T]

Dear battler,

It is not easy to explain how to use the Detrend or De-noise VIs. 

Basically it is a trial-and-error procedure.  You should have your own metrics to evaluate the results of detrending or denoising.  And the metrics depend on your applications.  If you are clear about your metrics, you can try the level of wavelet transform until you think your metrics are met.  You can try the wavelet family as well. However, in many real applications, the wavelet family does not tell much difference.  If you want to see whether the selected wavelet matches the signal, you can use the WA Mother Wavelet and Scaling Function VI to plot the shape of the mother wavelet.  

Regarding the implementation details, you can refer to the details in the Online Help of the VIs.

In summary, the VIs just implement the algorithms. The selection of parameters usually depends on the applications and not on the algorithms implemented.  If you are asking for the details of the algorithms, we can provide them.  If you are asking how to select the parameters, we might need to look at the signals you have.  If you can provide some signals and tell us some background, we would love to help.

[battler]

Dear Kalin,

Thank you for your response.  I apologise for the delay in mine.

We're working in the area of DNA microarrays.  We're trying to establish some potential applications of wavelets in the field.  We've had good results with the detrend and the denoise VIs.  We do wish to know all about the algorithms so it would be good if you could provide that information.  I must on a theoretical basis determine which wavelet family best suits the DNA microarray data.  Furthermore, the data is highly variant so it may be that for a certain a wavelet family may be unsuitable whereas on another it may well be optimal.  As part of the algorithm we wish to establish a test for this.  We must understand everything about your algorithms then.  It is not enough to trial and error since it is difficult to gauge the results.  Everything must be done on a theoretical basis.  The procedures have to be reproduceable and they must be explainable.

Again we would appreciate some further information from you.

Thank you,

J.Patrick.

[Jaspal]

Hi Battler,

I came across your post when trying to gather information on wavelet detrend of the Advanced signal processing toolkit. ALthough its been a while since u posted, it looks like you have a very good understanding on the topic and could possibly help me.

I have a r-r signal (time difference between consecutive heart beats)

I am estimating the Power spectral density of rr but boefore this I have to detrend the signal to remove baseline wandering.

I dont have the ASPT. I was wondering if this can be done using standard labview functions.