Findlevels Algorithm

Share icon

Hello,
I use the function "Findlevels" to treat my signals and it works very well. I wondered if it was possible to have the algorithm of this function because I want to reproduce exactly my program (procedure) in another programming language, C # for example. Is it possible ?
I'm looking for a simple algorithm to do the same work as "findlevels" function?
Many thanks,

August 31, 2016 at 07:50 am - Permalink

Having the exact code is not as important as getting consistent results. I can imagine that a search on a phrase such as "find a value in an array using C++ code" would give you a good starting point. Also, as far as I recall, the code in Igor Pro is referenced often to the handbook Numerical Recipes in C.

--
J. J. Weimer
Chemistry / Chemical & Materials Engineering, UAH

August 31, 2016 at 07:56 am - Permalink

Thank you jwiemer.
I would like to have the same code and keep the same parameters used in the function Find Levels such as "Box" and "Level".
I could not find a simple way to do it :(

September 2, 2016 at 05:57 am - Permalink

I would ask first ... Why go outside of Igor Pro when it already does exactly what you need?

As to your ending question ... I suggest you should contact WaveMetrics directly. I imagine that releasing code or even commenting that they might release it is not something they will do on a public forum. Finally, I might suggest that at best, you will get references to literature and still have to develop your own code. In other words, I seriously doubt that you will be handed a "code package" that you can just duplicate and use directly.

Which brings me back to my starting question.

--
J. J. Weimer
Chemistry / Chemical & Materials Engineering, UAH

September 3, 2016 at 07:34 am - Permalink

I come back here to propose a solution, maybe it could help someone.

jjweimer wrote:
I would ask first ... Why go outside of Igor Pro when it already does exactly what you need?



@jjweimer : We use Igor Pro for R&D. The goal was to test the algorithm on a large set of data then to make it an embedded software.

Since my last post here I have no longer worked on the issue. But, today I suddenly had a flash ;-).

Here is my solution

#pragma rtGlobals=3     // Use modern global access method and strict wave access.
 
//-----------------------------------------------------------------------------------------
//                  LOOK FOR INDEX 
//-----------------------------------------------------------------------------------------
//      This function act as the Igor's function "FindLevels"
//-----------------------------------------------------------------------------------------
 
 
Function/WAVE LookForIndex(dataWave,level, boxSize) 
    Wave dataWave
    Variable level      // threshold to be crossed
    Variable boxSize    // box size for sliding average, should preferably be an odd number 
        
    boxSize = ceil((boxSize-1)/2)   // take the half size of the box
        
    Variable length = numpnts(dataWave) 
    Variable i, j
    Variable slidingAverage 
    Variable lookForRisingEdge = 1
    
    Make/O/N=10000 indexWave // we assume that we have less than 10000 crossing level. If not, change this value. 
    
    for(i = 0; i < length; i+=1)    
        
        slidingAverage =  mean(dataWave,i-boxSize,i+boxSize)    // comput the sliding average  
            
        if ( (slidingAverage > level) && (lookForRisingEdge == 1))  // look for rising edge
            
            indexWave[j] = i            
            lookForRisingEdge = 0           
            j=j+1
            
        elseif ( (slidingAverage) < level && (lookForRisingEdge == 0 )) // look for falling edge
                    
            indexWave[j] = i            
            lookForRisingEdge = 1           
            j=j+1               
        endif
            
    endfor
        
    Redimension /N=(j) indexWave    
    
    return indexWave // All "X" values at which the "level" was crossed
    
End


Here is an example (attached image) with the same parameters on a noisy data:

LookForIndex(myData,1, B=151)
FindLevels/B=151/Q/P/D=levelWave myData,1
LookForIndex.png (11.68 KB)

December 2, 2016 at 08:18 am - Permalink