finding the intersection for a trace to a threshold

Hi I am trying to find the points of intersection of a trace compared to a threshold. Let me explain, I have the trace of the movement of the foot in the z axis (raising the toe and putting it back to the floor in a step) and I want to find the points of intersection of a determined threshold (which is 5 mm above the minimum point of the swing of the toe). I have tried codes that I have found around but I need something according to my data

    zstp=Rtoez(mRTO(i):mRHS(i));
    t1M=find(islocalmax(nzstp,'MaxNumExtrema',1));
    tmin=find(islocalmin(nzstp(t1M:end)))+t1M-1;
    tclear=zstp(tmin);ntclear=nzstp(tmin);
    traise=nzstp(t1M);
    
    %% Determining the initial threshold
    
    cutoff=round(ntclear+5)*ones(size(nzstp,1),1);
    if cutoff > (ntclear+5)
        cutoff=cutoff-1;
    end

So now I need the intersects of nzstp to cutoff, so I can calculate the area under the threshold but above the curve

[Marco Avalos]

[Marco Avalos]

I only need this area no the area above the threshold

Answer 1

[Daniel M]

You can get the closest data points on either side of the threshold very easily.

    location = threshold >= data;
% location is 1 below or on threshold, 0 above it
    crossover = diff(location);
% crossover is -1 when it goes from below to above, 0 when it stays the same, and +1 when it goes from above to below.

Now use find() to look for instances of -1 and +1 within the variable crossover. (Note that the length of crossover will be one fewer than the length of location). If all you need is the closest index position, this is good enough. If you need a more precise answer, do a linear interpolation between the two closest points above and below at each crossing. How you choose to calculate area under the curve at that point is up to you.

Alternatively, you could do the following (this is probably even simpler).

newdata = min(data,threshold);
% then integrate to get the area, possibly with:
myarea = trapz(newdata);

But again, you haven't indicated anything about precision, so it's hard to give a very precise answer.

[Marco Avalos]

Thank, you this helps me a lot. Interpolating more points gave me the exactitude needed.

Answer 2

[Turlough Hughes]

I'm not sure which variable you use to represent your x data, but lets call it x, where x is the same size as nzstp and cutoff.

You can find your intersection points with this handy function polyxpoly as follows:

[xi, yi]=polyxpoly(x,nzstp, [x(1) x(end)],[cutoff(1) cutoff(end)]);
hold on; plot(xi,yi,'^k')

The inputs are your curve data and the start and end points for your threshold line which I have written as

x,nzstp    % curve data
[x(1) x(end)],[cutoff(1) cutoff(end)] %start and end point of threshold line

Note you may need to download the mapping toolbox if you dont already have it installed in order to you polyxpoly.

[darova]

I love polyxpoly

[darova]

Intersections do the same also