Interp1 introduces spikes

Hello,

I would like to interpolate a set of data I have. It represents the indentation vs force of AFM measurement. My vector indentation is not sorted and might have some non unique values.

Here is the code I am using:

          [I,ia,~] = unique(Ii,'stable'); % Ii is the initial indentation vector
          F = Ff(ia); % Ff is the force I am measuring

          Force = interp1(I,F,I(1):0.1e-9:I(end),'pchip');
          IndentationCell{i} = I(1):0.1e-9:I(end);
          ForceCell{i} = detrend(Force);

However I am getting some spikes as shown in the picture (left side is plot(I,F), right side is plot(IndentationCell,ForceCell) - after interpolation)

Do you have any ideas to overcome this issue ? I already tried changing interpolation method.

Michael

[Adam]

You would probably have to do a piece-wise interpolation. The problem is that your curve goes back on itself rather than progressing monotonically along the x axis. It is in these places that it introduces the spikes.

If you break your curve up into segments each of which is monotonic along x and interpolate each of those and stitch them back together it should work better I think.

How to break up your curve in such a way is something you would need to think about though. It should be easy enough to locate the points at which the x-values change direction I would think, but is beyond the time I have available to suggest in more detail.

Answer 1

[Muhammad Usman Saleem]

here is the solution for this question

http://www.mathworks.com/matlabcentral/answers/265845-interpolations-nearest-neighbor-and-bilinear-interpolation-from-multiple-text-files

[Michael Lherbette]

I am sorry but I don't see how this is the solution of my problem

[Muhammad Usman Saleem]

interpolate1 interpolate data which is in 1 dimension. Some times its start to interpolate values based on next values. Please consider your data set stickily in 1 dimension. @Adam anwser is also right