data smoothing

[pande]

Hi,

I am poltting XY graph using some data. I want to smooth the data to have good Xy graph. Can anybody help me in thig regards. I am using Labview 8.0

[Karissap]

Try wiring your xy graph to the curve fitting express vi. With this vi you can choose from a range of interpolation algorithms.

It's in the Express>> Analysis Menu

[unicorn]

Smoothing is somehow different from curve fitting. Smoothing tries to remove noise and fast changes on the data to show the slow changes more clearly. Curve fitting tries to get as close as possible to the data with some mathematical description.

To make your data look more smooth look out in the net for "data smoothing". You will find descriptions of several algorithms. I haven't found a running mean in LV yet, but there are several filters which can be applied to your data. See palette Signal Processing > Windows

[JustinThomas]

Filtering works for me. Its the easiest solution

[Karissap]

Smoothing is somehow different from curve fitting. Smoothing tries to remove noise and fast changes on the data to show the slow changes more clearly. Curve fitting tries to get as close as possible to the data with some mathematical description.

To make your data look more smooth look out in the net for "data smoothing". You will find descriptions of several algorithms. I haven't found a running mean in LV yet, but there are several filters which can be applied to your data. See palette Signal Processing > Windows

I have used smoothing algorithms on data in LabVIEW before but these filters are based on IIR or FIR filters and as such are only applicable to time based signal types, as it is a X-Y graph that needs to be "smoothed" I think the curve fitting algorithms are more applicable. The curve fitting algorithms do work well to keep the shape of the data while removing a lot of the noise.

[lvoyster]

I'd look into Savitsky-Golay filters if I were you. LabVIEW has a very nice collection of subVIs.

[unicorn]

I have used smoothing algorithms on data in LabVIEW before but these filters are based on IIR or FIR filters and as such are only applicable to time based signal types, as it is a X-Y graph that needs to be "smoothed" I think the curve fitting algorithms are more applicable. The curve fitting algorithms do work well to keep the shape of the data while removing a lot of the noise.

Curve fitting needs often the knowledge of a mathematical description of the observed/measured values or at least on parts of the data set. Sometimes you can guess the behavior, e. g. if something decays (but do you see whether there are one or two decay rates?). Smoothing on the other hand just suppresses fast changing components by averaging consecutive data points using different weights.