Thin out data

classic Classic list List threaded Threaded
3 messages Options
Reply | Threaded
Open this post in threaded view
|

Thin out data

flambert
Hi,

I've a little problem. With an industrial robot and a lasersensor I measure an edge. That works good, but if the robot stands still I get a higher measurement noise and a higher amount of "edge-points" in this sector. Here two images:
Image of the whole edge
Zoom of the effected area

In the image you can see that I try to insert an spline. But in the effected sector the spline is very bumpy. The data amount in this area is much higher then the data amount in the "normal" areas. Is there a good method to reduce the amount of data. I think if I solve this my spline would be smoother. I also tried to make the spline smoother by its own functions, but the result is that the spline becomes smoother in the effected area but will be nearly linear in the "normal" area.

Thanks and regards,
     franz
Reply | Threaded
Open this post in threaded view
|

Re: Thin out data

Daπid
You could use binned_statistics, taking the average (mean, median...) over certain window. This is, in my experience, more robust than using a spline to smooth data.

In any case, your data seems to have suspicious patterns. Maybe you could exploit them to get better measurements.

/David.


On 25 October 2013 08:40, flambert <[hidden email]> wrote:
Hi,

I've a little problem. With an industrial robot and a lasersensor I measure
an edge. That works good, but if the robot stands still I get a higher
measurement noise and a higher amount of "edge-points" in this sector. Here
two images:
<http://scipy-user.10969.n7.nabble.com/file/n18816/wholeEdge.png>
<http://scipy-user.10969.n7.nabble.com/file/n18816/partOfEdge.png>

In the image you can see that I try to insert an spline. But in the effected
sector the spline is very bumpy. The data amount in this area is much higher
then the data amount in the "normal" areas. Is there a good method to reduce
the amount of data. I think if I solve this my spline would be smoother. I
also tried to make the spline smoother by its own functions, but the result
is that the spline becomes smoother in the effected area but will be nearly
linear in the "normal" area.

Thanks and regards,
     franz



--
View this message in context: http://scipy-user.10969.n7.nabble.com/Thin-out-data-tp18816.html
Sent from the Scipy-User mailing list archive at Nabble.com.
_______________________________________________
SciPy-User mailing list
[hidden email]
http://mail.scipy.org/mailman/listinfo/scipy-user


_______________________________________________
SciPy-User mailing list
[hidden email]
http://mail.scipy.org/mailman/listinfo/scipy-user
Reply | Threaded
Open this post in threaded view
|

Re: Thin out data

Matt Newville
In reply to this post by flambert
Hi,

On Fri, Oct 25, 2013 at 1:40 AM, flambert <[hidden email]> wrote:

> Hi,
>
> I've a little problem. With an industrial robot and a lasersensor I measure
> an edge. That works good, but if the robot stands still I get a higher
> measurement noise and a higher amount of "edge-points" in this sector. Here
> two images:
> <http://scipy-user.10969.n7.nabble.com/file/n18816/wholeEdge.png>
> <http://scipy-user.10969.n7.nabble.com/file/n18816/partOfEdge.png>
>
> In the image you can see that I try to insert an spline. But in the effected
> sector the spline is very bumpy. The data amount in this area is much higher
> then the data amount in the "normal" areas. Is there a good method to reduce
> the amount of data. I think if I solve this my spline would be smoother. I
> also tried to make the spline smoother by its own functions, but the result
> is that the spline becomes smoother in the effected area but will be nearly
> linear in the "normal" area.
>
> Thanks and regards,
>      franz

If you haven't done so already, try a Savitsky-Golay filter to smooth
the data: http://wiki.scipy.org/Cookbook/SavitzkyGolay
If the data in the very spiky section is oversampled, you may want to
interpolate (the filtered data) onto a uniform grid.

--Matt
_______________________________________________
SciPy-User mailing list
[hidden email]
http://mail.scipy.org/mailman/listinfo/scipy-user