Hello
I have a scatter points data that goes in Y axis direction for while than changes to X axis direction.
I want to perform Interpolation on it mainly trying to model its "shape". The linear interpolation does not work the way I want for the vertical form points and gives an average point of the data not following its "shape". But for the horizontal part linear interpolation works fine. I have tried inverting axis and got forms using interpolation for some parts only not the whole data.
I am searching for a method to extract the shape of the data as a model or as a set of points.
Cheers
Dear Salah,
welcome,
I do not see no scatter of y versus x. Scattering means for the same X there will be multiple points centered around an average value. Here you need only to interpolate your points to find the values between the original points. There curve fitting tool books in matlab that you can use them.
Best wishes
Your question is somewhat cryptic. Anyway, If the points are locations of a robot, then the most natural description would be: (X,Y)(t) = (f(t), g(t)) with t=time, and f and g some functions (probably defined on many connected time intervals).
Hello Abdelhalim Zekry
Thank you for the reply. I know it is little bit off. I could not find a better word but I think a 2D point cloud is more appropriate.
Thank you
Marek Wojciech Gutowski
Thank you for the reply. It is correct, the 2D data is a sum of a robot paths. I think the method you have suggested works if we have the time.
In case we don't have this parameter. We have only traces. how to deal with such a problem?! Human brain can extract easily the centerlane bases on this data.
Do you think using the boundaries of the 2D point cloud then using Voronoi algorithm to get the center line would be correct ?! the result would be prone to noisy paths, I think.
Thank you
Salah Eddine Ghamri,
You most likely don't need precise timing, it is enough to know which location was visited first, which one was the next and so on. The time ordering is important. Voronoi-related algorithm may be indeed susceptible to unpredictable failures.
Hello Marek Wojciech Gutowski
You have been a great help for me thank you. We need time and if we don't have time we need create one or another universal parameter. So I have created a common parameter "s" which is the distance from start till the end, common to all paths. Then using linear interpolation and averaging i have got:
(X,Y)(s) = (f(s), g(s)) with s=a distance
Cheers
Hi, Salah
I am very glad to be of some help, my great pleasure. By the way: you have just rediscovered the way mathematicians (at least some of them) define curves.
We have open-source tools to generate shape models as point clouds, for 3D shapes as well as contours/polylines. http://www.sci.utah.edu/software/shapeworks.html
You may contact directly there to get access to the code base which works with 2D/N-D contours.
- Badges
- Science topic