Re: point projection on polyline

[Juan Pablo Carbajal]

On Fri, Jan 22, 2016 at 7:51 AM, rcharan51 <address@hidden> wrote:

Hi, I am back!!. This is my pet project at office on which I spend some time
whenever I am free. The basic idea is to generate a 3D settlement(on Z axis)
profile because of tunneling. Settlement is a function of chainage (ie
projection of a particular point on tunnel) and  orthogonal distance of that
point from tunnel's   centerline (our polyline)  in transverse direction (ie
on XY plane). Here's a snap of polyline
<http://octave.1599824.n4.nabble.com/file/n4674483/tun_poly.jpeg>

My idea is to generate a grid of 10 meter spacing (can be fine tuned to 1
meter if a fast method is available), and find the proj and dist of each
point by arrayfun, but even that is not fast enough for some 5000(=15min)
points on a polyline (of 300 vertices) while calculating projection whereas
there is no problem with distancePointPolyline which is quiet faster. Thank
you for your idea to cull or split the polyline into small sections which
decreases the #of edges that need to be compared.

3D profile I want to produce will be something like
<http://octave.1599824.n4.nabble.com/file/n4674483/MMRL_model.png>

Can u give me some info about how to cull the polyline based on distance.
TIA.





--
View this message in context: http://octave.1599824.n4.nabble.com/point-projection-on-polyline-tp4673298p4674483.html
Sent from the Octave - General mailing list archive at Nabble.com.

_______________________________________________
Help-octave mailing list
address@hidden
https://lists.gnu.org/mailman/listinfo/help-octave

In short words. Do you want to find the perpendicular (to which curve?) distance between the two curves? 

You could try simplifying the polylines, reducing the sets of points to compare, and then iterate until you reach your desired resolution. Whether this si faster or not, it depends on how much you gain when you simplify your polyline. Otherwise full comparison is all you have. Check the algorithm is pretty simple, you can profile it and implement the bottle necks in C++.