Topic: Challenge : Contour to Elevation Points

[mailmaverick]

Dear All,

We have seen many threads / posts such as Triangulation and others where we try to make contours from Points or Elevations Text.

Can anyone do reverse of the same i.e. we have got contours and we wish to get corresponding points such that when we again draw the contours from these points, they exactly match the original contours.

Now for the challenge :- Challenge is to achieve it with minimum number of elevation points.

I am attaching a sample Contours File. The contours have been drawn using Civil 3D. Let us see who achieves minimum number of elevation points.

[ribarm]

I don't understand... Any 3 points obtained from such contours define its plane and elevation, so if you have contour, you just have to choose any 3 to satisfy such challenge - you can do this with (vlax-curve-getpointatparam) or (vlax-curve-getpointatdist)...

[mailmaverick]

Dear Marko

I think you have misunderstood the problem.

Kindly see the attached AutoCAD file where I have given contour polylines.

These have been generated from a set of points (having elevations) using Civil 3D.

Now using these contours, how can we generate the original set of points ?

I am attaching my original set of points too in this post.

[tombu]

Not possible! You'd be better off using those contours for the surface than points from them.  The quality of the surface data goes down with every conversion.

[mailmaverick]

I know that it is better to use contours for surface than points.

But this problem is more of a challenge than actual use.

[ymg]

mailmawerick,

What you are after is simply not possible.

The closest you can get is to use all contours as breaklines
and generate a TIN from this.

Your original points are loss, but you have a more or less
equivalent surface.

Maybe knowing the algo used to generate the contours,
we could reverse it some.  But that would be a very
intensive task.

ymg

[mailmaverick]

Dear ymg

I agree to what you are saying. But I dont want the TIN surface.

I want the original points.

And, if not original points, minimum number of points which can generate same contours.

[ymg]

mailmaverick,

Are your contours smoothed ?

If not finding sets of points that are parrallels
would certainly hint that that they were in
the same triangle.

If you use the contour segments instead of points
the problem is easier.

Say you are on contour x , look to contour x + interval
and find the parallel segment.

The 4 endpoints of those two segments are lying on
two edges of the triangle that generated it.

Then keep going and do intersections to get
the original points.

ymg

[mailmaverick]

Dear ymg

As I told you I am not looking into any particular problem. It is just academics / learning / fun.

Regarding smoothening of contours, let us start by assuming that they are not smoothened.

[ymg]

maimaverick,

Post the contour, instead of the point
and see the edit in above post.

Marko's point is also valid, we can do
a bunch of plane intersections we will generates
the edges of the original triangulation.

ymg

[mailmaverick]

Dear ymg

Contours have been posted in my first post i.e. Post No 1.

[ymg]

here what I mean:

[mailmaverick]

mailmaverick,

Are your contours smoothed ?

If not finding sets of points that are parrallels
would certainly hint that that they were in
the same triangle.

If you use the contour segments instead of points
the problem is easier.

Say you are on contour x , look to contour x + interval
and find the parallel segment.

The 4 endpoints of those two segments are lying on
two edges of the triangle that generated it.

Then keep going and do intersections to get
the original points.

ymg

here what I mean:

Dear ymg,

I have studied your posts in detail.

But I dont know how to do it in LISP.

Whenever you have time, please do the same.

Thanks in advance..

[ymg]

mailmaverick,

I have other things to do.

My point is that it is feasible,
but the Devil is in the details...

ymg

[ymg]

mailmaverick,

For a low-tech solution just draw lines that
seems to be side of a triangle.

Then fillet with zero radius.  With patience
you can reconstruct the whole triangulation.

See a partial reconstruction: