Hamish,
Getting closer (I was wondering how to accomplish the extraction of
nodes from a vector... thanks!). Unfortunately, this still leaves me
with the problem of a short line segment attached to the end of the
original line by a node rather than a vertex, which can put in a kink
that remains after smoothing with v.generalize.
I followed v.to.points/v.distance/v.patch with v.build.polylines and
v.generalize method=snakes, which works as long as the only desired
nodes are associated with one line and not two (i.e. they are terminal
nodes). However, I many have other nodes that connect two line
segments at a sharp corner such as a trail fork or switchback that I
don't want to lose (convert to a vertex).
Is there a way to only convert a two-line node to a vertex if one of
the lines is shorter than a threshold or the node is within a
threshold of a terminus or reference point? Alternatively, is it
possible to patch two vectors in a way that creates a polyline where
feasible?
I want to end up with the short line segment created by v.distance
patched to the original line by a vertex rather than a node, but
without losing nodes elsewhere in the original line.
-Dwight
On May 29, 2009, at 2:08 PM, Hamish wrote:
ok, try 'v.to.points -n in=lines out=line_nodes'
.... then 'v.distance from=gps to=line_nodes' + v.patch + v.clean ...
Hamish
Dwight wrote:
I am looking for a way to force the polylines
(from GPS
tracks) to terminate at the points (from averaged
waypoints). One possibility would be to snap each
terminal
node to the nearest point (within a threshold),
with the
point coordinates taking precedence.
Hi Dwight,
try 'v.distance output=' + v.patch + v.clean.
Hamish
Thanks, Hamish. That is very close to what I want, but I am
having some difficulty closing the gap.
In the simplest circumstance, the part of the line closest
to the point is the node on the end, so your suggested
procedure extends the line to the point. However, the
original node is still a node (rather than a vertex), so
there is a simple line segment attached to the end of a
polyline segment. I was hoping to find a way to end up with
a single polyline so that any kink generated at the end will
be straightened out using v.generalize
method=sliding_averaging.
Any ideas on how to accomplish this? Can two-line nodes be
converted to a vertex within a threshold distance of a
reference point or if one of the connected line segments is
shorter than a threshold?
Under other circumstances, the part of the line closest to
the point is somewhere along the line segment rather than at
the end, resulting in a "Y" configuration. I thought I could
use v.clean to break the original polyline where the
v.distance segment intersects, which would generate a dangle
that could then be removed with v.clean. However, I had
trouble with each of these steps. Here is what I did, using
a point near each terminus of a line.
v.distance -p from=terminus to=trail from_type=point
to_type=line output=trail_connections dmax=20 upload=dist
column=dist
v.patch input=trail,trail_connections output=trail_patched
v.clean input=trail_patched output=trail_break type=line
tool=break thresh=1
v.clean input=trail_break output=trail_rmdangle type=line
tool=rmdangle thresh=20
v.distance created lines that were about 6 and 13 ft
long that intersected the trail line at about 8 and 14 ft
away from each terminus, respectively (creating a "Y" on
each end).
v.clean tool=break clipped the trail at the second
intersection, but not the first (confirmed using v.digit).
What causes this difference in behavior?
v.clean tool=rmdangle removed both of the line segments
generated by v.distance, but did not remove what appears to
be a 14 ft dangle created by the v.clean tool=break step.
Why was the 14 ft dangle not removed (in v.digit it has a
green X on one end and a red X on the other, and it is
shorter than the threshold).
There are enough problems here that I am beginning to think
that I don't understand what should be happening. Can
someone straighten me out?
Thanks, -Dwight
_______________________________________________
grass-user mailing list
[email protected]
http://lists.osgeo.org/mailman/listinfo/grass-user
