search

fxbit / aforge

Automatically exported from code.google.com/p/aforge
Other
0 stars 0 forks source link

LineSegment.DistanceToPoint and GetIntersectionWith functions. #175

Closed GoogleCodeExporter closed 9 years ago

GoogleCodeExporter commented 9 years ago 
After some work (and more thought than I expected concerning the intersection 
of two segments) I finally have some more code.

I think I did better than last time at making sure the documentation matched 
the code.

There is a behavior difference between Line.GetIntersectionWith(Line) and the 
three forms that take at least one LineSegment; I'm not sure what (if anything) 
to do with it: Requesting the intersection of two parallel lines seems like 
exceptional behavior to me, and, as a result, this throws an exception. 
However, requesting the intersection of two non-intersecting segments does not 
seem like exceptional behavior, so an exception is not thrown in that case, 
even if the segments are collinear.
Line/Segment intersections follow Segment/Segment behavior: non-intersecting is 
not exceptional.
In all cases, if the two share multiple points, an exception is thrown.

Original issue reported on code.google.com by dales...@gmail.com on 16 Nov 2010 at 3:58

Attachments:

GoogleCodeExporter commented 9 years ago 
Merged. Committed in revision 1346. Will be released in version 2.1.5.

Found a bug in LineSegment.GetIntersectionWith( LineSegment other ) ...
Here is simple test to find it:
[Row( 0, 0, 6, 0, 5, 1, 5, 5, 5, 0, IntersectionType.SegmentA )]

For this two segments there is not intesection at all. However, here is what 
you do. You call result = GetIntersectionWith( other.line ), so you find 
intersection between a segment and a line. Then you check if the intesection is 
on the first segment. But you never check the second segment – you simply 
exteneded it to line without checking if the result belongs to it.

Since you trasnfer the call to LineSegment.GetIntersectionWith( Line other ), 
the check is actually useless, since same check is done in that method also. So 
in the segment-segment version you better check second segment, but not repeat 
check for the first segment again.

Some comments:

1)  The difference between parallel Line-Line intersection with other variants 
is a bit awkward. Why two parallel lines cannot have NULL as intersection 
point? Just not sure about the idea behind ... if parallel line-line 
intersection, then it is not valid and we throw exception.  But if parallel 
segment-line, line-segment or segment-segment, then it is OK, just no 
intersection. 

2)  In my opinion it is a bit unusual to call intersection the case, when two 
collinear segments have only one common point. Not really sure it is an 
intersection from geometry point of view. In my opinion all intersecting 
lines/segment should have different slope. Such implementation may be useful in 
some cases, but to me it looks like a hack.

What are going to do with #163? Just close it?

Original comment by andrew.k...@gmail.com on 16 Nov 2010 at 4:41

GoogleCodeExporter commented 9 years ago 
I've put the Pending status just for any further discussion/patch. Will not 
revert changes at this point. Just would like to clarify things, if possible.

Original comment by andrew.k...@gmail.com on 16 Nov 2010 at 4:42

GoogleCodeExporter commented 9 years ago 
> [Bug]
*headscratch*
Ah. Yes. And I see you've fixed it too; thanks.

1) Mostly because I did that before thinking too hard about intersections 
involving line segments. It wouldn't be hard to make 
Line.GetIntersectionWith(Line) return null for parallel and throw on identical. 
(I've gone ahead and attached a patch.)

2) I'm not sure. It may not be an "intersection", but it's definitely the 
single common point. I believe Wolfram 
(http://mathworld.wolfram.com/Intersection.html) says the intersection is the 
set of all common points; we're restricting that to the single common point. In 
the absence of a use case that says one way or another, I went for (what I 
think is) geometrically accurate.

> [Bug 163]
That was what I was thinking, maybe with a note that what was 
Point.DistanceToSegment exists here now?

Original comment by dales...@gmail.com on 23 Nov 2010 at 7:15

Attachments:

GoogleCodeExporter commented 9 years ago 
Done, merged. I think that is all about it.

Original comment by andrew.k...@gmail.com on 25 Nov 2010 at 2:02

GoogleCodeExporter commented 9 years ago

Original comment by andrew.k...@gmail.com on 12 Jan 2011 at 11:45