Le 07/09/2011 00:02, Curtis Jensen a écrit :
> Below is a simpler example. Here, set2 is a polygon completely
> encompassed by the set1 polygon. Yet the difference function produces
> a single polygon that doesn't seem to be a difference in any sense
> that I can understand. How are the verticies of a polygon suppose to
> be interpreted?
Sorry not to answer as fast as I want, I am a little busy.
I did not find the time yet to look at your example (only looked at the
polygons files you provided, not how [math] handles them).
Vertices have their straightforward meaning in [math]. However, the
underlying representation is BSP tree, and vertices are only a converted
representation. There may well be bugs in this conversion step. One way
to see if the underlying representation (i.e. the BSP tree) is good is
to generate a fine grained grid of points (xi, yi) and to use the
checkPoint method from the top level Region class to see which points
are inside and which points are outside. This could be a first step
understanding where the problem lies.
I'll try to have a look at this later on, I am sorry for the delay.
Luc
>
> Thanks,
> Curtis
>
>
> public void testdifferenceWithHole() {
>
> Vector2D[][] vertices1 = new Vector2D[][] {
> new Vector2D[] {
> new Vector2D(7.2, 3.1),
> new Vector2D(7.2, 0.1),
> new Vector2D(4.2, 0.1),
> new Vector2D(4.2, 3.1)
> }
> };
> PolygonsSet set1 = buildSet(vertices1);
>
> Vector2D[][] vertices2 = new Vector2D[][] {
> new Vector2D[] {
> new Vector2D(5.7, 1.6),
> new Vector2D(4.0, 1.0),
> new Vector2D(4.0, 2.0)
> }
> };
> PolygonsSet set2 = buildSet(vertices2);
>
> PolygonsSet setDiff = (PolygonsSet) new
> RegionFactory<Euclidean2D>().difference(set1.copySelf(),
> set2.copySelf());
> Vector2D[][] diffVerts = setDiff.getVertices();
> for (int i = 0; i< diffVerts.length; i++) {
> System.out.println("Verts: " + i);
>
> Vector2D[] set = diffVerts[i];
> for (Vector2D vertex : set) {
> System.out.println("\t" + vertex);
> }
> }
> }
>
>
> On Fri, Aug 26, 2011 at 11:41 AM, Curtis Jensen<curtis@thejensens.org> wrote:
>> Using math 3.0, I have two polygons with many points. One is
>> completely contained within the other. When I do a difference on the
>> two, I expected to get a polygon with a hole in it. However, I get 86
>> polygons, that roughly make up a polygon with a hole in it. If I
>> scale the points by a factor of 0.1, I get 7 polygons, and if I scale
>> it differently in the two directions, I get a different number of
>> polygons. Sometimes the resultant polygons don't seem to make a shape
>> resembling a polygon with a hole in it.
>>
>> How should I interpret the results of the difference method? i.e. How
>> do I process the 86 or 7 or however many polygons so that it resembles
>> 1 polygon with 1 hole in it?
>>
>> Thanks,
>> Curtis
>>
>>
>> Attached are two csv files with the points in CCW order. Also
>> attached is a plot of the points in the two files. Below is code I
>> added to the org.apache.commons.math.geometry.euclidean.twod.PolygonsSetTest
>> class to test with (It uses the Apache Common FileUtils too)
>>
>>
>>
>> @Test
>> public void testDifferenceManyPoints() throws IOException {
>> PolygonsSet set1 = csv2set(new File("src_ccw.csv"));
>> PolygonsSet set2 = csv2set(new File("inner_ccw.csv"));
>>
>> PolygonsSet set = (PolygonsSet) new
>> RegionFactory<Euclidean2D>().difference(set1.copySelf(),
>> set2.copySelf());
>> Vector2D[][] verts = set.getVertices();
>> System.out.println(verts.length);
>> }
>>
>> private PolygonsSet csv2set(File file) throws IOException {
>> List linesObj = FileUtils.readLines(file);
>>
>> Vector2D[][] verts = new Vector2D[1][linesObj.size()];
>> for (int i = 0; i< linesObj.size(); i++) {
>> String line = (String)linesObj.get(i);
>> String[] tokens = line.split(",");
>>
>> double x = Double.valueOf(tokens[0]);
>> double y = Double.valueOf(tokens[1]);
>>
>> verts[0][i] = new Vector2D(x, y);
>> }
>>
>> return buildSet(verts);
>> }
>>
>
> 
> To unsubscribe, email: userunsubscribe@commons.apache.org
> For additional commands, email: userhelp@commons.apache.org
>
>

To unsubscribe, email: userunsubscribe@commons.apache.org
For additional commands, email: userhelp@commons.apache.org
