Andrea Aime wrote:
Hum.... I'm wondering what changes did you made. It's easy to make
decimation more efficient
by making it more aggressive, but a properly working decimation
generates results that cannot
be distinguished from not having applied decimation at all.
When I coded it, I made sure the decimation distance was not over 1
pixel on screen, but I
don't really know what happened in the meantime.
It, basically, wasnt actually decimating. It would, however, generate
lines with 0 or 1 point in them, and polygon rings with 0, 1, 2 points
and 1st point != last point. These geometry issues were causing
problems, which this version solves.
Note - MathTransform in this code is always an affine transform (see my
earlier message about CRS being done in the wrong place).
Also note that this is always done on a copy of the geometry since its
destructive.
A better way to do this would be to generate a spanx/spany for each
geometry instead of one thats "good for all geometries", then
(hopefully) you'll be able to do this in 1 step instead of 2.
dave
/**
* 1. remove any points that are within the spanx,spany. We
ALWAYS keep 1st and last point
* 2. transform to screen coordinates
* 3. remove any points that are close (span <1)
*
* @param seq
* @param tranform
*/
private final void
decimateTransformGeneralize(LiteCoordinateSequence seq,MathTransform
transform) throws TransformException
{
//decimates before XFORM
int ncoords = seq.size();
double originalOrds[] = seq.getXYArray(); // 2*#of points
if (ncoords <2)
{
if (ncoords ==1) // 1 coordinate -- just xform it
{
double[] newCoordsXformed2= new double[2];
transform.transform(originalOrds, 0,
newCoordsXformed2, 0, 1);
seq.setArray(newCoordsXformed2);
return;
}
else
return; // ncoords =0
}
// unfortunately, we have to keep things in double precion
until after the transform or we could move things.
double[] allCoords = new double[ncoords*2]; // preallocate --
might not be full (throw away Z)
allCoords[0] = originalOrds[0]; //allways have 1st one
allCoords[1] = originalOrds[1];
int actualCoords = 1;
double lastX = allCoords[0];
double lastY = allCoords[1];
for (int t=1;t<(ncoords-1);t++)
{
//see if this one should be added
double x = originalOrds[t*2];
double y = originalOrds[t*2+1];
if ( (Math.abs(x-lastX )> spanx) || ( Math.abs(y-lastY
)) >spany)
{
allCoords[actualCoords*2] = x;
allCoords[actualCoords*2+1] = y;
lastX = x;
lastY = y;
actualCoords++;
}
}
allCoords[actualCoords*2] = originalOrds[(ncoords-1)*2];
//always have last one
allCoords[actualCoords*2+1] = originalOrds[(ncoords-1)*2+1];
actualCoords++;
double[] newCoordsXformed;
//DO THE XFORM
if ((transform == null) || (transform.isIdentity()) ) // no
actual xform
{
newCoordsXformed = allCoords;
}
else
{
newCoordsXformed= new double[actualCoords*2];
transform.transform(allCoords, 0, newCoordsXformed, 0,
actualCoords);
}
//GENERLIZE -- we should be in screen space so spanx=spany=1.0
// unfortunately, we have to keep things in double
precion until after the transform or we could move things.
double[] finalCoords = new double[ncoords*2]; //
preallocate -- might not be full (throw away Z)
finalCoords[0] = newCoordsXformed[0]; //allways have 1st one
finalCoords[1] = newCoordsXformed[1];
int actualCoordsGen = 1;
lastX = newCoordsXformed[0];
lastY = newCoordsXformed[1];
for (int t=1;t<(actualCoords-1);t++)
{
//see if this one should be added
double x = newCoordsXformed[t*2];
double y = newCoordsXformed[t*2+1];
if ( (Math.abs(x-lastX )> 1) || ( Math.abs(y-lastY
)) >1)
{
finalCoords[actualCoordsGen*2] = x;
finalCoords[actualCoordsGen*2+1] = y;
lastX = x;
lastY = y;
actualCoordsGen++;
}
}
finalCoords[actualCoordsGen*2] =
newCoordsXformed[(actualCoords-1)*2]; //always have last one
finalCoords[actualCoordsGen*2+1] =
newCoordsXformed[(actualCoords-1)*2+1];
actualCoordsGen++;
//stick back in
double[] seqDouble = new double[2*actualCoordsGen];
System.arraycopy(finalCoords,0,seqDouble, 0,
actualCoordsGen*2);
seq.setArray(seqDouble);
}
-------------------------------------------------------
This SF.Net email is sponsored by xPML, a groundbreaking scripting language
that extends applications into web and mobile media. Attend the live webcast
and join the prime developer group breaking into this new coding territory!
http://sel.as-us.falkag.net/sel?cmd=lnk&kid=110944&bid=241720&dat=121642
_______________________________________________
Geotools-devel mailing list
Geotools-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/geotools-devel
