Hi Mario,
I don't understand your question. Are you asking why you cannot create a
multipoint from the value of numpy.where(parr>0)?
Shapely is designed to initialize from N X 2 arrays (arrays of x, y
values) like
>>> a
array([[ 0., 0.],
[ 0., 2.],
[ 3., 3.]])
>>> m = MultiPoint(a)
>>> m
<shapely.geometry.multipoint.MultiPoint object at 0x8580e8c>
>>> m.wkt
'MULTIPOINT (0.0000000000000000 0.0000000000000000, 0.0000000000000000
2.0000000000000000, 3.0000000000000000 3.0000000000000000)'
If you can convert your data into such an array, I think you're on your way.
Sean
Mario Ceresa wrote:
> Ok I got it: converting the array to a list and then zipping will do the
> trick.
> ------------------------------------------------------
> from shapely.geometry import MultiPoint
> import numpy
>
> parr = numpy.zeros((6,6))
> parr[0,0] = 1
> parr[0,2] = 1
> parr[3,3] = 1
> print 'parr=',parr
> points = numpy.where(parr>0)
> print 'points=',points
> plist = numpy.vstack(points).tolist()
> print 'plist=',plist
> points_good = zip(plist[0],plist[1])
> print 'points_good=',points_good
> mp = MultiPoint(points_good)
> print 'list=',list((p.x,p.y) for p in mp.geoms)
> ------------------------------------------------------
> parr= [[ 1. 0. 1. 0. 0. 0.]
> [ 0. 0. 0. 0. 0. 0.]
> [ 0. 0. 0. 0. 0. 0.]
> [ 0. 0. 0. 1. 0. 0.]
> [ 0. 0. 0. 0. 0. 0.]
> [ 0. 0. 0. 0. 0. 0.]]
> points= (array([0, 0, 3]), array([0, 2, 3]))
> plist= [[0, 0, 3], [0, 2, 3]]
> points_good= [(0, 0), (0, 2), (3, 3)]
> list= [(0.0, 0.0), (0.0, 2.0), (3.0, 3.0)]
> -------------------------------------------------------
>
> I was working directly with array because I have to create a lot of
> objects and I was hoping to speed things up a bit: as always premature
> optimization is the root of all evil :)
>
> Anyway if someone still happen to know why it is not working with
> arrays or how to quickly convert binary images in shape objects I'd
> love to hear it!
>
> Thanks and regards
>
> Mario
>
> On Sun, Aug 17, 2008 at 1:57 PM, Mario Ceresa
> <[EMAIL PROTECTED]> wrote:
>> Hello again,
>> Apparently I found that using dtype=float instead of numpy.float32
>> better the whole thing:
>> -------------------------------------------------------
>> [[ 1. 0. 1. 0. 0. 0.]
>> [ 0. 0. 0. 0. 0. 0.]
>> [ 0. 0. 0. 0. 0. 0.]
>> [ 0. 0. 0. 1. 0. 0.]
>> [ 0. 0. 0. 0. 0. 0.]
>> [ 0. 0. 0. 0. 0. 0.]]
>> (array([0, 0, 3]), array([0, 2, 3]))
>> [[ 0. 0.]
>> [ 0. 2.]
>> [ 3. 3.]]
>> [(0.0, 0.0), (3.0, 0.0), (2.0, 3.0)]
>> ---------------------------------------------------------
>> so probably the last error was a casting problem (float32->float64).
>> Still I'd love to obtain [(0.0, 0.0), (0.0, 2.0), (3.0, 3.0)] as
>> coordinates in the multpoint...
>> It looks likes it is reading vertically column-by-column. Maybe
>> numpy.transpose() was a bad idea?
>>
>> *perplexed*
>>
>> Mario
>>
>> PS: BTW I use numpy 1.1.1
>>
>> On Sun, Aug 17, 2008 at 1:17 PM, Mario Ceresa
>> <[EMAIL PROTECTED]> wrote:
>>> Hi all,
>>> first thanks you all for shapely: I've just started using it and it's
>>> a wonderful product!
>>> But there is a things that leaves me puzzled: whenever i run the following
>>> code
>>>
>>> -------------------
>>> from shapely.geometry import MultiPoint
>>> import numpy
>>>
>>> parr = numpy.zeros((6,6))
>>> parr[0,0] = 1
>>> parr[0,2] = 1
>>> parr[3,3] = 1
>>> print parr
>>> points = numpy.where(parr>0)
>>> print points
>>> points_trans =
>>> numpy.asarray(numpy.transpose(numpy.vstack(points)),dtype=numpy.float32)
>>> print points_trans
>>> mp = MultiPoint(points_trans)
>>> print list((p.x,p.y) for p in mp.geoms)
>>> ---------------------
>>>
>>> I'm expecting that values in points_trans and mp.geoms should be equals.
>>> But this is actually the output I get:
>>>
>>> parr = [[ 1. 0. 1. 0. 0. 0.]
>>> [ 0. 0. 0. 0. 0. 0.]
>>> [ 0. 0. 0. 0. 0. 0.]
>>> [ 0. 0. 0. 1. 0. 0.]
>>> [ 0. 0. 0. 0. 0. 0.]
>>> [ 0. 0. 0. 0. 0. 0.]]
>>>
>>> points = (array([0, 0, 3]), array([0, 2, 3]))
>>>
>>> points_trans = [[ 0. 0.] [ 0. 2.] [ 3. 3.]]
>>>
>>> list = [(0.0, 5.325712092559326e-315), (32.000007629394531,
>>> 1.6304166312761136e-322), (9.2814421247954897e-317,
>>> 1.9762625833649862e-323)]
>>>
>>> This happen on fed core 9 - x64 - python 2.5 - Shapely 1.06, geos 3.0
>>> and on RHEL5 x64 - python 2.4 - Shapely 1.06 - geos 3.0.
>>> On this platform the last line is:
>>>
>>> list = [(0.0, 5.325712092559326e-315), (32.000007629394531,
>>> 1.6304166312761136e-322), (2.4278924170566564e-315,
>>> 2.4180091278772481e-315)]
>>>
>>> Basically I just want to take a binary image and create a shape object
>>> which represent the image . Is there something that I got wrong? or a
>>> simpler way to do it?
>>>
>>> Thanks and regards
>>>
>>> Mario Ceresa
> _______________________________________________
> Community mailing list
> [email protected]
> http://lists.gispython.org/mailman/listinfo/community
>
_______________________________________________
Community mailing list
[email protected]
http://lists.gispython.org/mailman/listinfo/community
