[Numpy-discussion] Count occurrences in a nested ndarray

[josef.pktd@gmai...]

On Tue, Oct 19, 2010 at 9:09 PM, Dewald Pieterse
<dewald.pieterse@gmail.com> wrote:
> Hi All
> I have a nested array created using:
> edges = scipy.misc.pilutil.imfilter(prent,'find_edges')
> edges
> array([[[  0, 255,   0],
>         [  0, 255,   0],
>         [  0, 255,   0],
>         ...,
>         [  0, 255,   0],
>         [  0, 255,   0],
>         [  0, 255,   0]],
>        [[  0, 255,   0],
>         [  0,   0,   0],
>         [  0,   0,   0],
>         ...,
>         [  0,   0,   0],
>         [  0,   0,   0],
>         [  0, 255,   0]],
>        [[  0, 255,   0],
>         [  0,   0,   0],
>         [  0,   0,   0],
>         ...,
>         [  0,   0,   0],
>         [  0,   0,   0],
>         [  0, 255,   0]],
>        ...,
>        [[  0, 255,   0],
>         [  0,   0,   0],
>         [  0,   0,   0],
>         ...,
>         [  0,   0,   0],
>         [  0,   0,   0],
>         [  0, 255,   0]],
>        [[  0, 255,   0],
>         [  0,   0,   0],
>         [  0,   0,   0],
>         ...,
>         [  0,   0,   0],
>         [  0,   0,   0],
>         [  0, 255,   0]],
>        [[  0, 255,   0],
>         [  0, 255,   0],
>         [  0, 255,   0],
>         ...,
>         [  0, 255,   0],
>         [  0, 255,   0],
>         [  0, 255,   0]]], dtype=uint8)
> I want to count the number of occurrences of certain unique elements in the
> array, I know what the elements are that I want to count [0,255,0],
> [255,0,0] and [0,0,255].
> I want to count the number of pixels of a particular color in a picture to
> determine the edge length and calculate areas and such.
> array does not have a array.count() method and trying to count the instances
> using a nested for loop like:
> for xiter in range(xindex):
> for yiter in range(yindex):
> if edges[xiter,yiter,:] == [255,0,0]:
> groenpixelarea = groenpixelarea + 1
> if edges[xiter,yiter,:] == [0,255,0]:
> rooipixelarea = rooipixelarea + 1
> if edges[xiter,yiter,:] == [0,0,255]:
> bloupixelarea = bloupixelarea + 1
> results in:
>     16 for xiter in range(xindex):
>      17         for yiter in range(yindex):
> ---> 18                 if edges[xiter,yiter,:].any() == [255,0,0]:
>      19                         groenpixelarea = groenpixelarea + 1
>      20                 if edges[xiter,yiter,:] == [0,255,0]:
> ValueError: The truth value of an array with more than one element is
> ambiguous. Use a.any() or a.all()
> WARNING: Failure executing file: <analiseerverwerkteprent.py>
>
> What am I doing wrong?

if (edges[xiter,yiter,:] == [255,0,0]).all()

makes element wise comparison and then joins them by all

but you should be able to do this for all pixels at one

 (edges == np.array([255,0,0])[None, None,:]).all(-1).sum()

or something like this

For more complex patterns there might be something fast in scipy.ndimage.

Josef


> Thanx
> --
> Dewald Pieterse
>
> _______________________________________________
> NumPy-Discussion mailing list
> NumPy-Discussion@scipy.org
> http://mail.scipy.org/mailman/listinfo/numpy-discussion
>
>