[Numpy-discussion] Bilateral filter

[Zachary Pincus]

Hi Stéfan,

>>> http://github.com/stefanv/bilateral.git
>>
>> Cool! Does this, out of curiosity, break things for you? (Or Nadav?)
>
> I wish I had some way to test.  Do you maybe have a short example that
> I can convert to a test?

Here's my test case for basic working-ness (e.g. non exception- 
throwing) of that bilateral code:

In [7]: bilateral.bilateral(numpy.arange(25).reshape((5,5)), 4, 10)

Out[7]:
array([[ 7,  7,  7,  8,  8],
        [ 9,  9,  9, 10, 10],
        [11, 11, 12, 12, 12],
        [13, 13, 14, 14, 14],
        [15, 15, 16, 16, 16]])

That's all I'd been using to provoke the errors before, so presumably  
if you get that far with the fixed code, then things should be good as  
far as cython's concerned?

>> I'm all for it. I've got a few other bits lying around that might be
>> good there too:
>>  - 2D iso-contour finding (sub-pixel precision)
>>  - 2D image warping via thin-plate splines
>
>> I also have some code for various geometric algorithms lying around:
>>  - calculating optimal rigid alignments of point-sets ("Procrustes
>> Analysis")
>>  - line intersections, closest points to lines, distance to lines,  
>> etc.
>>
>> if that would be of any use to anyone.
>
> Definitely.  In addition I have code for polygon clipping, hough
> transforms, grey-level co-occurrence matrices, connected components,
> shortest paths and linear position-invariant filtering.

Aah, fantastic. The co-occurrence matrix stuff will be very useful to  
me!

Zach