Mailman 3 December 2012 - scikit-image

Re: High order statistics/Gary level runlength matrices statistics
by Dan 23 Dec '12

23 Dec '12

Is this interesting option going to be implmented in the future release? El miércoles, 28 de noviembre de 2012 05:21:35 UTC-5, Dan escribió: > > Dear all, > > Is Scikit-image able to provide High order statistics/Gary level runlength > matrices statistics values(e.g., Short Run Emphasis, Long Run Emphasis, > Gray-Level Nonuniformity, Run Length Nonuniformity, Run Percentage, etc)?. > > Is there a way to calculate it using Scikit-image? > > Thanks a lot in advance.

2 1

Re: Fourier transform upsample/shift?
by Schönberger Johannes 23 Dec '12

23 Dec '12

Hi, I see you haven't benchmarked against the skimage.transform.* functions, those should be a lot faster compared to `ndimage.map_coordinates`. If your implementation provides a different kind of interpolation, I suggest to integrate them into the current functions and add a new parameter `mode="fourier"`. I'm not sure if there is an equivalent for your functions in all cases (see `transform/_warps.py`). Thoughts? Am 23.12.2012 um 08:30 schrieb Adam <keflavich(a)gmail.com>: > Benchmark tests created: > https://github.com/keflavich/image_registration/blob/master/examples/benchm… > https://github.com/keflavich/image_registration/blob/master/examples/benchm… > > They ought to be more extensive, but give the general results: > -fourier-based shifting is a little faster (maybe 50%), but follows > the same power-law > -fourier-based upsampling is marginally slower, but my test results so > far are questionable because of machine variability > > Of course, with fftw3, one can increase the number of processes used > for the fft, which may speed things up. > > On Sat, Dec 22, 2012 at 8:31 PM, Stéfan van der Walt <stefan(a)sun.ac.za> wrote: >> On Sat, Dec 22, 2012 at 5:24 PM, Adam Ginsburg <keflavich(a)gmail.com> wrote: >>> Reasonable idea. I think they are faster for at least some cases, but they >>> also behave a little differently than other interpolation techniques. So >>> put them all in a "fourier_interpolation.py" file, then put them somewhere >>> in transforms? >> >> Yes, I think some benchmarks would be interesting. Otherwise, we may >> want to add sinc interpolation to ndimage and see if that yields the >> same results. >> >> Thanks! >> Stéfan >> >> -- >> >> > > -- > Adam > > -- > >

1 0

Re: Fourier transform upsample/shift?
by Adam 23 Dec '12

23 Dec '12

Benchmark tests created: https://github.com/keflavich/image_registration/blob/master/examples/benchm… https://github.com/keflavich/image_registration/blob/master/examples/benchm… They ought to be more extensive, but give the general results: -fourier-based shifting is a little faster (maybe 50%), but follows the same power-law -fourier-based upsampling is marginally slower, but my test results so far are questionable because of machine variability Of course, with fftw3, one can increase the number of processes used for the fft, which may speed things up. On Sat, Dec 22, 2012 at 8:31 PM, Stéfan van der Walt <stefan(a)sun.ac.za> wrote: > On Sat, Dec 22, 2012 at 5:24 PM, Adam Ginsburg <keflavich(a)gmail.com> wrote: >> Reasonable idea. I think they are faster for at least some cases, but they >> also behave a little differently than other interpolation techniques. So >> put them all in a "fourier_interpolation.py" file, then put them somewhere >> in transforms? > > Yes, I think some benchmarks would be interesting. Otherwise, we may > want to add sinc interpolation to ndimage and see if that yields the > same results. > > Thanks! > Stéfan > > -- > > -- Adam

1 0

Re: Fourier transform upsample/shift?
by Adam Ginsburg 23 Dec '12

23 Dec '12

Reasonable idea. I think they are faster for at least some cases, but they also behave a little differently than other interpolation techniques. So put them all in a "fourier_interpolation.py" file, then put them somewhere in transforms? On Saturday, December 22, 2012 1:41:22 PM UTC-7, Stefan van der Walt wrote: > > Hi Adam > > On Fri, Dec 21, 2012 at 8:01 PM, Adam Ginsburg <kefl...(a)gmail.com<javascript:>> > wrote: > > > https://github.com/keflavich/image_registration/blob/master/image_registrat… > > > https://github.com/keflavich/image_registration/blob/master/image_registrat… > > > https://github.com/keflavich/image_registration/blob/master/image_registrat… > > (docs here: > > https://image-registration.readthedocs.org/en/latest/fourier_tools.html) > > > These look interesting; do they provide any real speed-up over the > existing routines? If so, how about using a `fourier_` prefix and > implement them all in the same file? > > Regards > Stéfan >

2 1

Re: Fourier transform upsample/shift?
by Stéfan van der Walt 22 Dec '12

22 Dec '12

Hi Adam On Fri, Dec 21, 2012 at 8:01 PM, Adam Ginsburg <keflavich(a)gmail.com> wrote: > https://github.com/keflavich/image_registration/blob/master/image_registrat… > https://github.com/keflavich/image_registration/blob/master/image_registrat… > https://github.com/keflavich/image_registration/blob/master/image_registrat… > (docs here: > https://image-registration.readthedocs.org/en/latest/fourier_tools.html) These look interesting; do they provide any real speed-up over the existing routines? If so, how about using a `fourier_` prefix and implement them all in the same file? Regards Stéfan

1 0

Re: problem with skimage.measure.perimeter (version 0.7.1)
by Schönberger Johannes 22 Dec '12

22 Dec '12

See https://github.com/scikit-image/scikit-image/pull/402.

1 0

Re: problem with skimage.measure.perimeter (version 0.7.1)
by Johannes Schönberger 22 Dec '12

22 Dec '12

Off course, you are right. I'll quickly fix this bug! Thanks for reporting. 2012/12/18 Patrice Munger <patricemunger(a)gmail.com>: > Comparing the code in regionprops.py with the reference > http://www.cs.qub.ac.uk/~d.crookes/webpubs/papers/perimeter.doc, > should line 450 be (1 + sqrt(2) )/ 2: (13, 23) rather than 1 + sqrt(2) / 2: > (13, 23) ? > > -- > >

1 0

Re: Stable vs Dev version, document image enhancement
by Marianne Corvellec 20 Dec '12

20 Dec '12

On December 14, Tony S Yu wrote: > It might be a good idea for us to point to the stable version instead of > the development version by default. +1 For both "Gallery" and "Documentation". On Monday, December 17, 2012 12:57:12 AM UTC+1, Juan Nunez-Iglesias wrote: > > Perhaps you can add a "coming soon" link to the dev gallery? > +1 Would act as a teaser. :) "Coming soon" link on the "Gallery" page of stable version pointing to "Gallery" page of dev version. On December 14, Stefan van der Walt wrote: > The development gallery is useful as an advertisement, but we could maybe make the link to the other versions clearer? +1 I think the above could resolve the issue. The "Versions" box on the right-hand side is visible and great, but again, imho it makes more sense to point to the stable version by default. Marianne

1 1

Re: Segmentation and Granulometry with scikit-image
by Dan 20 Dec '12

20 Dec '12

El lunes, 17 de diciembre de 2012 05:23:00 UTC+1, Tony S Yu escribió: > > > > On Sun, Dec 16, 2012 at 10:41 AM, Dan <dk...(a)walla.co.il <javascript:>>wrote: > >> Hi, >> >> I installed Windows binaries (kindly provided by Christoph Gohlke)- file >> name:scikit-image-0.7.2.win32-py2.7.exe >> >> Thanks >> >> El domingo, 16 de diciembre de 2012 10:36:13 UTC-5, Dan escribió: >> >>> Hi, >>> >>> The output: >>> Python 2.7.3 (default, Apr 10 2012, 23:31:26) [MSC v.1500 32 bit >>> (Intel)] on win32 >>> Type "copyright", "credits" or "license()" for more information. >>> >>> import skimage >>> >>> print skimage.__version__ >>> 0.7.2 >>> >>> print skimage.__file__ >>> C:\Python27\lib\site-packages\**skimage\__init__.pyc >>> >>> >>> >>> Thanks >>> >> > > Hmm... that all looks good to me. > > Maybe something went wrong during the install? This is bizarre, because > `_skeletonize` is a regular python file (not cython, so it's not a build > issue). If you look in > > "C:\Python27\lib\site-**packages\skimage\morphology\" > > > is there a "_skeletonize.py" and does it like the original<https://github.com/scikit-image/scikit-image/blob/v0.7.2/skimage/morphology…> (i.e. > does it have a `skeletonize` function)? > > -Tony > > >>> El domingo, 16 de diciembre de 2012 09:41:35 UTC-5, Tony S Yu escribió: >>>> >>>> On Sat, Dec 15, 2012 at 6:45 PM, Dan <dk...(a)walla.co.il> wrote: >>>> >>>>> Hi, >>>>> >>>>> Thanks for the rapid respnse, I am using 0.7.2 on windows. >>>>> >>>>> The error is now: >>>>> >>>>> Traceback (most recent call last): >>>>> File "C:/Users/DK/PProjects/**skimage_seg/scikit_seg.py", line 3, >>>>> in <module> >>>>> import skimage.morphology as morph >>>>> File "C:\Python27\lib\site-**packages\skimage\morphology\__**init__.py", >>>>> line 7, in <module> >>>>> from ._skeletonize import skeletonize, medial_axis >>>>> ImportError: cannot import name skeletonize >>>>> >>>>> Any idea why?. >>>>> >>>>> Thanks >>>>> >>>> >>>> >>>> Hi Dan, >>>> >>>> How did you install scikit-image? Also, what's the output of the >>>> following? >>>> >>>> >>>> import skimage >>>> print skimage.__version__ >>>> print skimage.__file__ >>>> >>>> >>>> -Tony >>>> >>>> >>>> >>>>> El sábado, 15 de diciembre de 2012 16:24:31 UTC-5, Tony S Yu escribió: >>>>> >>>>>> On Sat, Dec 15, 2012 at 1:58 PM, Dan <dk...(a)walla.co.il> wrote: >>>>>> >>>>>>> Dear Tony, >>>>>>> >>>>>>> Thanks for the script. >>>>>>> >>>>>>> I was running the script and I got this error: >>>>>>> Traceback (most recent call last): >>>>>>> File "C:/Users/DK/PProjects/**skimage**_seg/scikit_seg.py", line >>>>>>> 18, in <module> >>>>>>> img_background = morph.reconstruction(img-h, img) >>>>>>> AttributeError: 'module' object has no attribute 'reconstruction' >>>>>>> >>>>>>> Process finished with exit code 1 >>>>>>> >>>>>>> Any reason why? >>>>>>> >>>>>>> Thanks a lot, >>>>>>> Dan >>>>>>> >>>>>> >>>>>> Hi Dan, >>>>>> >>>>>> Which version of scikit-image are you running? You'll need at least >>>>>> 0.7 (the latest stable release). You can check the version with the >>>>>> following shell command: >>>>>> >>>>>> $ python -c "import skimage; print skimage.__version__" >>>>>> >>>>>> -Tony >>>>>> >>>>> -- >> >> >> > Dear Tony, > > The files were a match. I uninstalled python and the other libraries, > installed them again and it works fine now. I do not what was the problem. > > Thanks a lot for the help. > > Dan

1 0

Re: import of skimage.morphology.selem.disk fails
by Parand Darugar 18 Dec '12

18 Dec '12

Yes indeed, a git pull and rebuild has fixed the issue. Much thanks! Best, Parand On Monday, December 17, 2012 1:37:35 PM UTC-8, Tony S Yu wrote: > > > > On Mon, Dec 17, 2012 at 3:08 AM, Guillaume Gay < > guil...(a)mitotic-machine.org <javascript:>> wrote: > >> Le 17/12/2012 09:03, Stéfan van der Walt a écrit : >> >> Hi Prashand >>> >>> On Sun, Dec 16, 2012 at 11:50 PM, Parand Darugar <dar...(a)gmail.com<javascript:>> >>> wrote: >>> >>>> "from skimage.morphology.selem import disk" fails with the message >>>> "ImportError: cannot import name convex_hull_image". Subsequent >>>> attempts to >>>> import result in message "ImportError: cannot import name _template". >>>> >>> No problems on this end. Could you please send usyour build log? >>> >>> Thanks >>> Stefan >>> >>> Hi >> >> I had very similar import problems when I checked out the source from >> github last friday, that got sorted out by pulling it again sunday, >> so maybe it is enough for you to update your repository and build again... >> >> Guillaume >> >> > I think Guillaume is correct here. The error looks like a circular import > that was fixed in PR #394<https://github.com/scikit-image/scikit-image/pull/394>. > Does updating fix the issue? > > -Tony > >

1 0