Mailman 3 June 2011 - scikit-image

scikits.image sprints at EuroScipy and SciPy 2011
by Stéfan van der Walt 20 Aug '11

20 Aug '11

Hi everyone Just a heads-up about two upcoming scikits.image sprints. The first will be at SciPy 2011, held in Austin, Tx from 11 to 16 July. Then, we'll do the same at the EuroSciPy conference from 25 to 28 August in Paris. Thank you to Emmanuelle Guillart for taking care of the arrangements! At EuroSciPy, we'll be in the room next to the scikits.learn guys, so there should be some great cross-pollination and collaboration happening. If you're around, come and join us to finalise the 3.0 release! Here are the important dates again: * SciPy: July 11 - July 16 2011, Austin, Tx * EuroSciPy: August 25 - 28 2011, Paris Regards Stéfan

2 3

Re: Pull request: fix tv denoise
by Tony Yu 24 Jun '11

24 Jun '11

On Thu, Jun 23, 2011 at 10:47 AM, Maël Primet <mael.primet(a)gmail.com> wrote: > Why would you do that? People will usually change the weight, not eps, so > it is more natural to set weight as the second argument? The point of the patch isn't to set the order of arguments based on preference, but instead, based on consistency. The other functions (`tv_denoise` and `_tv_denoise_3d`) have `eps` followed by `weight`. Most importantly, the arguments passed to `_tv_denoise_2d` are in this order, so that setting the keyword arg `eps` in `tv_denoise` actually adjusts the `weight` argument of `_tv_denoise_2d` (and vice-versa). (See line 268 in https://github.com/stefanv/scikits.image/blob/master/scikits/image/filter/t… ) -Tony

3 2

Re: Pull request: fix tv denoise
by Tony Yu 23 Jun '11

23 Jun '11

I'd agree that `weight` is the more natural variable to change (that's actually the parameter I was trying to change when I found this error). I've added this change to a new branch and submitted this as a separate pull request. Others should probably weigh in on the argument order, but both options are available in the pull requests. -Tony On Thu, Jun 23, 2011 at 11:02 AM, Maël Primet <mael.primet(a)gmail.com> wrote: > Hum.. then I guess we should rather change the other functions to im, > weight, eps, no? > > it makes more sense to be able to call denoise(image, weight) directly > > On Thu, Jun 23, 2011 at 16:58, Tony Yu <tsyu80(a)gmail.com> wrote: > >> >> >> On Thu, Jun 23, 2011 at 10:47 AM, Maël Primet <mael.primet(a)gmail.com>wrote: >> >>> Why would you do that? People will usually change the weight, not eps, so >>> it is more natural to set weight as the second argument? >> >> >> The point of the patch isn't to set the order of arguments based on >> preference, but instead, based on consistency. The other functions >> (`tv_denoise` and `_tv_denoise_3d`) have `eps` followed by `weight`. Most >> importantly, the arguments passed to `_tv_denoise_2d` are in this order, so >> that setting the keyword arg `eps` in `tv_denoise` actually adjusts the >> `weight` argument of `_tv_denoise_2d` (and vice-versa). (See line 268 in >> https://github.com/stefanv/scikits.image/blob/master/scikits/image/filter/t… >> ) >> >> -Tony >> > >

1 0

Pull request: fix tv denoise
by Tony Yu 23 Jun '11

23 Jun '11

A while back, I sent a pull request for a fix to tv_denoise, but I just realized that I never actually emailed the group. The patch is a simple change: the order of arguments passed to a function was reversed. There are couple of other minor changes in addition, as detailed in the pull request: https://github.com/stefanv/scikits.image/pull/25 Best, -Tony

2 1

Re: Re: How to get the skeleton of a binary particle.
by jeanpatrick.pommier＠gmail.com 20 Jun '11

20 Jun '11

To me a particle is a set of 8-connected component, with or without hole. It can have anyshape like this: http://dip4fish.blogspot.com/2011/04/detecting-branched-points-in-skeleton.…. The aim i then to analyse the skeleton and the contour of the particle:http://dip4fish.blogspot.com/2011/06/open-curve-to-ordered-pixels-… I was wandering if there was something like scikits.image.thin( binaryparticle) could be directly done. Jean-Patrick Le , Maël Primet <mael.primet(a)gmail.com> a écrit : > what do you mean by particle? round? you would want to detect the center > then? skeletons will probably not be appropriate in this case because > they are sensible to noise. > There are many existing skeleton algorithms, either using simple > morphological operations, > distance transforms or EDP, but to my knowledge none have been > implemented in this package. > On Fri, Jun 17, 2011 at 18:11, jip jeanpatrick.pommier(a)gmail.com> wrote: > Hi, > Using scikits.image, is it possible to get the skeleton of a binary 2D > particle ? > Best regards > Jean-Patrick

4 3

Re: Re: How to get the skeleton of a binary particle.
by Jean-Patrick Pommier 18 Jun '11

18 Jun '11

2011/6/17 Maël Primet <mael.primet(a)gmail.com> > Then this is simply a connected component :) > > You can do morphological thinning > (see http://homepages.inf.ed.ac.uk/rbf/HIPR2/thin.htm or > http://microboone-docdb.fnal.gov/cgi-bin/RetrieveFile?docid=732&version=1&f… > ) > using simple operations, most of which should be either in sckits.image > where is it? If I do from a ipython/spyder console: In [4]: import scikits.image In [5]: import scikits.image as scim writting scim with a dot, I don't see so much modules, only:data_dir, getlog,test,version. > or easily implementable from numpy > > > Maël > > Do you know how to handle the "don't care points" in a structuring element when using the hit and miss operator provided by ndimage? I can do it under ubuntu with mahotas, but not under windows due to freeimage.dll issues. Thanks for your help. Jean-Patrick

1 0

Re: How to get the skeleton of a binary particle.
by Zachary Pincus 18 Jun '11

18 Jun '11

> Then this is simply a connected component :) > > You can do morphological thinning > (see http://homepages.inf.ed.ac.uk/rbf/HIPR2/thin.htm or http://microboone-docdb.fnal.gov/cgi-bin/RetrieveFile?docid=732&version=1&f… ) > using simple operations, most of which should be either in sckits.image or easily implementable from numpy > Also, CellProfiler has good skeletonization routines, amongst others: https://svn.broadinstitute.org/CellProfiler/trunk/CellProfiler/cellprofiler… It's built on top of scipy.ndimage where practical, and with extensions written in cython otherwise. Check out cpmorphology.py for the skeletonization stuff. (The cpmath is pretty stand-alone from the rest of the package.) Zach

2 2

Re: How to get the skeleton of a binary particle.
by Maël Primet 17 Jun '11

17 Jun '11

what do you mean by particle? round? you would want to detect the center then? skeletons will probably not be appropriate in this case because they are sensible to noise. There are many existing skeleton algorithms, either using simple morphological operations, distance transforms or EDP, but to my knowledge none have been implemented in this package. On Fri, Jun 17, 2011 at 18:11, jip <jeanpatrick.pommier(a)gmail.com> wrote: > Hi, > Using scikits.image, is it possible to get the skeleton of a binary 2D > particle ? > > Best regards > > Jean-Patrick

1 0

How to get the skeleton of a binary particle.
by jip 17 Jun '11

17 Jun '11

Hi, Using scikits.image, is it possible to get the skeleton of a binary 2D particle ? Best regards Jean-Patrick

1 0

A good article on optimization
by Pieter Holtzhausen 06 Jun '11

06 Jun '11

http://scikit-learn.sourceforge.net/dev/developers/performance.html from our scikits-learn brethren. And here is a google summer of code 2007 participant: http://www.python.org/dev/peps/pep-0368 I'm interested in how much one would gain in comparison with a numpy based approach.

3 2