Mailman 3 April 2013 - scikit-image

Re: multi-page TIFF writer
by Hadrien Mary 12 Apr '13

12 Apr '13

Hi Juan, You should have a look at the excellent tiffile.py: http://www.lfd.uci.edu/~gohlke/code/tifffile.py.html -- Hadrien Mary Ph.D student in Biology Tournier-Gachet Team CNRS - LBCMCP - UMR 5088 Université de Toulouse - Bât. 4R3B1 118, route de Narbonne - 31062 Toulouse On Fri, Apr 12, 2013 at 2:06 PM, Stéfan van der Walt <stefan(a)sun.ac.za>wrote: > Hi Juan > > On Fri, Apr 12, 2013 at 1:46 PM, Juan Nunez-Iglesias <jni.soma(a)gmail.com> > wrote: > > Do any of you know about writing a numpy array to a multi-page TIFF file > in > > Python? Last time I checked, you could only read those with PIL, not > write. > > A quick Googling led me to Tiffany, but it doesn't seem to be > particularly > > numpy-friendly... > > The freeimage plugin has code for this already, although we still need > to implement an "imsave_collection" wrapper. Grep for > "write_multipage". > > Regards > Stéfan > > -- > You received this message because you are subscribed to the Google Groups > "scikit-image" group. > To unsubscribe from this group and stop receiving emails from it, send an > email to scikit-image+unsubscribe(a)googlegroups.com. > For more options, visit https://groups.google.com/groups/opt_out. > > >

1 0

Re: multi-page TIFF writer
by Stéfan van der Walt 12 Apr '13

12 Apr '13

Hi Juan On Fri, Apr 12, 2013 at 1:46 PM, Juan Nunez-Iglesias <jni.soma(a)gmail.com> wrote: > Do any of you know about writing a numpy array to a multi-page TIFF file in > Python? Last time I checked, you could only read those with PIL, not write. > A quick Googling led me to Tiffany, but it doesn't seem to be particularly > numpy-friendly... The freeimage plugin has code for this already, although we still need to implement an "imsave_collection" wrapper. Grep for "write_multipage". Regards Stéfan

1 0

Re: Coordinate overlap in two labelled images
by Michael Aye 12 Apr '13

12 Apr '13

Great stuff, Juan, clear and concise. Thanks very much, it's very instructive! On Tuesday, April 2, 2013 9:33:11 PM UTC-7, Juan Nunez-Iglesias wrote: > > Suggestion number 3. I'll deal with each of three interpretations of your > question. Suppose your images are im1 and im2. Now, you might want 1) of > the regions in im1, the ones that intersect over at least 1 pixel with > regions of im2; 2) the regions that are part of regions in both im1 and im2 > (ie the intersection of regions in im1 and im2); or 3) the union of regions > of im1 and im2 that have at least one pixel intersection with the other > image. > > The last example assumes the union'ed regions don't touch each other. > > Then: > > In [1]: import scipy.ndimage as nd > > In [2]: im1 = np.array([[0, 0, 0, 0], [1, 1, 0, 2], [0, 0, 0, 2]]) > > In [3]: im1 > Out[3]: > array([[0, 0, 0, 0], > [1, 1, 0, 2], > [0, 0, 0, 2]]) > > In [4]: im2 = np.array([[0, 1, 0, 0], [0, 1, 0, 0], [0, 1, 0, 0]]) > > In [5]: im2 > Out[5]: > array([[0, 1, 0, 0], > [0, 1, 0, 0], > [0, 1, 0, 0]]) > > In [6]: intersect = im1.astype(bool) * im2.astype(bool) > > In [7]: intersect > Out[7]: > array([[False, False, False, False], > [False, True, False, False], > [False, False, False, False]], dtype=bool) > > In [8]: overlap_labels1 = np.unique(im1[intersect]) > > In [9]: selector = np.zeros(im1.max() + 1, int) > > In [10]: selector[overlap_labels1] = arange(1, len(overlap_labels1) + 1) > > In [11]: reg1 = selector[im1] > > In [12]: reg1 > Out[12]: > array([[0, 0, 0, 0], > [1, 1, 0, 0], > [0, 0, 0, 0]]) > > In [13]: reg2 = nd.label(intersect)[0] > > In [14]: reg2 > Out[14]: > array([[0, 0, 0, 0], > [0, 1, 0, 0], > [0, 0, 0, 0]], dtype=int32) > > In [15]: overlap_labels2 = np.unique(im2[intersect]) > > In [16]: selector2 = np.zeros(im2.max() + 1, int) > > In [17]: selector2[overlap_labels2] = 1 > > In [18]: nonzero_pixels = reg1 + selector2[im2] > > In [19]: reg3 = nd.label(nonzero_pixels)[0] > > In [20]: reg3 > Out[20]: > array([[0, 1, 0, 0], > [1, 1, 0, 0], > [0, 1, 0, 0]], dtype=int32) > > > > > > On Wed, Apr 3, 2013 at 5:52 AM, Stéfan van der Walt <ste...(a)sun.ac.za<javascript:> > > wrote: > >> Hi Stuart >> >> On Tue, Apr 2, 2013 at 7:04 PM, Stuart Mumford <little...(a)gmail.com<javascript:>> >> wrote: >> > I am using your (amazingly fast) label image routines and regionprops >> to extract the coordinates of the regions in two images. I need to filter >> these objects out if they do not overlap in both the images, then renumber >> all the remaining regions. At the moment I am doing something unspeakable >> with nested loops and sets which is pretty damn slow; I was hoping someone >> here would have a really clever idea on how to solve this. Failing that >> does anyone know how I can find the number of coordinates (output from >> regionprops) that overlap in two numpy arrays? >> >> You can use join_segmentations to find an intersection of your two >> label maps. Then, walk through the newly formed regions and check at >> each point whether the object exists in both images (that way, you >> don't have so many comparisons to do). >> >> >> http://scikit-image.org/docs/dev/auto_examples/plot_join_segmentations.html >> >> If that's not exactly what you had in mind, you can use this code to >> find all the connected regions in an image: >> >> https://github.com/stefanv/lulu/blob/master/lulu/base.pyx#L21 >> >> These regions are encoded sparsely, and testing their intersection is >> very fast (see the algorithm section of this paper: >> https://scholar.sun.ac.za/handle/10019.1/4883). >> >> I hope that helps! >> Stéfan >> >> -- >> You received this message because you are subscribed to the Google Groups >> "scikit-image" group. >> To unsubscribe from this group and stop receiving emails from it, send an >> email to scikit-image...(a)googlegroups.com <javascript:>. >> For more options, visit https://groups.google.com/groups/opt_out. >> >> >> >

2 1

multi-page TIFF writer
by Juan Nunez-Iglesias 12 Apr '13

12 Apr '13

Hi everyone, Do any of you know about *writing* a numpy array to a multi-page TIFF file in Python? Last time I checked, you could only read those with PIL, not write. A quick Googling led me to Tiffany<https://bitbucket.org/pydica/tiffany>, but it doesn't seem to be particularly numpy-friendly... Thanks! Juan.

1 0

Re: GSOC 2013 Proposal
by siddharth kherada 12 Apr '13

12 Apr '13

Thanks for the feedback. a) I want to point out that the domain transform filter is better than bilateral filter and can be used in various applications in which bilateral filter is used. b) Image denoising using non-local means method. Is this already implemented? http://inf.ufrgs.br/~eslgastal/AdaptiveManifolds/Gastal_Oliveira_SIGGRAPH20… c) Some Image Debluring techniques: 1) Deblur image using blind deconvolution 2) Deblur image using Lucy-Richardson method 3) Deblur image using regularized filter 4) Deblur image using Wiener filter What is your opinion about the above techniques? Best Regards, Siddharth On Tuesday, 9 April 2013 17:01:15 UTC+5:30, Stefan van der Walt wrote: > > Hi Siddharth > > On Tue, Apr 9, 2013 at 11:32 AM, siddharth kherada > <siddharth...(a)gmail.com <javascript:>> wrote: > > I am currently pursuing MS by Research in the field of image processing > at > > IIIT- Hyderabad, India. I am interested in working for Scikit-image this > > summer in a GSoC project. > > Thanks for getting in touch. Both these algorithms are cute, and > we'll consider PRs for them if they come in, but I don't think they > are major underlying parts of the library we'd want to spend a GSoC > on. If you are interested in this sort of thing, image deconvolution > may be more relevant. For GSoC, we'd like to implement algorithms > that will be useful across a wide array of disciplines, if possible. > > Also, we require that students who would like to participate in GSoC > first make other contributions to the project, so be sure to look at > the list of tickets and try to fix some of them, at a minimum. > > Regards > Stéfan >

2 3

Re: Annoying gstreamer bug
by Stéfan van der Walt 11 Apr '13

11 Apr '13

Hi Juan On Thu, Apr 11, 2013 at 10:28 AM, Juan Nunez-Iglesias <jni.soma(a)gmail.com> wrote: > Is no one concerned that anyone writing a command-line tool that leverages > skimage.io will have their help menu overridden? (If they have GStreamer > installed.) This is a bug--I have it fixed in one of my branches, but looks like I forgot to make a PR. I'll see if I can track it down. I think what stopped me is that I realized that the gstreamer video reading is somewhat broken, and tried to get it up and running properly (which I never did). Stéfan

1 0

Re: Annoying gstreamer bug
by Juan Nunez-Iglesias 11 Apr '13

11 Apr '13

Is no one concerned that anyone writing a command-line tool that leverages skimage.io will have their help menu overridden? (If they have GStreamer installed.) On Wed, Apr 10, 2013 at 12:23 AM, Juan Nunez-Iglesias <jni.soma(a)gmail.com>wrote: > Hi All, > > Consider the following code, saved in a file called test.py: > > import gst > > I will bet that few among you would predict this output: > > $ python test.py -h > Usage: > test.py [OPTION...] - GStreamer initialization > > Help Options: > -h, --help Show help options > --help-all Show all help options > --help-gst Show GStreamer Options > > > It seems that someone forgot an > > if __name__ == '__main__' > guard somewhere! > > Now, this would be none of our concern except for the following corollary: > > $ python -c 'from skimage.io.collection import alphanumeric_key' -h > Usage: > -c [OPTION...] - GStreamer initialization > > Help Options: > -h, --help Show help options > --help-all Show all help options > --help-gst Show GStreamer Options > > > Now, I don't know how to fix this, because it make sense that `io` would > import `video` (although `from .video import *` is sloppy). But any > suggestions are very very welcome. > > Juan. >

1 0

Re: Histogram backprojection
by Tony Yu 11 Apr '13

11 Apr '13

On Wed, Apr 10, 2013 at 9:12 PM, abid rahman <abidrahman2(a)gmail.com> wrote: > > But it makes another question : If scipy.ndimage is there and scikit is > using scipy, what is the need of scikit-image ? scipy.ndimage itself is a > good library. > Hi Abid, You should think of `scikit-image` as a complement to `scipy.ndimage`: `scikit-image` adds additional capabilities to what is provided by `scipy.ndimage`. In fact, the maintainers of `scipy.ndimage`<https://github.com/scipy/scipy/blob/master/doc/MAINTAINERS.rst.txt> are Stefan van der Walt, the creator of `scikit-image`, and Thouis (Ray) Jones, a frequent contributor. Part of the rationale for the development of scikits (not just `scikit-image`) was a faster, independent development cycle. This has allowed the SciPy Stack (I think that's the tentative name for the entire scientific python ecosystem, not just `scipy` the package) to grow much faster than before. Cheers, -Tony > Regards > > > On Thu, Apr 11, 2013 at 6:08 AM, Josh Warner <silvertrumpet999(a)gmail.com>wrote: > >> I believe we generally leverage the convolution tools in `scipy.ndimage`. >> On a high level, the scikit-image project is an extension toolkit for SciPy >> so we generally* don't reinvent the wheel. >> >> * there are always exceptions. >> >> >> On Wednesday, April 10, 2013 6:13:52 AM UTC-5, abid rahman wrote: >>> >>> Doesn't scikit-image has a 2d convolution function like cv2.filter2D in >>> OpenCV which enables convolution of an image (or any numpy array) with >>> custom kernel ? >>> >>> On Tuesday, 9 April 2013 22:10:26 UTC+5:30, abid rahman wrote: >>>> >>>> Hi, >>>> >>>> I would like to add a feature on histogram backprojection as proposed >>>> by *Swain & Michael J.* in their paper *"Indexing via color histograms" >>>> *.. Earlier, I had written a simple article on that ( it was focussed >>>> for OpenCV) in my blog : http://opencvpython.blogspot.** >>>> in/2013/03/histograms-4-back-**projection.html.<http://opencvpython.blogspot.in/2013/03/histograms-4-back-projection.html.> >>>> >>>> I can have some simple modification to it to integrate it into >>>> scikit-image. So I would like to know the opinion of senior developers on >>>> this proposal. >>>> >>>> Regards >>>> Abid K. >>>> >>> -- >> You received this message because you are subscribed to the Google Groups >> "scikit-image" group. >> To unsubscribe from this group and stop receiving emails from it, send an >> email to scikit-image+unsubscribe(a)googlegroups.com. >> For more options, visit https://groups.google.com/groups/opt_out. >> >> >> > > > > -- > Abid K. > opencvpython.blogspot.com > > -- > You received this message because you are subscribed to the Google Groups > "scikit-image" group. > To unsubscribe from this group and stop receiving emails from it, send an > email to scikit-image+unsubscribe(a)googlegroups.com. > For more options, visit https://groups.google.com/groups/opt_out. > > >

1 0

Re: Histogram backprojection
by abid rahman 11 Apr '13

11 Apr '13

Hi, I ran into a problem with scikit installation in windows 7. I am using python 2.7 32 bit. python setup.py install shows error : Unable to find vcvarsall.bat. Any idea what it is? I googled and tried a lot of hacks, but still the same error. Any idea what it is? Meanwhile, I will switch into Fedora for backprojection work !!! Regards On Tuesday, 9 April 2013 22:26:52 UTC+5:30, Stefan van der Walt wrote: > > On Tue, Apr 9, 2013 at 6:51 PM, abid rahman <abidr...(a)gmail.com<javascript:>> > wrote: > >> > > http://opencvpython.blogspot.in/2013/03/histograms-4-back-projection.html. > > > I got hold of it now, thanks. That looks very interesting, as does > > http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.132.5942 > > We look forward to your pull request! Let us know if you need any help. > > Regards > Stéfan >

2 1

Montréal scikit-image overview (in French)
by Stéfan van der Walt 10 Apr '13

10 Apr '13

Hi all Marianne Corvellec gave a very nice overview of scikit-image at the Montréal Python user group: http://youtu.be/fqCu6kkPNxU Have a look! Stéfan

1 0

Annoying gstreamer bug
by Juan Nunez-Iglesias 09 Apr '13

09 Apr '13

Hi All, Consider the following code, saved in a file called test.py: import gst I will bet that few among you would predict this output: $ python test.py -h Usage: test.py [OPTION...] - GStreamer initialization Help Options: -h, --help Show help options --help-all Show all help options --help-gst Show GStreamer Options It seems that someone forgot an if __name__ == '__main__' guard somewhere! Now, this would be none of our concern except for the following corollary: $ python -c 'from skimage.io.collection import alphanumeric_key' -h Usage: -c [OPTION...] - GStreamer initialization Help Options: -h, --help Show help options --help-all Show all help options --help-gst Show GStreamer Options Now, I don't know how to fix this, because it make sense that `io` would import `video` (although `from .video import *` is sloppy). But any suggestions are very very welcome. Juan.

1 0

Re: Histogram backprojection
by Stéfan van der Walt 09 Apr '13

09 Apr '13

On Tue, Apr 9, 2013 at 6:51 PM, abid rahman <abidrahman2(a)gmail.com> wrote: >> > http://opencvpython.blogspot.in/2013/03/histograms-4-back-projection.html. I got hold of it now, thanks. That looks very interesting, as does http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.132.5942 We look forward to your pull request! Let us know if you need any help. Regards Stéfan

1 0

Re: Histogram backprojection
by abid rahman 09 Apr '13

09 Apr '13

Sorry about that. I don't know what is the problem. It works fine for me. Please try base url : www.opencvpython.blogspot.in On Tue, Apr 9, 2013 at 10:18 PM, Stéfan van der Walt <stefan(a)sun.ac.za>wrote: > Hi Abid > > On Tue, Apr 9, 2013 at 6:40 PM, abid rahman <abidrahman2(a)gmail.com> wrote: > > I would like to add a feature on histogram backprojection as proposed by > > Swain & Michael J. in their paper "Indexing via color histograms".. > Earlier, > > I had written a simple article on that ( it was focussed for OpenCV) in > my > > blog : > > > http://opencvpython.blogspot.in/2013/03/histograms-4-back-projection.html. > > For some reason, I get a blank blog entry page when I try to load > that. Can anyone else see it? > > Stéfan > > -- > You received this message because you are subscribed to the Google Groups > "scikit-image" group. > To unsubscribe from this group and stop receiving emails from it, send an > email to scikit-image+unsubscribe(a)googlegroups.com. > For more options, visit https://groups.google.com/groups/opt_out. > > > -- Abid K. opencvpython.blogspot.com

1 0

Re: Histogram backprojection
by Puneeth Chaganti 09 Apr '13

09 Apr '13

On Tue, Apr 9, 2013 at 10:18 PM, Stéfan van der Walt <stefan(a)sun.ac.za> wrote: > Hi Abid > > On Tue, Apr 9, 2013 at 6:40 PM, abid rahman <abidrahman2(a)gmail.com> wrote: >> I would like to add a feature on histogram backprojection as proposed by >> Swain & Michael J. in their paper "Indexing via color histograms".. Earlier, >> I had written a simple article on that ( it was focussed for OpenCV) in my >> blog : >> http://opencvpython.blogspot.in/2013/03/histograms-4-back-projection.html. > > For some reason, I get a blank blog entry page when I try to load > that. Can anyone else see it? It takes a short while (~20s?) to load. But, I can see it.

1 0

Re: GSOC 2013 Proposal
by Stéfan van der Walt 09 Apr '13

09 Apr '13

Hi Siddharth On Tue, Apr 9, 2013 at 11:32 AM, siddharth kherada <siddharthkherada27(a)gmail.com> wrote: > I am currently pursuing MS by Research in the field of image processing at > IIIT- Hyderabad, India. I am interested in working for Scikit-image this > summer in a GSoC project. Thanks for getting in touch. Both these algorithms are cute, and we'll consider PRs for them if they come in, but I don't think they are major underlying parts of the library we'd want to spend a GSoC on. If you are interested in this sort of thing, image deconvolution may be more relevant. For GSoC, we'd like to implement algorithms that will be useful across a wide array of disciplines, if possible. Also, we require that students who would like to participate in GSoC first make other contributions to the project, so be sure to look at the list of tickets and try to fix some of them, at a minimum. Regards Stéfan

1 0

Re: [OT] open-data movies of science experiments
by Marianne Corvellec 07 Apr '13

07 Apr '13

Hi Emmanuelle, On the biology side again (but biophysics is a big thing ;)), I'm fascinated with the OpenSPIM project and their data might be suitable: http://openspim.org/Gallery#OpenSPIM_data I haven't checked though. Marianne

1 0

[OT] open-data movies of science experiments
by Emmanuelle Gouillart 06 Apr '13

06 Apr '13

Hello, sorry for the slightly off-topic e-mail. I'm currently working on a tutorial on image processing with skimage aimed at undergraduate physics teachers. To show them how skimage can be useful for them and their students, I wish to process images taken from a video of a physics experiment (Brownian motion of particles, granular segregation / Brazil nut effect, rising of bubbles in a viscous fluids etc.). However, I'm having a hard time finding open data with a suitable license (such as creative commons). Would you know a website that gathers a large number of open-data educational / scientific movies, and in particular of science experiments? That'd be very helpful. Cheers, Emmanuelle

3 2

Re: imageio progress
by Ralf Gommers 06 Apr '13

06 Apr '13

On Wed, Mar 27, 2013 at 3:33 PM, Almar Klein <almar.klein(a)gmail.com> wrote: > Dear all, > > Several months ago, I decided to take the code for skimage's freeimage > plugin (which was originally written by Zach Pincus) and turn it into a > stand-alone library. I discussed with Zach and a few others how to proceed > with that, and it seemed that time was a limiting factor for many, so I > ended up doing most coding on my own. > > The project is called imageio <http://imageio.readthedocs.org/> and it is > now a plugin-based library with a relatively small core. The freeimage > plugin that started it all currently provides most file formats. Its now at > a stage where I think the core is pretty much finished, and focus can shift > towards different plugins for various scientific formats. But I am not > going to do that alone :) > > I could use some help with: > * People just trying it out. > * People checking the design of the classes<http://imageio.readthedocs.org/en/latest/classes.html>and plugin > system <http://imageio.readthedocs.org/en/latest/plugins.html>. > * If possible, people checking out the code a bit. > * And of course, people writing plugins! > > The way I see it, imageio can become a common backend for reading and > writing images for several libraries (including skimage), or can function > stand-alone. > > Further, I plan on moving it to github to easier attract developers. > > Any help and comments is much appreciated, > Hi Almar, this looks interesting. I'm not really working with images anymore, but was tempted to try imageio out anyway. Installation was smooth, and the 16-bit TIFF image I tried loaded just fine (where PIL still chokes). So in it's current form I suspect it's already quite useful for some people. I haven't looked in enough detail to be able to comment on the classes/plugin design. An option to use imageio.imread/imsave (and/or skimage.io.imread/imsave) as an alternative to PIL from scipy.misc.imread/imshow would make sense to me. Cheers, Ralf

1 0

Re: SciPy 2013 Tutorial
by Emmanuelle Gouillart 06 Apr '13

06 Apr '13

Hi Tony, I've heard that there has been a deadline extension (from people that were also behind schedule for submitting their tutorial :-), so I think it's still time to submit an abstract. Cheers, Emmanuelle On Mon, Apr 01, 2013 at 10:26:07PM -0500, Tony Yu wrote: > Emmanuelle, > Thanks so much for passing along your proposal... Unfortunately, I got > caught up with other things and didn't have time to put together a > proposal for SciPy. :( > -Tony > On Fri, Mar 29, 2013 at 2:51 AM, Emmanuelle Gouillart > <[1]emmanuelle.gouillart(a)nsup.org> wrote: > Hi Tony, > I'm not going to SciPy 2013, but I've agreed to give a 1h30-long > tutorial > on image processing with scikit-image at Euroscipy 2013 (August 21-24). > I'd happy to share course materials if you're interested. > I copy below the abstract I've given to the organizers. > Cheers, > Emmanuelle > Image processing with scikit-image and the SciPy stack > ====================================================== > Numerical image processing involves the manipulation and transformation > of images, often in order to extract information of interest from the > images. NumPy nd-arrays provide an efficient container for image data, > that can therefore be processed using the SciPy toolstack. > In this tutorial, we will first briefly see how simple operations on > images (e.g. cropping, framing) can be performed using NumPy. More > complex image processing operations will be handled principally by the > scikit-image module (occasionally by the ndimage submodule of > SciPy). Compared to other image processing modules for Python, > scikit-image is designed to work transparently with numpy nd-arrays, and > is written in pure Python and some Cython in order to promote > readability > and maintainability. > The different subtopics of this tutorial include: > * input and output of images (file formats...) > * image enhancing / denoising > * image segmentation (separating an image in labeled regions) > * extraction of geometrical features (edges, lines, spheres, > skeleton...) > * extraction of features for image classification > This tutorial will consist mostly of hands-on examples; no mathematical > justification of the algorithms will be given during the tutorial. > Besides the main image processing tasks, the tutorial will demonstrate > how NumPy advanced features (masks, broadcasting, manipulation of > subsets of indices) and scikit-image utilities functions make image > processing easier. The tutorial will also address the visualization of > image processing results (contours, etc.). > On Thu, Mar 28, 2013 at 09:06:04PM -0500, Tony Yu wrote: > > � �I was wondering if anyone has plans to submit a [1]tutorial > proposal for > > � �SciPy 2013. The proposal is due on Monday (Apr. 1st). It'd be great > to > > � �have scikit-image represented there; especially since medical > imaging is > > � �one of the mini-symposia topics. > > � �If no one else is able, I'll probably submit something, but I have > a > > � �feeling that many of you have more (i.e. any) experience teaching > image > > � �processing than I do. > > � �Cheers, > > � �-Tony

3 2

Re: [OT] open-data movies of science experiments
by Emmanuelle Gouillart 06 Apr '13

06 Apr '13

On Sat, Apr 06, 2013 at 02:13:56PM +0200, Nelle Varoquaux wrote: > > Would you know a website that gathers a large number of open-data > > educational / scientific movies, and in particular of science > > experiments? That'd be very helpful. > Would you be interested in live cell imaging ? Well, this is not what I was primarily looking for (my audience will consist mostly of physics teachers) but it could be interesting for other training sessions, or for the biology teachers. Do you know such a website for live cell imaging? Thanks for your help, Emmanuelle

2 1

Re: Coordinate overlap in two labelled images
by Stuart Mumford 04 Apr '13

04 Apr '13

Hello, Thanks for your replies everyone, I am looking through them but this is a reasonably low-priority project. I am sure I will have a follow up question or two soon!! Thanks again Stuart

1 0

Re: Coordinate overlap in two labelled images
by Juan Nunez-Iglesias 03 Apr '13

03 Apr '13

Suggestion number 3. I'll deal with each of three interpretations of your question. Suppose your images are im1 and im2. Now, you might want 1) of the regions in im1, the ones that intersect over at least 1 pixel with regions of im2; 2) the regions that are part of regions in both im1 and im2 (ie the intersection of regions in im1 and im2); or 3) the union of regions of im1 and im2 that have at least one pixel intersection with the other image. The last example assumes the union'ed regions don't touch each other. Then: In [1]: import scipy.ndimage as nd In [2]: im1 = np.array([[0, 0, 0, 0], [1, 1, 0, 2], [0, 0, 0, 2]]) In [3]: im1 Out[3]: array([[0, 0, 0, 0], [1, 1, 0, 2], [0, 0, 0, 2]]) In [4]: im2 = np.array([[0, 1, 0, 0], [0, 1, 0, 0], [0, 1, 0, 0]]) In [5]: im2 Out[5]: array([[0, 1, 0, 0], [0, 1, 0, 0], [0, 1, 0, 0]]) In [6]: intersect = im1.astype(bool) * im2.astype(bool) In [7]: intersect Out[7]: array([[False, False, False, False], [False, True, False, False], [False, False, False, False]], dtype=bool) In [8]: overlap_labels1 = np.unique(im1[intersect]) In [9]: selector = np.zeros(im1.max() + 1, int) In [10]: selector[overlap_labels1] = arange(1, len(overlap_labels1) + 1) In [11]: reg1 = selector[im1] In [12]: reg1 Out[12]: array([[0, 0, 0, 0], [1, 1, 0, 0], [0, 0, 0, 0]]) In [13]: reg2 = nd.label(intersect)[0] In [14]: reg2 Out[14]: array([[0, 0, 0, 0], [0, 1, 0, 0], [0, 0, 0, 0]], dtype=int32) In [15]: overlap_labels2 = np.unique(im2[intersect]) In [16]: selector2 = np.zeros(im2.max() + 1, int) In [17]: selector2[overlap_labels2] = 1 In [18]: nonzero_pixels = reg1 + selector2[im2] In [19]: reg3 = nd.label(nonzero_pixels)[0] In [20]: reg3 Out[20]: array([[0, 1, 0, 0], [1, 1, 0, 0], [0, 1, 0, 0]], dtype=int32) On Wed, Apr 3, 2013 at 5:52 AM, Stéfan van der Walt <stefan(a)sun.ac.za>wrote: > Hi Stuart > > On Tue, Apr 2, 2013 at 7:04 PM, Stuart Mumford <littlemumford(a)gmail.com> > wrote: > > I am using your (amazingly fast) label image routines and regionprops to > extract the coordinates of the regions in two images. I need to filter > these objects out if they do not overlap in both the images, then renumber > all the remaining regions. At the moment I am doing something unspeakable > with nested loops and sets which is pretty damn slow; I was hoping someone > here would have a really clever idea on how to solve this. Failing that > does anyone know how I can find the number of coordinates (output from > regionprops) that overlap in two numpy arrays? > > You can use join_segmentations to find an intersection of your two > label maps. Then, walk through the newly formed regions and check at > each point whether the object exists in both images (that way, you > don't have so many comparisons to do). > > http://scikit-image.org/docs/dev/auto_examples/plot_join_segmentations.html > > If that's not exactly what you had in mind, you can use this code to > find all the connected regions in an image: > > https://github.com/stefanv/lulu/blob/master/lulu/base.pyx#L21 > > These regions are encoded sparsely, and testing their intersection is > very fast (see the algorithm section of this paper: > https://scholar.sun.ac.za/handle/10019.1/4883). > > I hope that helps! > Stéfan > > -- > You received this message because you are subscribed to the Google Groups > "scikit-image" group. > To unsubscribe from this group and stop receiving emails from it, send an > email to scikit-image+unsubscribe(a)googlegroups.com. > For more options, visit https://groups.google.com/groups/opt_out. > > >

1 0

Coordinate overlap in two labelled images
by Stuart Mumford 02 Apr '13

02 Apr '13

Hello, I have a problem I was hoping someone here would be able to help me with! I am using your (amazingly fast) label image routines and regionprops to extract the coordinates of the regions in two images. I need to filter these objects out if they do not overlap in both the images, then renumber all the remaining regions. At the moment I am doing something unspeakable with nested loops and sets which is pretty damn slow; I was hoping someone here would have a really clever idea on how to solve this. Failing that does anyone know how I can find the number of coordinates (output from regionprops) that overlap in two numpy arrays? Thanks a lot Stuart

3 2

Re: SciPy 2013 Tutorial
by Emmanuelle Gouillart 02 Apr '13

02 Apr '13

Hi Tony, I'm not going to SciPy 2013, but I've agreed to give a 1h30-long tutorial on image processing with scikit-image at Euroscipy 2013 (August 21-24). I'd happy to share course materials if you're interested. I copy below the abstract I've given to the organizers. Cheers, Emmanuelle Image processing with scikit-image and the SciPy stack ====================================================== Numerical image processing involves the manipulation and transformation of images, often in order to extract information of interest from the images. NumPy nd-arrays provide an efficient container for image data, that can therefore be processed using the SciPy toolstack. In this tutorial, we will first briefly see how simple operations on images (e.g. cropping, framing) can be performed using NumPy. More complex image processing operations will be handled principally by the scikit-image module (occasionally by the ndimage submodule of SciPy). Compared to other image processing modules for Python, scikit-image is designed to work transparently with numpy nd-arrays, and is written in pure Python and some Cython in order to promote readability and maintainability. The different subtopics of this tutorial include: * input and output of images (file formats...) * image enhancing / denoising * image segmentation (separating an image in labeled regions) * extraction of geometrical features (edges, lines, spheres, skeleton...) * extraction of features for image classification This tutorial will consist mostly of hands-on examples; no mathematical justification of the algorithms will be given during the tutorial. Besides the main image processing tasks, the tutorial will demonstrate how NumPy advanced features (masks, broadcasting, manipulation of subsets of indices) and scikit-image utilities functions make image processing easier. The tutorial will also address the visualization of image processing results (contours, etc.). On Thu, Mar 28, 2013 at 09:06:04PM -0500, Tony Yu wrote: > I was wondering if anyone has plans to submit a [1]tutorial proposal for > SciPy 2013. The proposal is due on Monday (Apr. 1st). It'd be great to > have scikit-image represented there; especially since medical imaging is > one of the mini-symposia topics. > If no one else is able, I'll probably submit something, but I have a > feeling that many of you have more (i.e. any) experience teaching image > processing than I do. > Cheers, > -Tony

3 2

Google Summer of Code 2013
by Stéfan van der Walt 01 Apr '13

01 Apr '13

Hi everyone It looks like we will be able to participate in Google Summer of Code 2013 under the PSF umbrella. If you have any project ideas, please add details here: https://github.com/scikit-image/scikit-image/wiki/GSoC-2013 *Prospective GSoC students: *We follow the same policy as scikit-learn, in that you have to have at least one patch contributed to the project *before*applying for a summer of code project. Regards Stéfan

1 0