Mailman 3 April 2013 - scikit-image

Re: Describing superpixels
by Brickle Macho April 22, 2013

April 22, 2013

Hi Juan, Thanks. Makes more sense now when I inspect the mask. Michael. -- On 22/04/13 4:17 PM, Juan Nunez-Iglesias wrote: > Hi Brickle, > > Cool problem. =) iirc the return type of these algorithms is an M x N > integer-type numpy array (where the input image is an M x N x 3 numpy > array). Every pixel with the same value belongs in the same > superpixel. So, all pixels with value 1 make up the 1st superpixel, > all pixels with value 2 make up the 2nd, and so on until the nth > superpixel. > > Does that answer your question? > > Juan. > > > On Mon, Apr 22, 2013 at 6:02 PM, Brickle Macho <bricklemacho(a)gmail.com > <mailto:bricklemacho@gmail.com>> wrote: > > Hi, > > I am new to python and image processing, which may be my problem, > but I don't understand how to interpret/use the integer mask > indicating segment labels output from the SLIC and Quickshift > algorithms. > > I have a RGB-D image. Using only RGB I segment the image into > superpixels using SLIC and Quickshift algorithms provided in > scikit-image. I am trying visit each superpixel, calculate some > depth features for each superpixel. Specifically I want to > calculate the surface normal of the superpixel and the average > angular difference with the neighbouring superpixels. > Eventually I plan to combine the superpixels based on these depth > features. > > Could someone explain the segment_mask format/structure and how I > should use the mask? > > Thanks in advance. > > Brickle. > -- > > -- > 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 > <mailto:scikit-image%2Bunsubscribe@googlegroups.com>. > For more options, visit https://groups.google.com/groups/opt_out. > > > > -- > 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

Describing superpixels
by Brickle Macho April 22, 2013

April 22, 2013

Hi, I am new to python and image processing, which may be my problem, but I don't understand how to interpret/use the integer mask indicating segment labels output from the SLIC and Quickshift algorithms. I have a RGB-D image. Using only RGB I segment the image into superpixels using SLIC and Quickshift algorithms provided in scikit-image. I am trying visit each superpixel, calculate some depth features for each superpixel. Specifically I want to calculate the surface normal of the superpixel and the average angular difference with the neighbouring superpixels. Eventually I plan to combine the superpixels based on these depth features. Could someone explain the segment_mask format/structure and how I should use the mask? Thanks in advance. Brickle. --

3 2

Re: Morphological techniques
by Chintak Sheth April 21, 2013

April 21, 2013

Hi, So I have been able to implement the convex_hull_object function. The implementation is in the "convex_hull.py" file. I have also included two tests in "test_convex_hull.py" file itself. I have submitted a PR for the same. https://github.com/scikit-image/scikit-image/pull/522 Johannes and Stefan, could you'll review it please ? Also if there is a faster algorithm for the same ? I was also thinking about having just the current convex_hull_image() with an additional parameter, method. When method='union', i get the convex hull of the entire image, and if method='object', i get the convex hull of individual objects. Though this approach of defining an another function and giving an additional option of returning convex hull of individual segmented objects, seemed interesting. Also helps keep the code clean. Any comments on this ? Thanks, Chintak On Mon, Apr 22, 2013 at 1:56 AM, Chintak Sheth <chintaksheth(a)gmail.com>wrote: > Hi, > > So I have been > > Chintak > On Apr 15, 2013 11:40 PM, "Stéfan van der Walt" <stefan(a)sun.ac.za> wrote: > >> Hi Chintak >> >> On Mon, Apr 15, 2013 at 7:55 PM, Chintak Sheth <chintaksheth(a)gmail.com> >> wrote: >> > I am really excited about being a part of this community and have been >> > pretty fascinated by morphological techniques in particular. I would >> really >> > want to contribute something in this direction. Can someone please >> point me >> > to possible bugs or algorithms in this field that need implementations ? >> >> Welcome! There are several places to start, e.g. we need a >> convex_hull_objects function, as well as adapting rank filters to >> handle int16 correctly (not as 12-bit integers). >> >> Otherwise, I'd also keep an eye on IEEE Transactions on Image >> Processing, and what other leading image processing packages >> implement. >> >> There are some pretty interesting segmentation and feature detection >> applications of morphology too. The sky's the limit :) >> >> Cheers >> Stéfan >> >> P.S. There are plenty of bugs to fix too: >> https://github.com/scikit-image/scikit-image >> >> -- >> 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: Morphological techniques
by Chintak Sheth April 21, 2013

April 21, 2013

Hi, So I have been Chintak On Apr 15, 2013 11:40 PM, "Stéfan van der Walt" <stefan(a)sun.ac.za> wrote: > Hi Chintak > > On Mon, Apr 15, 2013 at 7:55 PM, Chintak Sheth <chintaksheth(a)gmail.com> > wrote: > > I am really excited about being a part of this community and have been > > pretty fascinated by morphological techniques in particular. I would > really > > want to contribute something in this direction. Can someone please point > me > > to possible bugs or algorithms in this field that need implementations ? > > Welcome! There are several places to start, e.g. we need a > convex_hull_objects function, as well as adapting rank filters to > handle int16 correctly (not as 12-bit integers). > > Otherwise, I'd also keep an eye on IEEE Transactions on Image > Processing, and what other leading image processing packages > implement. > > There are some pretty interesting segmentation and feature detection > applications of morphology too. The sky's the limit :) > > Cheers > Stéfan > > P.S. There are plenty of bugs to fix too: > https://github.com/scikit-image/scikit-image > > -- > 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: Challenge: estimate a blurring kernel
by abid rahman April 21, 2013

April 21, 2013

Hi, That is really mind blowing !!! Here is more better video of that presentation : http://www.youtube.com/watch?v=LBIJj13H7uk Do you consider it as a GSoC project ? Isn't it patented ? Regards Abid K On Friday, 14 October 2011 04:06:56 UTC+5:30, Stefan van der Walt wrote: > > Hi all, > > At a recent Adobe MAX conference, people went crazy over their > deblurring demo [1]. I want it in scikits-image! The authors were > even kind enough to leave us some clues as to how [2]. > > Who is up for the challenge? > > Stéfan > > [1] http://www.youtube.com/watch?v=xxjiQoTp864t > [2] > http://people.csail.mit.edu/sparis/publi/2011/cvpr_radon/Cho_11_Blur_Kernel… > >

2 1

Re: Edge detectors in filter package
by Johannes Schönberger April 21, 2013

April 21, 2013

Basically, "filter" is a very universal term that may be used for lots of image processing related tasks… On the one hand, edge operators probably fit well into the feature package, on the other hand Tony just used the very common term > edge filters So, I am +-0 on this… Johannes Schönberger Am 21.04.2013 um 17:56 schrieb Tony Yu <tsyu80(a)gmail.com>: > > On Sun, Apr 21, 2013 at 8:15 AM, Ankit Agrawal <aaaagrawal(a)gmail.com> wrote: > Hi everyone, > > I would like to know the motivation behind keeping edges.py(edge detectors) in filters package. Based on the Image Processing terminology , I think features package is a more appropriate place for it. Thank you. > > > I've always thought that edge filters belonged in the features package, too. There's always the question of whether deprecating the current behavior is worth the trouble. I'm +1, despite the pain of deprecation. > > Cheers, > -Tony > > -- > 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. > >

2 1

Edge detectors in filter package
by Ankit Agrawal April 21, 2013

April 21, 2013

Hi everyone, I would like to know the motivation behind keeping edges.py(edge detectors) in filters package. Based on the Image Processing terminology , I think features package is a more appropriate place for it. Thank you. Regards, Ankit Agrawal, Communication and Signal Processing, IIT Bombay.

2 1

Re: imageio progress
by Juan Nunez-Iglesias April 21, 2013

April 21, 2013

I wrote a lot of image io stuff for my own projects before I became aware of skimage. I spent some time thinking about how people could add and maintain their own modules, and came up with the following: io/ __init__.py tiff.py read() write() png.py read() write() etc. This isn't coded, it was just on my todo. Adding new formats is then a matter of dumping in a file containing read() and write() functions, and importing that in the __init__.py file. This results in very clean syntax: io.png.read() There could also be io.read() and io.write() defined in __init__ or some generic name, e.g. io.auto.read(), that would call the right format based on filename. Thoughts? On Fri, Apr 19, 2013 at 10:47 PM, Stéfan van der Walt <stefan(a)sun.ac.za>wrote: > Hi Almar > > On Wed, Mar 27, 2013 at 4:33 PM, Almar Klein <almar.klein(a)gmail.com> > wrote: > > The project is called imageio 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 :) > > Is there a way to add plugins by simply "dumping" plugin files into > place? This was part of the thinking when we designed the skimage.io > plugin infrastructure (so it would be easy for, e.g., Debian to add > new formats by simply unpacking into the right location). > > I am not entirely happy with the complexity of the io module in > skimage, and would love to hear ideas about how it can be simplified, > perhaps integrating with imageio. > > 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

Test failures in graph subpackage
by Tony Yu April 20, 2013

April 20, 2013

Can anyone reproduce the test failures that the Travis buildbot is raising on master: https://travis-ci.org/scikit-image/scikit-image/jobs/6472929#L2274 The tests pass on my setup, so I can't really figure out what's wrong. Thanks, -Tony

1 0

Re: multi-page TIFF writer
by Tony Yu April 20, 2013

April 20, 2013

On Thu, Apr 18, 2013 at 10:47 AM, Stéfan van der Walt <stefan(a)sun.ac.za>wrote: > On Thu, Apr 18, 2013 at 5:12 PM, Zachary Pincus <zachary.pincus(a)yale.edu> > wrote: > > Also, BTW, I get the below test failure when I have matplotlib but no > PIL installed. I think test_imread_url or the test right before it must be > making some incorrect assumptions. (If I re-run the tests after this > failure, then also all of the tifffile tests fail, so it seems that some > test isn't doing proper teardown of the image loading plugins?) > > Ah, I see what's happening. This one might be a job for @tonysyu -- > Tony, this looks weird because matplotlib is supposed to be able to > handle pngs (and you are reading in a png in that test). So perhaps a > bug in matplotlib when providing a "file-like object"? > > Fixed in PR 515: https://github.com/scikit-image/scikit-image/pull/515 Unfortunately, I can't reproduce Zach's issue with the test failure leaking into into the tifffile tests. Best, -Tony

1 0