Mailman 3 July 2014 - scikit-image

Re: Non-linear distance based filter
by Eduardo Henrique Arnold 16 Jul '14

16 Jul '14

Yes, setting the vertical sampling value to a high value (i.e. 10000) did the trick. It makes sense since it is going to sample the the next pixel in a vertical position that does not exist in the image. With the right distance matrix I could use Josh's idea, as he suggested on the first reply. That was simpler than implementing something with several loops. Thank you both, those were very quick and effective answers. :) On Wednesday, July 16, 2014 1:16:34 AM UTC+1, Juan Nunez-Iglesias wrote: > > @JDWarner, actually, instead of iterating, setting one of the sampling > dimensions to a very large number should produce the desired output, no? =) > > > On Tue, Jul 15, 2014 at 11:13 AM, Josh Warner <silvertr...(a)gmail.com > <javascript:>> wrote: > >> I don’t believe sampling works like this. The sampling= kwarg tells the >> algorithm the spacing of each pixel/voxel in each spatial dimension. It >> then calculates the distance transform. So if you put a zero in, it will >> traverse in that dimension with a distance cost of zero for each step. This >> will generate a result that has mostly spurious zero distances, rather than >> select a particular axis to calculate the distance transform along. >> >> To calculate the correct distance function along a specific axis probably >> requires repeated calls along that axis; since scipy.ndimage is by >> definition n-dimensional it should be able to efficiently handle rank-1 >> arrays without issue. >> >> I use the super handy Chrome extension “Markdown Here” to write in >> Markdown and then have my free entry fields converted to pretty HTML prior >> to posting. At this point I pretty much think in Markdown… >> >> Josh >> >> -- >> 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/d/optout. >> > >

1 0

Non-linear distance based filter
by Eduardo Henrique Arnold 15 Jul '14

15 Jul '14

Hi there. I need to implement a filter that takes each white pixel in a binary image and look for the distance between the top and bottom closest background (black) pixels. If this distance is smaller than a threshold T all the pixels between this two background pixels should be set to 0 (background pixels). I know I could iterate through all the image columns and pixels, but I think this would yield a poor performance. Is there any suggestion of how I could develop this filter in a more efficient way? Thanks.

3 4

Re: Potential contribution: subpixel cross correlation
by jim＠rybarski.com 12 Jul '14

12 Jul '14

His contact info is all here: http://www.optics.rochester.edu/workgroups/fienup/Professor.htm On Saturday, July 12, 2014 3:55:37 PM UTC-5, Stefan van der Walt wrote: > > On Sat, Jul 12, 2014 at 10:11 PM, <j...(a)rybarski.com <javascript:>> > wrote: > > Where are things at with this? The method works incredibly well and my > > current project is pretty much impossible without it. > > If anyone can provide me the original author's contact information, > I'd be happy to contact him and ask permission. I'll be traveling for > a few days, so I may not be able to respond immediately. > > Regards > Stéfan >

1 0

Re: Potential contribution: subpixel cross correlation
by jim＠rybarski.com 12 Jul '14

12 Jul '14

Where are things at with this? The method works incredibly well and my current project is pretty much impossible without it. On Tuesday, May 13, 2014 1:16:02 AM UTC-5, Stefan van der Walt wrote: > > Hi Michael > > On Tue, May 13, 2014 at 6:02 AM, Michael Sarahan <msar...(a)gmail.com > <javascript:>> wrote: > > I wrote to Prof. Fienup last Tuesday, but haven't heard back. The work > is > > published, and to the best of my knowledge unpatented (Google patent > search > > turns up nothing.) Let me know if you'd still be up for a PR. > > I would prefer to get Prof. Fienup's approval of the work. However, > if that cannot be obtained we can still publish code if we can show > that it was derived purely from the published information (not from > the "tainted" code the current version is based on). > > Regards > Stéfan >

2 1

Re: Make Pillow a requirement
by Stéfan van der Walt 11 Jul '14

11 Jul '14

I imagine it would take only one email to Illan Schnell to get it included.

1 0

Re: Make Pillow a requirement
by Juan Nunez-Iglesias 11 Jul '14

11 Jul '14

Hey Stefan, Pillow is in conda, imread is not. My experimentation over the last 15 minutes suggests that it will be much more of a challenge to install imread than pillow. (I'm also not sure whether imread supports Python 3 and that's why things are failing.) I also just saw @jjhelmus's email so maybe PIL should throw its hat into the ring... On Fri, Jul 11, 2014 at 10:44 AM, Stéfan van der Walt <stefan(a)sun.ac.za> wrote: > Hi Juan > > On Fri, Jul 11, 2014 at 5:32 PM, Juan Nunez-Iglesias <jni.soma(a)gmail.com> > wrote: > > Cheap solution: make Pillow a requirement. > > My suggestion last time was to make 'imread' a requirement--what do you > think? > > 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/d/optout. >

1 0

Re: Make Pillow a requirement
by Jonathan Helmus 11 Jul '14

11 Jul '14

On 7/11/14, 10:44 AM, Stéfan van der Walt wrote: > Hi Juan > > On Fri, Jul 11, 2014 at 5:32 PM, Juan Nunez-Iglesias <jni.soma(a)gmail.com> wrote: >> Cheap solution: make Pillow a requirement. > My suggestion last time was to make 'imread' a requirement--what do you think? > > Stéfan > From my experience, imread is not as widely available as PIL and pillow. PIL and pillow are both provided in the Anaconda repositories but imread is not. I set up a test environment on OS X and Windows with the current master and tried running the skimage unit tests with the required dependencies as well as with pillow and PIL, results are: Only Required Dependencies: 2 Errors (Win); 2 Errors/2 Fails (OS X) "" + pillow: 106 Errors, 4 Fails (Win); 1 Error (OS X) "" + PIL: tests pass (Win); 1 Error (OS X). All packages were from Anaconda's repo, 32 bit Windows, OS X 10.9. - Jonathan Helmus

1 0

Re: Make Pillow a requirement
by Stéfan van der Walt 11 Jul '14

11 Jul '14

Hi Juan On Fri, Jul 11, 2014 at 5:32 PM, Juan Nunez-Iglesias <jni.soma(a)gmail.com> wrote: > Cheap solution: make Pillow a requirement. My suggestion last time was to make 'imread' a requirement--what do you think? Stéfan

1 0

Re: Make Pillow a requirement
by François Boulogne 11 Jul '14

11 Jul '14

Le 11/07/2014 11:32, Juan Nunez-Iglesias a écrit : > @jjhelmus and I are having a discussion regarding IO and how tests > fail if you don't have any of Pillow, PyQt, freeimage, etc, none of > which are actually a requirement. As one of the sprint topics is > fixing up IO, this would be a great thing to fix. > > Cheap solution: make Pillow a requirement. > > I remember I submitted a PR (accepted) to add Pillow in DEPENDS.txt. But it's not un requierements.txt... I follow the Pillow project, and I believe it's definitely the solution to invest in. :) Best, -- François Boulogne. http://www.sciunto.org GPG: 32D5F22F

1 0

Re: Make Pillow a requirement
by Juan Nunez-Iglesias 11 Jul '14

11 Jul '14

In a related note, we just found that Travis *installs FreeImage and Pillow for ALL tests*!!! This is completely ludicrous given that these are not requirements!!! On Fri, Jul 11, 2014 at 10:32 AM, Juan Nunez-Iglesias <jni.soma(a)gmail.com> wrote: > @jjhelmus and I are having a discussion regarding IO and how tests fail if > you don't have any of Pillow, PyQt, freeimage, etc, none of which are > actually a requirement. As one of the sprint topics is fixing up IO, this > would be a great thing to fix. > > Cheap solution: make Pillow a requirement. > > Expensive solution: write/pilfer our own IO modules, however inefficient > they may be. (e.g. tifffile.py for tifs, maybe similar PNG) > > I don't see any *other* solution. Making mpl a requirement is not useful > since their IO is also broken (e.g. images convert to float by default). > > Juan. >

1 0