Mailman 3 April 2014 - scikit-image

Re: Potential contribution: subpixel cross correlation
by Tony Yu 26 Apr '14

26 Apr '14

> > > On Sat, Apr 26, 2014 at 12:02 PM, Stéfan van der Walt <stefan(a)sun.ac.za>wrote: > >> Hi Mike >> >> On Sat, Apr 26, 2014 at 8:24 PM, Mike Sarahan <msarahan(a)gmail.com> wrote: >> > >> http://www.mathworks.com/matlabcentral/fileexchange/18401-efficient-subpixe… >> >> This sounds interesting, but I'm a bit worried by the license. Most posts on the File Exchange have a BSD license, but this particular contribution has "No BSD License" written next to it. I'm not sure what the protocol is for derivative works, but it's something to keep in mind. Best, -Tony

1 0

Re: Potential contribution: subpixel cross correlation
by Michael Sarahan 26 Apr '14

26 Apr '14

Hi Stefan, As I understand it, it's phase correlation, but with a clever way of upsampling the DFT to achieve subpixel registration. The algorithm does one cross correlation in the frequency domain, finds the maximum, then expands around that maximum to arbitrary (user-specified) precision. My experience with template matching is to use it for finding a smaller image in a larger image. I've used only OpenCV for that, so please forgive my lack of understanding of scikit-image's template match. I use this for finding the shift that registers two equally-sized images. I can't use OpenCV's template matching for that, because it always returns a cross correlation that is the size of the larger image minus the size of the smaller image, which for equally sized images is 0. Based on the docs, it looks like that's also true for scikit-image? Phase correlation works pretty well in these cases, but I've found that this particular algorithm works better than OpenCV's phase correlation, thanks to its approach to upsampling. I'll put things together for a PR, and I'll look forward to comments/revisions... Best, Mike On Sat, Apr 26, 2014 at 12:02 PM, Stéfan van der Walt <stefan(a)sun.ac.za>wrote: > Hi Mike > > On Sat, Apr 26, 2014 at 8:24 PM, Mike Sarahan <msarahan(a)gmail.com> wrote: > > > http://www.mathworks.com/matlabcentral/fileexchange/18401-efficient-subpixe… > > Are you doing phase correlation, or cross-correlation in the image > domain? If so, I wonder if you can build on top of the existing > template matching algorithm to use that as an initial value from which > to start the subpixel search? > > Anyway, always keen to be able to do better alignment, and would love > to build out the registration capabilities in scikit-image. > > Thanks for getting in touch! > 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: Real-time interaction forum
by Stuart Mumford 26 Apr '14

26 Apr '14

Hello, Partly motivated by this conversation, and partly by my need to experiment, I have written a module for the IRC bot we use in SunPy ( https://github.com/embolalia/willie/) so that you can tell it to send you pushover (http://pushover.net) or email notifications when your nick is mentioned in a channel the bot is in. My code so far is here: https://github.com/Cadair/williemodules it is very basic but it works. I would be happy to run a bot and a logger as already mention if you guys decide to set up a IRC channel, so this would at least give you notifications. Stuart On 23 April 2014 23:15, Dan Farmer <dfarmernv(a)gmail.com> wrote: > Another alternative is HipChat: https://www.hipchat.com/ > > Seems the IPython guys are using that (if you go to to ipython.org and > look at the right for "Help Chat Room" ) > > > On Wed, Apr 23, 2014 at 11:58 AM, Stuart Mumford <stuart(a)mumford.me.uk>wrote: > >> Hi, >> >> Ah I see. Unfortunately because of the way IRC works you have to have >> some kind of client connected because the server will not give you >> notifications. >> >> The ZNC server I run is effectively a headless irc client that you can >> connect other clients to. There are lots of other similar type solutions to >> this other than znc. >> >> This, as has been pointed out is the major disadvantage to IRC. >> >> Stuart >> On 23 Apr 2014 19:50, "Vighnesh Birodkar" <vighneshbirodkar(a)gmail.com> >> wrote: >> >>> They are , but Stefan mentioned push notifications . >>> I am assuming that , if your IRC client is off and you'd still want to >>> get a notification about a private message >>> On Apr 24, 2014 12:07 AM, "Stuart Mumford" <stuart(a)mumford.me.uk> wrote: >>> >>>> I don't really understand that, android notifications are generated by >>>> apps on the device. >>>> >>>> Stuart >>>> On 23 Apr 2014 19:11, "Vighnesh Birodkar" <vighneshbirodkar(a)gmail.com> >>>> wrote: >>>> >>>>> Push notifications are linked to your google account. I don't think >>>>> you can send Push notifications without knowing a persons Google >>>>> credentials. So unless the IRC server or a Bot knows your Google >>>>> credentials, push notifications can be possible when offline. >>>>> >>>>> On Wednesday, April 23, 2014 11:07:36 PM UTC+5:30, Vighnesh Birodkar >>>>> wrote: >>>>>> >>>>>> Yeah >>>>>> >>>>>> I will test out a few other IRC clients and get back to this. >>>>>> >>>>>> On Wednesday, April 23, 2014 11:01:32 PM UTC+5:30, Stuart Mumford >>>>>> wrote: >>>>>>> >>>>>>> Hello, >>>>>>> >>>>>>> > This seems to do it >>>>>>> > https://play.google.com/store/apps/details?id=fi.iki.murgo. >>>>>>> irssinotifier >>>>>>> >>>>>>> I presume that needs a constant irssi session running somewhere? It >>>>>>> is basically a less complex znc type arrangement. >>>>>>> >>>>>>> Stuart >>>>>>> >>>>>> -- >>>>> 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. >>>>> >>>> -- >>>> You received this message because you are subscribed to a topic in the >>>> Google Groups "scikit-image" group. >>>> To unsubscribe from this topic, visit >>>> https://groups.google.com/d/topic/scikit-image/bD5WRCjr2FY/unsubscribe. >>>> To unsubscribe from this group and all its topics, send an email to >>>> scikit-image+unsubscribe(a)googlegroups.com. >>>> For more options, visit https://groups.google.com/d/optout. >>>> >>> -- >>> 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. >>> >> -- >> 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. >> > > -- > 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

Potential contribution: subpixel cross correlation
by Mike Sarahan 26 Apr '14

26 Apr '14

Hi there, I've implemented this code in Python for one of my projects: http://www.mathworks.com/matlabcentral/fileexchange/18401-efficient-subpixe… It seems to me like this might fit in scikits-image. If you guys would like, I'll clean up my code and do a PR. I'm not sure what the general use cases would be, though - for me, all I care about is the subpixel registration value (how much to shift the images so that they are aligned). Would anyone have any other use cases for this sort of thing? I see this as not quite the same as template matching, but please correct me if I'm wrong. Best, Michael

3 2

Getting distance and cost from an MCP
by David Orme 25 Apr '14

25 Apr '14

Hi, I'm running some road building simulations in Python across a cost surface. What I'd like to be able to do is to get both the cost of paths to a set of starting points and the actual physical distance that those paths would cover. I can easily find the cumulative costs of paths using MCP_Geometric from the graph module but I can't figure out how to get a surface of cumulative distance along the least cost paths. I could iterate over each cell in the cost array in turn to get the cheapest path and fetch the distance using pythogoras on the steps, but my intuition from the documentation is that the MCP_Flexible class in 0.10dev - particularly the travel_cost method - might allow me to get at this more quickly. In the toy example, below I'd like to find costs from a start point at [0,1] in the cost array. Most of the array is expensive to traverse, but the valley means that there are big differences in the length of the cheapest road needed to get between points. If the entries in the cost array represent cells with a resolution of 250 metres across, the path from [0,1] to [0,5] has a cost of 32.0 along a road of length 1.5 km (4 steps E). In contrast, the path from [0,1] to [6,7] has a cost of 35.04 along a road of length 6.21 km (12 steps S - 3km, 1 step SE - 0.354km, 4 steps E - 1km, 1 step NE - 0.354km, and 6 steps N - 1.5km). Any suggestions would be very welcome. Cheers, David import numpy from skimage import graph cost = numpy.array([[9,1,9,9,9,9,9,9,9], [9,1,9,9,9,9,9,9,9], [9,2,9,9,9,9,9,9,9], [9,1,9,9,9,9,9,9,9], [9,1,9,9,9,9,9,9,9], [9,1,9,9,9,9,9,9,9], [9,3,9,9,9,9,9,1,9], [9,1,9,9,9,9,9,1,9], [9,1,9,9,9,9,9,2,9], [9,1,9,9,9,9,9,1,9], [9,2,9,9,9,9,9,1,9], [9,1,9,9,9,9,9,3,9], [9,1,9,9,9,9,9,1,9], [9,1,2,1,1,3,1,1,9], [9,9,9,9,9,9,9,9,9]], dtype='float') costMCP = graph.MCP_Geometric(cost, fully_connected=True) cumcost, trace = costMCP.find_costs([(0,1)])

2 2

Re: Real-time interaction forum
by Dan Farmer 24 Apr '14

24 Apr '14

Another alternative is HipChat: https://www.hipchat.com/ Seems the IPython guys are using that (if you go to to ipython.org and look at the right for "Help Chat Room" ) On Wed, Apr 23, 2014 at 11:58 AM, Stuart Mumford <stuart(a)mumford.me.uk>wrote: > Hi, > > Ah I see. Unfortunately because of the way IRC works you have to have some > kind of client connected because the server will not give you notifications. > > The ZNC server I run is effectively a headless irc client that you can > connect other clients to. There are lots of other similar type solutions to > this other than znc. > > This, as has been pointed out is the major disadvantage to IRC. > > Stuart > On 23 Apr 2014 19:50, "Vighnesh Birodkar" <vighneshbirodkar(a)gmail.com> > wrote: > >> They are , but Stefan mentioned push notifications . >> I am assuming that , if your IRC client is off and you'd still want to >> get a notification about a private message >> On Apr 24, 2014 12:07 AM, "Stuart Mumford" <stuart(a)mumford.me.uk> wrote: >> >>> I don't really understand that, android notifications are generated by >>> apps on the device. >>> >>> Stuart >>> On 23 Apr 2014 19:11, "Vighnesh Birodkar" <vighneshbirodkar(a)gmail.com> >>> wrote: >>> >>>> Push notifications are linked to your google account. I don't think you >>>> can send Push notifications without knowing a persons Google credentials. >>>> So unless the IRC server or a Bot knows your Google credentials, push >>>> notifications can be possible when offline. >>>> >>>> On Wednesday, April 23, 2014 11:07:36 PM UTC+5:30, Vighnesh Birodkar >>>> wrote: >>>>> >>>>> Yeah >>>>> >>>>> I will test out a few other IRC clients and get back to this. >>>>> >>>>> On Wednesday, April 23, 2014 11:01:32 PM UTC+5:30, Stuart Mumford >>>>> wrote: >>>>>> >>>>>> Hello, >>>>>> >>>>>> > This seems to do it >>>>>> > https://play.google.com/store/apps/details?id=fi.iki.murgo. >>>>>> irssinotifier >>>>>> >>>>>> I presume that needs a constant irssi session running somewhere? It >>>>>> is basically a less complex znc type arrangement. >>>>>> >>>>>> Stuart >>>>>> >>>>> -- >>>> 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. >>>> >>> -- >>> You received this message because you are subscribed to a topic in the >>> Google Groups "scikit-image" group. >>> To unsubscribe from this topic, visit >>> https://groups.google.com/d/topic/scikit-image/bD5WRCjr2FY/unsubscribe. >>> To unsubscribe from this group and all its topics, send an email to >>> scikit-image+unsubscribe(a)googlegroups.com. >>> For more options, visit https://groups.google.com/d/optout. >>> >> -- >> 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. >> > -- > 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: Real-time interaction forum
by Stuart Mumford 23 Apr '14

23 Apr '14

Hi, Ah I see. Unfortunately because of the way IRC works you have to have some kind of client connected because the server will not give you notifications. The ZNC server I run is effectively a headless irc client that you can connect other clients to. There are lots of other similar type solutions to this other than znc. This, as has been pointed out is the major disadvantage to IRC. Stuart On 23 Apr 2014 19:50, "Vighnesh Birodkar" <vighneshbirodkar(a)gmail.com> wrote: > They are , but Stefan mentioned push notifications . > I am assuming that , if your IRC client is off and you'd still want to get > a notification about a private message > On Apr 24, 2014 12:07 AM, "Stuart Mumford" <stuart(a)mumford.me.uk> wrote: > >> I don't really understand that, android notifications are generated by >> apps on the device. >> >> Stuart >> On 23 Apr 2014 19:11, "Vighnesh Birodkar" <vighneshbirodkar(a)gmail.com> >> wrote: >> >>> Push notifications are linked to your google account. I don't think you >>> can send Push notifications without knowing a persons Google credentials. >>> So unless the IRC server or a Bot knows your Google credentials, push >>> notifications can be possible when offline. >>> >>> On Wednesday, April 23, 2014 11:07:36 PM UTC+5:30, Vighnesh Birodkar >>> wrote: >>>> >>>> Yeah >>>> >>>> I will test out a few other IRC clients and get back to this. >>>> >>>> On Wednesday, April 23, 2014 11:01:32 PM UTC+5:30, Stuart Mumford wrote: >>>>> >>>>> Hello, >>>>> >>>>> > This seems to do it >>>>> > https://play.google.com/store/apps/details?id=fi.iki.murgo. >>>>> irssinotifier >>>>> >>>>> I presume that needs a constant irssi session running somewhere? It is >>>>> basically a less complex znc type arrangement. >>>>> >>>>> Stuart >>>>> >>>> -- >>> 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. >>> >> -- >> You received this message because you are subscribed to a topic in the >> Google Groups "scikit-image" group. >> To unsubscribe from this topic, visit >> https://groups.google.com/d/topic/scikit-image/bD5WRCjr2FY/unsubscribe. >> To unsubscribe from this group and all its topics, send an email to >> scikit-image+unsubscribe(a)googlegroups.com. >> For more options, visit https://groups.google.com/d/optout. >> > -- > 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: Real-time interaction forum
by Vighnesh Birodkar 23 Apr '14

23 Apr '14

They are , but Stefan mentioned push notifications . I am assuming that , if your IRC client is off and you'd still want to get a notification about a private message On Apr 24, 2014 12:07 AM, "Stuart Mumford" <stuart(a)mumford.me.uk> wrote: > I don't really understand that, android notifications are generated by > apps on the device. > > Stuart > On 23 Apr 2014 19:11, "Vighnesh Birodkar" <vighneshbirodkar(a)gmail.com> > wrote: > >> Push notifications are linked to your google account. I don't think you >> can send Push notifications without knowing a persons Google credentials. >> So unless the IRC server or a Bot knows your Google credentials, push >> notifications can be possible when offline. >> >> On Wednesday, April 23, 2014 11:07:36 PM UTC+5:30, Vighnesh Birodkar >> wrote: >>> >>> Yeah >>> >>> I will test out a few other IRC clients and get back to this. >>> >>> On Wednesday, April 23, 2014 11:01:32 PM UTC+5:30, Stuart Mumford wrote: >>>> >>>> Hello, >>>> >>>> > This seems to do it >>>> > https://play.google.com/store/apps/details?id=fi.iki.murgo. >>>> irssinotifier >>>> >>>> I presume that needs a constant irssi session running somewhere? It is >>>> basically a less complex znc type arrangement. >>>> >>>> Stuart >>>> >>> -- >> 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. >> > -- > You received this message because you are subscribed to a topic in the > Google Groups "scikit-image" group. > To unsubscribe from this topic, visit > https://groups.google.com/d/topic/scikit-image/bD5WRCjr2FY/unsubscribe. > To unsubscribe from this group and all its topics, send an email to > scikit-image+unsubscribe(a)googlegroups.com. > For more options, visit https://groups.google.com/d/optout. >

1 0

Re: Real-time interaction forum
by Vighnesh Birodkar 23 Apr '14

23 Apr '14

Yeah I will test out a few other IRC clients and get back to this. On Wednesday, April 23, 2014 11:01:32 PM UTC+5:30, Stuart Mumford wrote: > > Hello, > > > This seems to do it > > https://play.google.com/store/apps/details?id=fi.iki.murgo.irssinotifier > > I presume that needs a constant irssi session running somewhere? It is > basically a less complex znc type arrangement. > > Stuart >

2 2

Re: Real-time interaction forum
by Vighnesh Birodkar 23 Apr '14

23 Apr '14

This seems to do it https://play.google.com/store/apps/details?id=fi.iki.murgo.irssinotifier And logging can be set up via a bot http://umashanthi.blogspot.in/2010/10/create-your-irc-channel-and-make-logs… On Wednesday, April 23, 2014 10:43:24 PM UTC+5:30, Stefan van der Walt wrote: > > > On 23 Apr 2014 16:44, "Vighnesh Birodkar" <vighnesh...(a)gmail.com<javascript:>> > wrote: > > > > Should I try setting up #scikitimage on freenode ? > > Until we have a good way to record what happens in #skimage / > #scikit-image, I don't think it makes much sense (for me, personally, at > least). Also, one would have to have a way of getting push notifications on > Android and iPhone on usage of your nick. Does any such solution exist out > there? > > To be clear, I have no objection to others using IRC--these are just the > factors that would compel me to do so myself. > > Stéfan >

2 1