Mailman 3 January 2015 - scikit-image

Question on tests
by Kevin Keraudren 25 Jan '15

25 Jan '15

Hi, I have two questions related to tests in scikit-image: 1) is there a way to run just a specific test file, as "make test" runs all tests? I tried: python skimage/feature/tests/test_peak.py but it dies with "NameError: name 'version' is not defined" 2) What about using conditional skips for tests in skimage/viewer/tests otherwise these tests fail with "cannot connect to X server" if we are using ssh without -X? StackOverflow suggests checking os.environ['DISPLAY'] (http://stackoverflow.com/questions/12122671/how-to-determine-from-a-python-…) Kind regards, Kevin

2 2

border artefacts with skimage.exposure.equalize_adapthist
by Kai Wiechen 24 Jan '15

24 Jan '15

<https://lh3.googleusercontent.com/-JDkNXe7k44s/VMOE-xwgPXI/AAAAAAAABLA/STWx…> <https://lh3.googleusercontent.com/-sgqpXIXHhF0/VMOFDbcK8sI/AAAAAAAABLI/iHcF…> Hello, I am new to scikit-image (and python too), but found the combination of scikit-image implementation of algorithms together with ndarrays extremely powerful. For now, I am trying to find nuclei in microscopic images (color deconvolution, find seeds with blob_log) This works fairly well but due to low contrast some nuclei are not detected. nuclei_clahe = equalize_adapthist(nuclei) gives much better local contrast but introduces linear artefacts at the test image borders. Is it possible to avoid/reduce this? Best regards, Kai

3 2

Release timeline, v0.11
by Stéfan van der Walt 21 Jan '15

21 Jan '15

Hi everyone We didn't quite hit New Year for the 0.11 release, but I'd like to get it out the door as soon as possible. I've triaged most of the milestone tickets, so please help wherever you can with reviewing and closing up any remaining PRs. Thanks! Stéfan

1 0

minimum cost MCP with IPython Parallel
by OOM 19 Jan '15

19 Jan '15

Hi, I am trying to use graph.MCP_Geometric(costs) on a cluster. The script and methods I use work on my personal computer which has 64GB memory. I am now trying to implement the same code on a cluster using IPython parallel. Here's the problem, each node/task only has about 20GB memory so they cannot even complete the mcp=graph.MCP_Geometric(costs) command. My cost data has shape (12837, 43345). I am trying IPython parallel since I think it should allow me to overcome the lack of memory in each node. E.g. If I define a parallel function like @dview.parallel(block=True) def costsnx(costs): mcp=graph.MCP_Geometric(costs) return mcp and run it mcpp=costsnx(costs) I get a list with mcp objects like <skimage.graph._mcp.MCP_Geometric at 0x2d6fdd0>, one for each process I created. So in principle this would solve the memory problem, but the individual mcp objects do not seem to work e.g. executing mcpp[30].find_costs([[0,0]]) generates TypeError: object of type 'NoneType' has no len() Also, there is no way to join them to recreate the real graph. I tried also including the mcp.find_costs(location) command in the parallel function, but it cannot find the location. I was thinking that perhaps there would be a way to use the intermediate steps of the graph.MCP_Geometric function to get the indices or something that could be constructed in parallel and then joined back into a unique element to compute minimum_costs. Any ideas on how this could be implemented? Or any idea on how to tackle the problem at all? I appreciate any help or pointers. Thanks OOM

1 0

skimage.feature.peak_local_max behaviour
by Sigmund 16 Jan '15

16 Jan '15

Hi, My first post a few hours ago somehow didn't appear. It was with examples but I trink I can make my paint without. If two peaks with the same value are within the min_distance range they are not detectes as too closer and reduced to one. Just use the provided examples in the doc and change the peaks to the same value.

4 5

Re: Segmentation of multi-channel(>3) images
by Juan Nunez-Iglesias 15 Jan '15

15 Jan '15

So... pass convert2lab=False! ;) On Thu, Jan 15, 2015 at 10:00 PM, Ankit Agrawal <aaaagrawal(a)gmail.com> wrote: > Hi Juan, > Apparently the docs of slic are updated with `multi-channel = True`, > but when I tried it for my case, it returned this error - > In [24]: ps = seg.slic(pn, n_segments=100, multichannel=True) > --------------------------------------------------------------------------- > ValueError Traceback (most recent call last) > <ipython-input-24-68d7c9154670> in <module>() > ----> 1 ps = seg.slic(pn, n_segments=100, multichannel=True) > /usr/local/lib/python2.7/dist-packages/skimage/segmentation/slic_superpixels.pyc > in slic(image, n_segments, compactness, max_iter, sigma, spacing, > multichannel, convert2lab, enforce_connectivity, min_size_factor, > max_size_factor, slic_zero) > 142 if convert2lab and multichannel: > 143 if image.shape[3] != 3: > --> 144 raise ValueError("Lab colorspace conversion requires a > RGB image.") > 145 image = rgb2lab(image) > 146 > ValueError: Lab colorspace conversion requires a RGB image. > Ankit Agrawal, > IIT Bombay. > On Thu, Jan 15, 2015 at 4:26 PM, Juan Nunez-Iglesias <jni.soma(a)gmail.com> > wrote: >> Hi Ankit, >> >> iirc SLIC was updated to work for any number of channels with >> `multichannel=True`! Looks like the docstring needs updating! Feel free to >> submit a PR but in the meantime, give it a shot! =) >> >> Juan. >> >> >> >> >> On Thu, Jan 15, 2015 at 9:44 PM, Ankit Agrawal <aaaagrawal(a)gmail.com> >> wrote: >> >>> Hi everyone, >>> >>> I am working on a problem a small step of which requires me to >>> perform segmentation of a multi-channel/multi-band image. That is, like RGB >>> images which are 2D images with 3 channels, I have 2D images with 5 >>> channels. I looked at the segmentation API >>> <http://scikit-image.org/docs/dev/api/skimage.segmentation.html> and it >>> seems to me that all the current implement algorithms work on 2D single >>> channel(grayscale) or/and 2D three channel(RGB). I don't have a thorough >>> knowledge of the implemented segmentation algorithms, so I would like to >>> know if any of them can be extended to multi-channel(>3) images. Thanks. >>> >>> Regards, >>> Ankit. >>> >>> -- >>> 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/_WIES8XQrN8/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.

2 1

Re: Matching images with rotation/scaling invariant match
by Johannes Schoenberger 15 Jan '15

15 Jan '15

Number of matched features would be a good measure. Number of matched features is the number of rows of the match matrix. Depending on what you do, I would also suggest some geometric verification of the matches.

1 0

Matching images with rotation/scaling invariant match
by Hrishi Ganu 15 Jan '15

15 Jan '15

I have a template image and want to match it with a set of images. Since the template could be in a rotated position or scaled in the image I want to use ORB. I have extracted the ORB descriptor for the template image and for each of the images to be matched. Question-how should I assess the degree of match between the template's ORB descriptor and that for the image? Can I use len() ndarray returned by match_descriptors since I'm specifying the same # keypoints for both the templete and the images? Hrishi

1 0

Segmentation of multi-channel(>3) images
by Ankit Agrawal 15 Jan '15

15 Jan '15

Hi everyone, I am working on a problem a small step of which requires me to perform segmentation of a multi-channel/multi-band image. That is, like RGB images which are 2D images with 3 channels, I have 2D images with 5 channels. I looked at the segmentation API <http://scikit-image.org/docs/dev/api/skimage.segmentation.html> and it seems to me that all the current implement algorithms work on 2D single channel(grayscale) or/and 2D three channel(RGB). I don't have a thorough knowledge of the implemented segmentation algorithms, so I would like to know if any of them can be extended to multi-channel(>3) images. Thanks. Regards, Ankit.

3 3

Gsoc 2015 Aspirant Introduction
by GAURAV MISHRA 13 Jan '15

13 Jan '15

Hi everyone, I am Gaurav Mishra, an undergraduate in Computer Science and Engineering .I am doing research in Computer Vision,Machine Learning and have a keen interest in Image Processing. My skills include C/C++/Python/Matlab/Opencv.I would like to contribute to your organisation.Please suggest me as to how should i start contributing.Also is there any irc channel for your organisation? Your guidance will be highly appreciated.

4 4