Mailman 3 February 2016 - scikit-image

Re: skimage and opencv
by Emmanuelle Gouillart Sept. 22, 2016

Sept. 22, 2016

Hi FranÃ¯Â¿Â½ois, that's an excellent question, and not a troll :-). Opencv is a very powerful library, but it focuses primarily on computer vision (feature detection and extraction, classification, ...), as opposed to image processing in general (with other tasks such as denoising, segmentation, ...). The other big difference is that skimage builds on numpy ndarrays, and uses the full power of the numpy API (including of course the basic facilities for processing arrays as images that come with numpy), as well as some of scipy functions (you could have added scipy.ndimage to your list -- a few functions in skimage are wrappers around scipy.ndimage, that exist for the sake of completeneness). One important consequence is that algorithms working for 3-d or even n-d images can be easily implemented in 3-d/n-d in skimage, whereas opencv is restricted to 2-D images (as far as I know). Thanks to the use of numpy arrays, the API of skimage is also quite pleasant for a numpy user, more than the API of opencv. A related difference is that skimage is written in python and cython, whereas opencv is a C++ library. The two libraries attract a different crowd of developers, and a Python/Cython toolkit based on numpy arrays is easier to develop and maintain inside the Scientific Python ecosystem. I'm sure that other devs/users will have things to add to this discussion! Cheers, Emmanuelle On Thu, Dec 27, 2012 at 02:06:08PM -0800, FranÃ¯Â¿Â½ois wrote: > Hi users and devs, > It came to my knowledge that another python library (based on C++ and C > codes) for image processing exists too : opencv > I understand that numpy intregrates some basic features and we need some > advanced features but I have the feeling that skimages is redoundant with > opencv in some ways. > What's the position of skimage about that? (Don't read this question as a > troll but like a real question). > I mean that similar features exist in both. Would not be possible to > reuse/integrate opencv or merge? what's the reason for keeping them apart? > My observation is there is 4 libraries to manipulate images: > * PIL > * numpy > * skimages > * opencv > That's a lot. > Cheers,

3 2

Re: img_as_float
by Michael Aye April 29, 2016

April 29, 2016

I am not opposed to supporting range preservation, but we do have to > think a bit about the implications: > > - what do you do when the data-type of values change? > What are the situations where they *have* to change? > - what do you do when your operation due to, e.g., rounding issues > push values outside the input range? > Return a new object instead of changing the original, maybe? > - what do you do when you need to know the full potential range of the > data? > > don't understand, do you mean the full potential range per data-type? isn't that defined by the data-type the input image has? The ``preserve_range`` flag has allowed us to do whatever we do > normally, unless the user gave explicit permission to change data > types, ranges, etc. It also serves as a nice tag for "I, the > developer, thought about this issue". > > And that's quite cool that that's offered, but the question is, I guess, which default is best and why? Which default setting would confuse the least new (and old) users? Michael

3 2

Feature extraction for 3D model
by Zichao Li April 29, 2016

April 29, 2016

I have about 1Gb 3d models contained in .obj file and now I'd like to do some feature extraction: color, texture, shape etc. I wonder how can i do it with scikit-image? I can't find relative explanation in documents. Thanks a lot!

2 3

Re: Extracting data from an image of plots
by Josh Warner Feb. 27, 2016

Feb. 27, 2016

You may want to look into the work of Matt Terry (@mrterry) from SciPy 2013 and earlier. There are a number of tools he created, using what may now be outdated matplotlib interfaces but which could save you time. One of which is yoink (https://github.com/mrterry/yoink), which is particularly good with rastered data behind a color map. Using tools like Inkscape and/or the GIMP you should be able to crop any arbitrary figure out of a scanned paper/image and transform it so it's at least relatively rectilinear. Save that out as an image and these tools start to be useful - though there may be some customization left to go depending on what type of figure you're digitizing. Matt gave a lightning talk at SciPy 2013 about yoink: https://youtu.be/ywHqIEv3xXg?t=1890 Josh On Friday, February 26, 2016 at 9:12:07 AM UTC-7, Himanshu Mishra wrote: > > WebPlotDigitizer works very well for an image with single plot. But when I > try it with a research paper with text on it and more than one plot, it > works miserably. I don't know if such tool is available online, but I would > love to hear suggestions on how to make one. > > Thank you, > Himanshu Mishra >

2 1

HOG enhancement to accept color images #1963
by Pradyumna Kumar N Feb. 26, 2016

Feb. 26, 2016

Hi all, I implemented the HOG for color images based on the excerpt from the original paper "For colour images, we calculate separate gradients for each colour channel, and take the one with the largest norm as the pixel’s gradient vector"[Dalal and Triggs] and the CVPR presentation "For color image, pick the color channel with the highest gradient magnitude for each pixel"[Dalal and Triggs]. To allow this change, I refactored the code so that image gradients are computed in a separate method. If the image has more than 1 channels, this method is called on each channel. Gradient magnitude is calculated for each channel and for each pixel, the gradient vector with maximum magnitude is considered. This is my first open source contribution and I implemented the change based on my understanding. I pushed my changes to Github. Please review my code and let me know if I implemented the functionality correctly. Please comment on coding style too.

1 0

Extracting data from an image of plots
by Himanshu Mishra Feb. 26, 2016

Feb. 26, 2016

Hello everyone, We have a pdf page which contains one or more figures which are two-dimensional plots of experimental results. The figures may or may not be embedded in text. Each plot has the x and y axis with their labels and unit measurements marked in the plot. Inside each figure are one or more plots, each with a different color. How can we convert the plot into a table of corresponding x and y values (say for 100 points) ? Looking forward to a solution, Thank you, Himanshu Mishra

2 2

Re: Livewire segmentation in scikit-image
by Pavlo Dyban Feb. 25, 2016

Feb. 25, 2016

Hello Stéfan! You are right. Livewire segmentation approach is often used for interactive segmentation when the contour is updated and displayed while user is moving the mouse. However, the approach itself is not limited to interactive segmentations, but can just as well be used as a semi-automatic segmentation approach. This is the functionality that the iPython ntebook in my repository on Github demonstrates (https://github.com/pdyban/livewire/blob/master/notebooks/Livewire.ipynb). Even though scikit-image is majorilly used as a static tool (wihtout on-the-fly user interaction), it can nonetheless be used in the latter case. In both cases I imagine that Liewire segmentation would enrich the segmentation arsenal in scikit-image. The algorithm is simple and follows along this article: https://en.wikipedia.org/wiki/Livewire_Segmentation_Technique. An enhanced search for the shortest path has been implemented based on a thread here: http://code.activestate.com/recipes/119466-dijkstras-algorithm-for-shortest…, see comment 1. Please let me know if you have any concerns. Thanks! On Tuesday, 23 February 2016 03:56:30 UTC+1, stefanv wrote: > > Hi Pavlo > > On 22 February 2016 at 13:43, Pavlo Dyban <pdy...(a)gmail.com <javascript:>> > wrote: > > Hello! I am a great fan of scikit-learn and have used it in a number of > > various projects so far. This time, I want to contribute to the project. > I > > have implemented Livewire segmentation algorithm for image segmentation > and > > would like to bring it into the main package. My code is hosted on > github, > > it is documented with sphinx and tested. > > Thanks for your interest in contributing to scikit-image! > > > Livewire segmentation technique deduces object boundaries in the image > by > > converting the image to a weighted graph where edges' weights are > computed > > from the gradient image. The shortest path algorithm minimises the total > > cost function, thus avoiding steep gradients (i.e. object boundaries in > the > > original image). An example of how the algorithm works you will find in > my > > repository: https://github.com/pdyban/livewire. > > I've only heard of livewire in terms of interactive work. Is it > commonly used as a stand-alone technique? > > > Do you think this algorithm would be needed inside scikit-image? If yes, > > would that belong inside segmentation module? Could someone assist me in > my > > first open source contribution? It would be great if I could contribute > to > > your project! Thanks! > > It may well be--I will rely on our segmentation specialists Emmanuelle > and Juan to take a look. Do you perhaps have a citation for this > method? > > We're happy to help you get pull requests off the ground. In the mean > time, take a look at: > > http://scikit-image.org/docs/stable/contribute.html > > Thanks! > Stéfan >

2 1

Re: New Contributor - Want to work on #1875 - Walker fails if a seed is surrounded by masked pixels
by Pradyumna Kumar N Feb. 25, 2016

Feb. 25, 2016

Hi Stéfan, If you have some bandwidth, can you mentor me so that I can start contributing to scikit-image. On Friday, 22 January 2016 12:43:19 UTC-7, stefanv wrote: > > Hi Pradyumna > > On Fri, Jan 22, 2016 at 11:39 AM, Pradyumna Kumar N > <pradyu...(a)gmail.com <javascript:>> wrote: > > I am Pradyumna, pursuing my PhD at Colorado State University in the > field of > > Computer Vision. I would like to contribute to scikit-image project. I > > already forked the project and I would like to work on #1875. I am in > the > > process of reading the random walk paper. In the mean time, any pointers > or > > help on #1875 would be appreciated. > > > > In the long run I would like to implement DTW routine if it is not > already > > implemented. > > Welcome, and thanks for working on these issues! At the moment we're > in release mode, but if all goes well that should be done soon and > have a bit more bandwidth to spend on mentoring. > > Regards > Stéfan >

2 1

Re: 0.12 release
by Emmanuelle Gouillart Feb. 25, 2016

Feb. 25, 2016

Hi StÃ©fan, > #1937 highlights a bug in our tests (and failing tests, that aren't > currently showing up), so I would consider that the last blocker. > I'll try to get around to #1905, but if I don't that's also OK. OK ! For #1937, I think the bug highlighted by test_orb.test_keypoints_orb_desired_no_of_keypoints should be fixed before the PR can be merged. Otherwise we'll have Travis failing all the time, for all new PRs. Am I wrong? And many thanks for all the explanations! Cheers, Emma

2 1

Undergraduate team
by Stéfan van der Walt Feb. 25, 2016

Feb. 25, 2016

Hi, everyone Thanks to Berkeley's Undergraduate Research Assistants Program (URAP), you may be seeing a few new faces active on GitHub. Some of the URAP team members are new to open source development, so I'd like to thank you in advance for your patience and time in giving them a helping hand. The URAP team members are: Seth Park Emily Pedersen Jenny Dai Ben Gee Shaun Singh Anand Kuchibotla Thanks! Stéfan

1 0