[scikit-learn] Announcement: scikit-image 0.13.0

Juan Nunez-Iglesias jni.soma at gmail.com
Wed Mar 29 14:22:10 EDT 2017


We're happy to (finally) announce the release of scikit-image v0.13.0!
Special thanks to our many contributors for making it possible! This
release is the result of over a year of work, with over 200 pull requests
by 82 contributors.

Linux and macOS wheels are available now on PyPI
<https://pypi.python.org/pypi/scikit-image>, together with a source
distribution. Use "pip install -U scikit-image" to get the latest version!
Packages on conda-forge, Windows wheels, and Debian packages should be
available within the next few days.

scikit-image is an image processing toolbox for SciPy that includes
algorithms for segmentation, geometric transformations, color space
manipulation, analysis, filtering, morphology, feature detection, and more.
For more information, examples, and documentation,
please visit our website:

http://scikit-image.org

and our gallery of examples

http://scikit-image.org/docs/dev/auto_examples/

Highlights
----------

- Improved n-dimensional image support. This release adds nD support to:

  * ``regionprops`` computation for centroids (#2083)
  * ``segmentation.clear_border`` (#2087)
  * Hessian matrix (#2194)

- In addition, the following new functions support nD images:

  * new wavelet denoising function, ``restoration.denoise_wavelet`` (#1833,
#2190, #2238, #2240, #2241, #2242, #2462)
  * new thresholding functions, ``filters.threshold_sauvola`` and
``filters.threshold_niblack`` (#2266, #2441)
  * new local maximum, local minimum, hmaxima, hminima functions (#2449)

- Grey level co-occurrence matrix (GLCM) now works with uint16 images
- ``filters.try_all_threshold`` to rapidly see output of various
thresholding methods
- Frangi and Hessian filters (2D only) (#2153)
- New *compact watershed* algorithm in ``segmentation.watershed`` (#2211)
- New *shape index* algorithm in ``feature.shape_index`` (#2312)

New functions and features
--------------------------

- Add threshold minimum algorithm (#2104)
- Implement mean and triangle thresholding (#2126)
- Add Frangi and Hessian filters (#2153)
- add bbox_area to region properties (#2187)
- colorconv: Add rgba2rgb() (#2181)
- Lewiner marching cubes algorithm (#2052)
- image inversion (#2199)
- wavelet denoising (from #1833) (#2190)
- routine to estimate the noise standard deviation from an image (#1837)
- Add compact watershed and clean up existing watershed (#2211)
- Added the missing 'grey2rgb' function. (#2316)
- Shape index (#2312)
- Fundamental and essential matrix 8-point algorithm (#1357)
- Add YUV, YIQ, YPbPr, YCbCr colorspaces
- Detection of local extrema from morphology (#2449)
- shannon entropy (#2416)

Documentation improvements
--------------------------

- add details about github SSH keys in contributing page (#2073)
- Add example for felzenszwalb image segmentation (#2096)
- Sphinx gallery for example gallery (#2078)
- Improved region boundary RAG docs (#2106)
- Add gallery Lucy-Richardson deconvolution algorithm (#2376)
- Gallery: Use Horse to illustrate Convex Hull (#2431)
- Add working with OpenCV in user guide (#2519)

Code improvements
-----------------

- Remove lena image from test suite (#1985)
- Remove duplicate mean calculation in skimage.feature.match_template
(#1980)
- Add nD support to clear_border (#2087)
- Add uint16 images support for co-occurrence matrix (#2095)
- Add default parameters for Gaussian and median filters (#2151)
- try_all to choose the best threshold algorithm (#2110)
- Add support for multichannel in Felzenszwalb segmentation (#2134)
- Improved SimilarityTransform, new EuclideanTransform class (#2044)
- ENH: Speed up Hessian matrix computation (#2194)
- add n-dimensional support to denoise_wavelet (#2242)
- Speedup ``inpaint_biharmonic`` (#2234)
- Update hessian matrix code to include order kwarg (#2327)
- Handle cases for label2rgb where input labels are negative and/or
nonconsecutive (#2370)
- Added watershed_line parameter (#2393)

API Changes
-----------

- Remove deprecated ``filter`` module. Use ``filters`` instead. (#2023)
- Remove ``skimage.filters.canny`` links. Use ``feature.canny`` instead.
(#2024)
- Removed Python 2.6 support and related checks (#2033)
- Remove deprecated {h/v}sobel, {h/v}prewitt, {h/v}scharr,
roberts_{positive/negative} filters (#2159)
- Remove deprecated ``_mode_deprecations`` (#2156)
- Remove deprecated None defaults in ``rescale_intensity`` (#2161)
- Parameters ``ntiles_x`` and ``ntiles_y`` have been removed from
``exposure.equalize_adapthist``
- The minimum NumPy version is now 1.11, and the minimum SciPy version is
now 0.17

Deprecations
------------

- clip_negative will be set to false by default in version 0.15 (func:
dtype_limits) (#2228)
- Deprecate "dynamic_range" in favor of "data_range" (#2384)
- The default value of the ``circle`` argument to ``radon`` and ``iradon``
transforms will be ``True`` in 0.15 (#2235)
- The default value of ``multichannel`` for ``denoise_bilateral`` and
``denoise_nl_means`` will be ``False`` in 0.15
- The default value of ``block_norm`` in ``feature.hog`` will be
L2-Hysteresis in 0.15.
- The ``threshold_adaptive`` function is deprecated. Use
``threshold_local`` instead.
- The default value of ``mode`` in ``transform.swirl``, ``resize``, and
``rescale`` will be "reflect" in 0.15.

For a complete list of contributors and pull requests merged in this
release, please see our release notes online:

https://github.com/scikit-image/scikit-image/blob/master/doc/release/release_0.13.rst

Please spread the word, including on Twitter
<https://twitter.com/jnuneziglesias/status/847145183250665472>!

Enjoy!

Juan.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/scikit-learn/attachments/20170329/e39735f2/attachment.html>


More information about the scikit-learn mailing list