I have implemented a simple Chan-Vese Algorithm in the segmentation module based mostly on matlab code by Yue Wu freely available at http://www.mathworks.com/matlabcentral/fileexchange/23445-chan-vese-active-contours-without-edges

Let me know what needs to be changed, improved or corrected and whether I have made any mistakes in the actual implementation of the algorithm.

This is part of a student project supervised by Olivier Debeir who is a contributor for this library. This is also my first time contributing to such a project, so please bear with me. .

For the sake of keeping a more readily accessibly reference to this controversial addition, here's a PR.

Discussion has thus far been at: https://groups.google.com/forum/#!topic/scikit-image/fDbdOgHgolI

This PR should not be merged until licensing issues discussed in that email thread have been addressed.

Example Code

from skimage.feature import get_blobs
from skimage import data
from matplotlib import pyplot as plt
import math

img = data.coins()
blobs = get_blobs(img)

fig = plt.figure()
ax = fig.add_subplot(1, 1, 1)
plt.imshow(img,cmap='gray')


for b in blobs:
    x,y = b[0],b[1]
    r = math.sqrt( b[2]/math.pi )
    c = plt.Circle((y,x),r,color='#ff0000',lw = 2,fill = False)
    ax.add_patch(c)

plt.show()

Results

This is my first PR to scikit-image. I would be glad if someone could go through it and suggest improvements

I haven't used Cython before.Could re writing _blob_overlap in Cython provide speed improvements ? The main bottle neck in this case is computing the Gaussian Filter.

@cdeil , @adonath Please share your thoughts.

LoG will be just minor additions to this. Once this is accepted I'll move onto that and the remaining ones.

Description

Use Pythran instead of cython to compile several kernels, while keeping somewhat high level and pythonic code.

Checklist

[It's fine to submit PRs which are a work in progress! But before they are merged, all PRs should provide:]

• [x] Clean style in the spirit of PEP8
• [x] Docstrings for all functions
• [ ] Gallery example in ./doc/examples (new features only)
• [x] Benchmark in ./benchmarks, if your changes aren't covered by an existing benchmark
• [x] Unit tests

[For detailed information on these and other aspects see scikit-image contribution guidelines]

References

#2956

For reviewers

(Don't remove the checklist below.)

• [ ] Check that the PR title is short, concise, and will make sense 1 year later.
• [ ] Check that new functions are imported in corresponding __init__.py.
• [ ] Check that new features, API changes, and deprecations are mentioned in doc/release/release_dev.rst.
• [ ] Consider backporting the PR with @meeseeksdev backport to v0.14.x

• Test against python 2.6, 2.7, 3.2, 3.3, and 3.4 on Travis. All but 3.2 use Anaconda environments.
• Make networkx and matplotlib optional dependencies for testing. Test first without, and then with them.
• Make PIL a required dependency - we could optionally allow one of a number of io plugins to be installed and make the tests pass with each
• Make compatible with numpy 1.6 - no np.pad and astype() takes not arguments.
• Make compatible py2.6: removed dict generators, could not use masked arrays as a Cython buffer in restoration\unwrap.
• Add version_requirements helpers in skimage._shared, with tests
• Bump required version of Cython to 0.19 due to relative imports error (see below)
• imread and pyfits do not support py3.2, so they are not included in that test.

Cython Error that required Cython >= 0.19:

Traceback (most recent call last):
  File "setup.py", line 151, in <module>
    cmdclass={'build_py': build_py},
  File "/home/silvester/anaconda/envs/py26test/lib/python2.6/site-packages/numpy/distutils/core.py", line 152, in setup
    config = configuration()
  File "setup.py", line 52, in configuration
    config.add_subpackage('skimage')
  File "/home/silvester/anaconda/envs/py26test/lib/python2.6/site-packages/numpy/distutils/misc_util.py", line 1002, in add_subpackage
    caller_level = 2)
  File "/home/silvester/anaconda/envs/py26test/lib/python2.6/site-packages/numpy/distutils/misc_util.py", line 971, in get_subpackage
    caller_level = caller_level + 1)
  File "/home/silvester/anaconda/envs/py26test/lib/python2.6/site-packages/numpy/distutils/misc_util.py", line 908, in _get_configuration_from_setup_py
    config = setup_module.configuration(*args)
  File "skimage/setup.py", line 16, in configuration
    config.add_subpackage('filter')
  File "/home/silvester/anaconda/envs/py26test/lib/python2.6/site-packages/numpy/distutils/misc_util.py", line 1002, in add_subpackage
    caller_level = 2)
  File "/home/silvester/anaconda/envs/py26test/lib/python2.6/site-packages/numpy/distutils/misc_util.py", line 971, in get_subpackage
    caller_level = caller_level + 1)
  File "/home/silvester/anaconda/envs/py26test/lib/python2.6/site-packages/numpy/distutils/misc_util.py", line 908, in _get_configuration_from_setup_py
    config = setup_module.configuration(*args)
  File "skimage/filter/setup.py", line 18, in configuration
    cython(['rank/generic_cy.pyx'], working_path=base_path)
  File "/home/silvester/workspace/scikit-image/skimage/_build.py", line 41, in cython
    cythonize(pyxfile)
  File "/home/silvester/anaconda/envs/py26test/lib/python2.6/site-packages/Cython/Build/Dependencies.py", line 603, in cythonize
    aliases=aliases)
  File "/home/silvester/anaconda/envs/py26test/lib/python2.6/site-packages/Cython/Build/Dependencies.py", line 546, in create_extension_list
    kwds = deps.distutils_info(file, aliases, base).values
  File "/home/silvester/anaconda/envs/py26test/lib/python2.6/site-packages/Cython/Build/Dependencies.py", line 457, in distutils_info
    return (self.transitive_merge(filename, self.distutils_info0, DistutilsInfo.merge)
  File "/home/silvester/anaconda/envs/py26test/lib/python2.6/site-packages/Cython/Build/Dependencies.py", line 467, in transitive_merge
    node, extract, merge, seen, {}, self.cimported_files)[0]
  File "/home/silvester/anaconda/envs/py26test/lib/python2.6/site-packages/Cython/Build/Dependencies.py", line 478, in transitive_merge_helper
    for next in outgoing(node):
  File "/home/silvester/anaconda/envs/py26test/lib/python2.6/site-packages/Cython/Utils.py", line 29, in wrapper
    res = cache[args] = f(self, *args)
  File "/home/silvester/anaconda/envs/py26test/lib/python2.6/site-packages/Cython/Build/Dependencies.py", line 406, in cimported_files
    pxd_file = self.find_pxd(module, filename)
  File "/home/silvester/anaconda/envs/py26test/lib/python2.6/site-packages/Cython/Utils.py", line 29, in wrapper
    res = cache[args] = f(self, *args)
  File "/home/silvester/anaconda/envs/py26test/lib/python2.6/site-packages/Cython/Build/Dependencies.py", line 388, in find_pxd
    raise NotImplementedError("New relative imports.")
NotImplementedError: New relative imports.

TO DO :

Replace the variables keypoints, scales, orientations, response with kpts_recarray:

    keypoints = np.array(np.dstack((row, col, octave, orientation, response)),
                         dtype=[('row', np.double), ('col', np.double),
                         ('octave', np.double), ('orientation', np.double),
                         ('response', np.double)])
    kpts_recarray = keypoints.view(np.recarray)

Description

Adds an example to implement the rolling ball algorithm. Relevant Issue: #3538 Relevant Paper: Biomedical Image Processing

For reviewers

• Check that the PR title is short, concise, and will make sense 1 year later.
• Check that new functions are imported in corresponding __init__.py.
• Check that new features, API changes, and deprecations are mentioned in doc/release/release_dev.rst.

• arbitrary channel combinations possible (not necessarily 3-channel images)
• improved speed by cythonized cluster center computation and only using necessary channels (for grayscale images I see a speed improvement of 4x)

I am highly recommending to revert some of the multi-channel magic. For me it was very confusing...

We must at least improve the documentation about the behavior of sigma and the multichannel parameters.

At the moment one of the test cases is failing, I have not had time to look into it.

This tries to simplify the requirements and also updates install instructions. It will most probably not work out of the box. Fixes #1506 (when all errors are ironed out).

I'm open to suggestions!

Following a message on the mailing-list, here is a first attempt at contributing my non-local means denoising code. It still lacks an example for the gallery. I would also welcome suggestions to improve the speed, since this algorithm is very slow...

Description

This is a new feature for skimage: connected operators or attribute openings / closings. At the moment, I have only implemented area openings/closings, diameter openings and closings and volume fill (which is is less used). These operators allow are based on image flooding, where the flooding stops as soon as the corresponding region fulfills a certain criterion. In the case of area closings, this is the surface of the region, in the case of diameter closings, this is the maximal extension of the region. New criteria can be added by implementing new Cython classes providing functions for initialize, update, fusion and stop criterion.

It is still work in progress, as I am currently looking at max-tree implementations of the same thing which might be more general.

Checklist

[It's fine to submit PRs which are a work in progress! But before they are merged, all PRs should provide:]

• [x] Clean style in the spirit of PEP8
• [x] Docstrings for all functions
• [x] Gallery example in ./doc/examples (new features only)
• [x] Unit tests

References

1. Breen, E.J., Jones, R. (1996). Attribute openings, thinnings and granulometries. Computer Vision and Image Understanding 64 (3), 377-389.
2. Vincent L., Proc. "Grayscale area openings and closings, their efficient implementation and applications", EURASIP Workshop on Mathematical Morphology and its Applications to Signal Processing, Barcelona, Spain, pp.22-27, May 1993.
3. Vachier, C., Meyer, F. (1995). Extinction values: a new measurement of persistence. In: Proceedings of the IEEE Workshop on Non Linear Signal and Image Processing (pp. 254-257).

For reviewers

(Don't remove the checklist below.)

• [ ] Check that the PR title is short, concise, and will make sense 1 year later.
• [ ] Check that new functions are imported in corresponding __init__.py.
• [ ] Check that new features, API changes, and deprecations are mentioned in doc/release/release_dev.rst.

Description:

I was in a situation where I wanted to do local thresholding on a 2D image, but I had mistakenly given a 1D array as the input argument to skimage.filters.ran.otsu(). The error made it seem like the function only accepted 3D arrays. I believe this could be fixed by running expanding the if tree a bit, like below. This pattern is repeated in a number of functions in this module.

This seems like a small fix, but I have never contributed to a library as big as this one and am afraid setting up a local dev environment would take too much of my time at the moment.

Current behaviour:

> import numpy as np
> from skimage.filters.rank import otsu
> from skimage.morphology import disk
> otsu(np.eye(10).ravel(), disk(3))

ValueError: The parameter `image` must be a 3-dimensional array

Desired behaviour:

> import numpy as np
> from skimage.filters.rank import otsu
> from skimage.morphology import disk
> otsu(np.eye(10).ravel(), disk(3))

ValueError: The parameter `image` must be a 2- or 3-dimensional array

Proposed solution

# scikit-image/skimage/filters/rank/generic.py

def otsu(image, ...):
    np_image = np.asanyarray(image)
    if np_image.ndim == 2:
        return _apply_scalar_per_pixel(...)
    elif np_image.ndim == 3:  # Added check
        return _apply_scalar_per_pixel_3D(...)
    else:
        raise ValueError('The parameter `image` must be a 2- or 3-dimensional array') # Added error

I am getting these type of error while doing the object detection with mrcnn in kangaroo dataset,, i have saw your post regarding this one but i cant get it how to solve that error in my program plzzzz solve this error..................

Description:

When using skeletonize on a 2d numpy array with an invalid method parameter, an error message is shown falsely stating that there is a mismatch in the array dimension.

I would expect the error message to highlight the incorrect method as the issue rather than incorrect array dimensionality.

Let me know if I can provide any further detail :smi

Way to reproduce:

import numpy as np
from skimage.morphology import skeletonize

result = skeletonize(np.ones((5, 5)), method='foo')

Traceback or output:

Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "C:\Users\james\anaconda3\envs\skan\lib\site-packages\skimage\morphology\_skeletonize.py", line 87, in skeletonize
    raise ValueError(f'skeletonize requires a 2D or 3D image as input, '
ValueError: skeletonize requires a 2D or 3D image as input, got 2D.

Version information:

3.10.8 | packaged by conda-forge | (main, Nov 24 2022, 14:07:00) [MSC v.1916 64 bit (AMD64)]
Windows-10-10.0.19044-SP0
scikit-image version: 0.19.3
numpy version: 1.23.5

Description

Closes #6643. This private function is used in the plot_j_invariant_tutorial gallery example. It's useful so make it public and more consistent with the other denoise_* functions. The "Notes" and "Examples" section are adapted from the same gallery example. I opted to keep the function in this module rather than moving it to _denoise.py to keep git blame intact.

Checklist

• Docstrings for all functions
• Gallery example in ./doc/examples (new features only)
• Benchmark in ./benchmarks, if your changes aren't covered by an existing benchmark
• Unit tests
• Clean style in the spirit of PEP8
• Descriptive commit messages (see below)

For reviewers

• Check that the PR title is short, concise, and will make sense 1 year later.
• Check that new functions are imported in corresponding __init__.py.
• Check that new features, API changes, and deprecations are mentioned in doc/release/release_dev.rst.
• There is a bot to help automate backporting a PR to an older branch. For example, to backport to v0.19.x after merging, add the following in a PR comment: @meeseeksdev backport to v0.19.x
• To run benchmarks on a PR, add the run-benchmark label. To rerun, the label can be removed and then added again. The benchmark output can be checked in the "Actions" tab.

Description:

ransac() in measure/fit.py raises an exception in case the number of provided samples equals min_samples, see https://github.com/scikit-image/scikit-image/blob/main/skimage/measure/fit.py#L808:

    if not (0 < min_samples < num_samples):
        raise ValueError(f"`min_samples` must be in range (0, {num_samples})")

The documentation of the parameter, see https://github.com/scikit-image/scikit-image/blob/main/skimage/measure/fit.py#L673:

    min_samples : int in range (0, N)
        The minimum number of data points to fit a model to.

The issue

The inherent meaning of "minimum" and the documentation suggests == to be valid. This aligns with my understanding of RANSAC and transformations that == should be accepted. Also, skimming over the code doesn't give me any reason to reject ==.

In addition, (0, N) in documentation and error message can be interpreted as open interval or tuple, both ambiguous. Proposal: Use 0 < ... ≤ N or in {1, ..., N} to bring clarity.

If there is a reason to go up to N - 1 only, improve the documentation (even though the variable name cannot be improved for compatibility reasons).

Way to reproduce:

No response

Traceback or output:

No response

Version information:

No response

Description:

I tried to make blob_dog work on the human_mitosis dataset and it was hard. @tlambert03 gave the hint that this function may only work with images of type float. We were now wondering if this might be a bug, or if a comment should be added to the documentation. My tests furthermore suggest that also blog_doh and blog_log suffer from the same problem.

I propose to make sure that results are same or similar independent from the pixel data type.

Way to reproduce:

from skimage.data import human_mitosis
from skimage.feature import blob_dog
from skimage.io import imshow

data = human_mitosis()

print("\nDoG")
print(len(blob_dog(data,               min_sigma=1, max_sigma=10)))
print(len(blob_dog(data.astype(float), min_sigma=1, max_sigma=10)))
print("\nLoG")
print(len(blob_log(data,               min_sigma=1, max_sigma=10)))
print(len(blob_log(data.astype(float), min_sigma=1, max_sigma=10)))
print("\nDoH")
print(len(blob_doh(data,               min_sigma=1, max_sigma=10)))
print(len(blob_doh(data.astype(float), min_sigma=1, max_sigma=10)))

Traceback or output:

DoG
0
471

LoG
52
10131

DoH
19
5317

Version information:

3.9.15 | packaged by conda-forge | (main, Nov 22 2022, 08:41:22) [MSC v.1929 64 bit (AMD64)]
Windows-10-10.0.19045-SP0
scikit-image version: 0.19.3
numpy version: 1.23.5