All's in the title... regmax is currently generated using SAFE_SWITCH_ON_INTEGER_TYPES_OF and checks for integer-ness of the input; is there any specific reason why it doesn't handle float types?

By the way, the docstring of regmax doesn't mention this limitation.

I ran regmax on a 600x600x600 bool array and once it started, the memory usage slowly increased passed the amount of physical memory that I have (32GB). I realized that I needed to specify a new structuring element for 3D, but even now regmax is using 10GB on a file that is smaller than 1GB. Any ideas as to why this is?

For the original problem, maybe you could just add

assert Bc.ndim == f.ndim

into your code

Another edge detection method besides sobel. DoG can approximate Laplacian of Gaussian operator. It works as an efficient edge detector.

Will add a demo soon.

Hi there!

I thought you might be interested in adding these LGTM code quality badges to your project. They indicate how you care about code quality and encourage your future contributors to do the same. To get an idea of the analyses reflected by these grades, check the alerts discovered by LGTM.

N.B.: I am on the team behind LGTM.com, I'd appreciate your feedback on this initiative, whether you're interested or not, if you find time to drop me a line. Thanks.

When mahotas is installed with pip install mahotas --user building wheels fails with the following error on Ubuntu 16.04 with Python 2.7.12:

    x86_64-linux-gnu-gcc -pthread -DNDEBUG -g -fwrapv -O2 -Wall -Wstrict-prototypes -fno-strict-aliasing -Wdate-time -D_FORTIFY_SOURCE=2 -g -fstack-protector-strong -Wformat -Werror=format-security -fPIC -DNPY_NO_DEPRECATED_API=NPY_1_7_API_VERSION -DPY_ARRAY_UNIQUE_SYMBOL=Mahotas_PyArray_API_Symbol -I[] -I/usr/include/python2.7 -c mahotas/_histogram.cpp -o build/temp.linux-x86_64-2.7/mahotas/_histogram.o
    cc1plus: warning: command line option '-Wstrict-prototypes' is valid for C/ObjC but not for C++
    In file included from mahotas/numpypp/array.hpp:17:0,
                     from mahotas/_histogram.cpp:10:
    mahotas/numpypp/numpy.hpp:11:33: fatal error: numpy/ndarrayobject.h: No such file or directory
    compilation terminated.
    error: command 'x86_64-linux-gnu-gcc' failed with exit status 1

  ----------------------------------------
  Failed building wheel for mahotas
  Running setup.py clean for mahotas
Failed to build mahotas
Installing collected packages: numpy, mahotas
  Running setup.py install for mahotas ... done
Successfully installed mahotas-1.4.3 numpy-1.12.0

This fails:

docker run -t hackermd/ubuntu-xenial-python pip install mahotas --user

This works:

docker run -t hackermd/ubuntu-xenial-python sudo pip install mahotas

I try to convert a huge file of raw video data into many multipage TIFF files by using the "freeimage.write_multipage()" function of the freeimage package from the Mahotas library (Python 2.7). Unfortunately, it seems that this "very easy to use" function doesn't release memory when running the script. So, my script works fine for small input raw files (less than 1 GB) but crashes with bigger files (a 3 GB input file crashes with Win XP pro 32 - ram 3.2 GB). My goal is to convert input files up to 1.5 TB. An extract of the script below:

When I run my script, the Windows Task manager shows an increase of the used ram, output file after output file until the crash which release all the used ram :

An extract of the reported error :

Currently, the output dtype of cwatershed is the same as the dtype of the surface argument (I think this happens in the declaration of res_a in py_cwatershed), which creates a wraparound issue if e.g. there are more than 255 markers but surface is of dtype uint8 (which is actually the case in the examples given in the doc). I think that a quick fix would be to just use the dtype of the markers argument instead.

I get this error on Mac 10.8:

$ python3.2 setup.py build
Traceback (most recent call last):
  File "setup.py", line 40, in <module>
    long_description = open('docs/source/readme.rst').read()
  File "/opt/local/Library/Frameworks/Python.framework/Versions/3.2/lib/python3.2/encodings/ascii.py", line 26, in decode
    return codecs.ascii_decode(input, self.errors)[0]
UnicodeDecodeError: 'ascii' codec can't decode byte 0xc3 in position 5169: ordinal not in range(128)

I was attempting to process a 2304x2304x2475 voxel tomogram using cwatershed. I'd keep getting an array of zeros out. I eventually found that just processing 100 slices worked fine, so I kept increasing the number of slices until it failed. It seems like the issue is that as soon as the no. of voxels exceeds 2**31, cwatershed fails (either giving an array of zeros, or once even causing Python to segfault). Looking at _morph.cpp, I think the problem is that int is used for indices, e.g. line 599, instead of int64_t or uint64_t.

I need to calculate the Euler characteristic with periodic boundary conditions, which works very nicely with 'wrap' mode.

BTW, I implemented an earlier algorithm for the characteristic (described in http://dx.doi.org/10.1016/S0010-4655(00)00139-9) and it happily gives the same answer as mahotas. But, mahotas is so much faster.

FAIL: mahotas.tests.test_interpolate.test_shift_ratio
----------------------------------------------------------------------
Traceback (most recent call last):
  File "X:\Python27\lib\site-packages\nose\case.py", line 197, in runTest
    self.test(*self.arg)
  File "X:\Python27\lib\site-packages\mahotas\tests\test_interpolate.py", line 37, in test_shift_ratio
    assert np.abs(ratio - 1.) < .01
AssertionError

Strangely, this fails on win32-py2.7 only, not win32-py2.6, win-amd64-py2.7 or any other platform I tested, event though they use the same compiler and compiler settings.

A reduced example that fails is:

f = np.zeros((3,3))
f[1,1] = 128
output = interpolate.shift(f,(1,1))

On win32-py2.7 the output is

[[ -7.78937000e+05  -7.78937000e+05  -7.78937000e+05]
 [ -7.78937000e+05   0.00000000e+00  -7.99360578e-15]
 [ -7.78937000e+05  -8.88178420e-15   1.28000000e+02]]

(note the -7.78937000e+05 on the borders) while on other platforms output is close to the expected result:

 [[  1.48200610e-307   1.48200610e-307   1.48200610e-307]
 [  1.48200610e-307   0.00000000e+000  -7.99360578e-015]
 [  1.48200610e-307  -8.88178420e-015   1.28000000e+002]]

Hi, Python 3.11 is already out. It will be nice if wheels for python 3.11 will be uploaded to pypi.

I also see that you use NumPy as a build requirement which means using the latest available NumPy may lead to errors as described in #132 (the author of this issue has to old NumPy version).

I suggest switching to oldest-supported-numpy metapackage that will install the oldest supported NumPy for a given python version.

Hi all,

I was using this library as a lightweight alternative to scipy, but I ran into a few inconveniences. I usually work with 2D images that are 32 bit floats and normalized to [0..1]. However, it seems some functions, notably erode and dilate, do not accept floating point arrays. Is there a particular reason to why erode and dilate only accepts integer data types? If so, it is no biggie as I can simply emulate floating point arithmetics with 32 bit integers.

Kind regards, Erich

Problem

On aarch64, ‘pip install mahotas’ builds the wheels from source code and then installs it. It requires the user to have a development environment installed on his system. Also, it takes some time to build the wheels than downloading and extracting the wheels from pypi.

Resolution

On aarch64, ‘pip install mahotas’ should download the wheels from pypi

@luispedro and Team Please let me know your interest in releasing aarch64 wheels. I can help in this.

I am trying to get Zernike moments for some images but I keep getting the following error

AttributeError                            Traceback (most recent call last)
<ipython-input-33-6eb9b0e4463a> in <module>()
----> 1 get_moments(img, 10)

<ipython-input-32-b3923b47676f> in get_moments(image, radius)
      1 def get_moments(image, radius):
----> 2     moments = mahotas.features.zernike_moments(image, radius)
      3     return moments

AttributeError: module 'mahotas' has no attribute 'features'

Can you suggest the fix for this?

Stack Trace -

Traceback (most recent call last):
  File "C:/MS/Courses/CSE 515 MWDB/Project/Phase1/main.py", line 41, in <module>
    elbp.local_binary_pattern(image1)
  File "C:\MS\Courses\CSE 515 MWDB\Project\Phase1\feature_extractor.py", line 66, in local_binary_pattern
    feature = lbp(image, radius=3, points=16)
  File "C:\Users\fateh\anaconda3\envs\Phase1\lib\site-packages\mahotas\features\lbp.py", line 151, in lbp
    codes = lbp_transform(image, radius, points, ignore_zeros=ignore_zeros, preserve_shape=False)
  File "C:\Users\fateh\anaconda3\envs\Phase1\lib\site-packages\mahotas\features\lbp.py", line 75, in lbp_transform
    select(shift(image, [radius*dy,radius*dx], order=1)))
  File "C:\Users\fateh\anaconda3\envs\Phase1\lib\site-packages\mahotas\interpolate.py", line 289, in shift
    _interpolate.zoom_shift(array, None, shift, output, order, mode2int[mode], cval)
RuntimeError: mahotas.zoom_shift: input data has unexpected types. This may be a bug in mahotas.

Code -

from mahotas.features.lbp import lbp
feature = lbp(image, radius=3, points=16)

print(image1.shape)

gives (64, 64) which is as expected (a 2d array)

print(type(image1)) 

gives <class 'numpy.ndarray'> which is as expected (i.e. 2-D numpy ndarray)