在python包skimage中导入错误

Question

I want to use skimage.restoration.denoise_wavelet to denoise a image. But problem occurs on importing.

from skimage.restoration import denoise_nl_means,denoise_wavelet

ImportError Traceback (most recent call last) <ipython-input-2-161a32d32528> in <module>() ----> 1 from skimage.restoration import denoise_nl_means,denoise_wavelet ImportError: cannot import name denoise_wavelet

There is no problem in importing denoise_nl_means which is in the same category with denoise_wavelet . It doesn't make sense.

I used pip for installing and updating the skimage package and jupyter notebook for coding. I installed all requirements before installing scikit-image 0.12.3. The requirement items I installed were:

matplotlib 1.5.1,numpy 1.11.1,scipy 0.18.1,six 1.10.0,networkx 1.11,pillow 3.4.1,dask 0.10.0,PyWavelets 0.4.0 .

Answer 1

As you installed this library with pip, it probably installed the latest stable release, which is something within the 0.12 branch.

There are different documentations for different versions of scikit-learn. The function which you want to import does not exist yet in any stable branch if we interpret the following correctly:

Search for function-name in docs of development-branch : OK!

Search for function-name in docs of 0.12 branch No hits!

Sadly, the docs of scikit-image do not say when a function was added to the library (as done by scipy).

Further analysis can be done on github + blame here .

So if you need this function, grab the development-branch and install from sources! Always read the corresponding doc-versions!

Answer 2

使用以下命令：

pip install -U https://github.com/scikit-image/scikit-image/archive/master.zip