From c6e5c7095350e931709acd894a54b32c42375822 Mon Sep 17 00:00:00 2001
From: Scott Sievert <me@scottsievert.com>
Date: Sat, 23 Jul 2016 15:09:12 -0500
Subject: [PATCH] makes PyWavelets required

---
 optional_requirements.txt                 |  1 -
 requirements.txt                          |  1 +
 skimage/restoration/_denoise.py           | 25 ++++++++---------------
 skimage/restoration/tests/test_denoise.py |  6 +++---
 4 files changed, 13 insertions(+), 20 deletions(-)

diff --git a/optional_requirements.txt b/optional_requirements.txt
index 6f923675..8bc53bb0 100644
--- a/optional_requirements.txt
+++ b/optional_requirements.txt
@@ -4,4 +4,3 @@ SimpleITK; python_version != '3.4' and python_version != '3.5'
 astropy
 tifffile
 imageio
-PyWavelets
diff --git a/requirements.txt b/requirements.txt
index 1e6921be..b05402dc 100644
--- a/requirements.txt
+++ b/requirements.txt
@@ -5,3 +5,4 @@ six>=1.7.3
 networkx>=1.8
 pillow>=2.1.0
 dask[array]>=0.5.0
+PyWavelets>=0.4.0
diff --git a/skimage/restoration/_denoise.py b/skimage/restoration/_denoise.py
index 33fac776..94a11608 100644
--- a/skimage/restoration/_denoise.py
+++ b/skimage/restoration/_denoise.py
@@ -5,6 +5,7 @@ from .. import img_as_float
 from ..restoration._denoise_cy import _denoise_bilateral, _denoise_tv_bregman
 from .._shared.utils import skimage_deprecation, warn
 import warnings
+import pywt
 
 
 def denoise_bilateral(image, win_size=None, sigma_color=None, sigma_spatial=1,
@@ -336,6 +337,7 @@ def denoise_tv_chambolle(im, weight=0.1, eps=2.e-4, n_iter_max=200,
 
 def _wavelet_threshold(img, wavelet, threshold=None, sigma=None, mode='soft'):
     """Performs wavelet denoising.
+
     Parameters
     ----------
     img : ndarray (2d or 3d) of ints, uints or floats
@@ -372,8 +374,8 @@ def _wavelet_threshold(img, wavelet, threshold=None, sigma=None, mode='soft'):
     .. [2] D. L. Donoho and I. M. Johnstone. "Ideal spatial adaptation
            by wavelet shrinkage." Biometrika 81.3 (1994): 425-455.
            DOI: 10.1093/biomet/81.3.425
+
     """
-    import pywt
     coeffs = pywt.wavedecn(img, wavelet=wavelet)
     detail_coeffs = coeffs[-1]['d' * img.ndim]
 
@@ -393,7 +395,8 @@ def _wavelet_threshold(img, wavelet, threshold=None, sigma=None, mode='soft'):
 
 
 def denoise_wavelet(img, sigma=None, wavelet='db1', mode='soft'):
-    r"""Performs wavelet denoising on an image.
+    """Performs wavelet denoising on an image.
+
     Parameters
     ----------
     img : ndarray (greater than 2d) of ints, uints or floats
@@ -406,7 +409,7 @@ def denoise_wavelet(img, sigma=None, wavelet='db1', mode='soft'):
         standard deviation is estimated via the method in [2]_.
     wavelet : string, optional
         The type of wavelet to perform and can be any of the options
-        [pywt.wavelist]_ outputs. The default is `'db1'`. For example,
+        ``pywt.wavelist`` outputs. The default is `'db1'`. For example,
         ``wavelet`` can be any of ``{'db2', 'haar', 'sym9'}`` and many more.
     mode : {'soft', 'hard'}, optional
         An optional argument to choose the type of denoising performed. It
@@ -424,15 +427,6 @@ def denoise_wavelet(img, sigma=None, wavelet='db1', mode='soft'):
     wavelet domain. Correspondingly, many pixel values of an image are 0 after
     taking the wavelet transform.
 
-    By wavelet denoising, we are enforcing that many of the wavelet
-    coefficients are 0 while keeping the error small. When we use soft
-    thresholding, our estimate is
-
-    .. math:: \widehat{x} = \arg \min_x ||z - x||_2^2 + \lambda ||x||_1
-
-    where :math:`z` is the input image wavelet coefficients and :math:`\lambda`
-    is the threshold.
-
     This function performs wavelet denoising on each color plane separately.
     The output is clipped between 0 and 1.
 
@@ -451,11 +445,10 @@ def denoise_wavelet(img, sigma=None, wavelet='db1', mode='soft'):
     >>> from skimage import color, data
     >>> img = img_as_float(data.astronaut())
     >>> img = color.rgb2gray(img)
-    >>> img += 0.5 * img.std() * np.random.randn(*img.shape)
+    >>> img += 0.1 * np.random.randn(*img.shape)
     >>> img = np.clip(img, 0, 1)
-    >>> denoised_img = denoise_wavelet(img)
-    >>> assert denoised_img.min() >= 0.0
-    >>> assert denoised_img.max() <= 1.0
+    >>> denoised_img = denoise_wavelet(img, sigma=0.1)
+
     """
 
     if not img.dtype.kind == 'f':
diff --git a/skimage/restoration/tests/test_denoise.py b/skimage/restoration/tests/test_denoise.py
index 664f89a3..206f1923 100644
--- a/skimage/restoration/tests/test_denoise.py
+++ b/skimage/restoration/tests/test_denoise.py
@@ -313,11 +313,11 @@ def test_wavelet_denoising():
         img = img.copy() + 0.1 * np.random.randn(*(img.shape))
         img = np.clip(img, 0, 1)
         # less energy in signal
-        assert restoration.denoise_wavelet(img).sum() <= img.sum()
+        assert restoration.denoise_wavelet(img, sigma=0.3).sum() <= img.sum()
 
         # test changing noise_std (higher threshold, so less energy in signal)
-        assert (restoration.denoise_wavelet(img, noise_stdev=0.2).sum() <=
-                restoration.denoise_wavelet(img, noise_stdev=0.1).sum())
+        assert (restoration.denoise_wavelet(img, sigma=0.2).sum() <=
+                restoration.denoise_wavelet(img, sigma=0.1).sum())
 
         # This works for this particular image
         assert (np.sum(restoration.denoise_wavelet(img, wavelet='db2')) <