codespell (#165)

Author: schwehr

VERSIONS.txt

@@ -126,7 +126,7 @@ Release date: 2019.10.06
 New Features
 ------------
 
-* Support for ECOSTRESS spectral library (supercedes ASTER library).
+* Support for ECOSTRESS spectral library (supersedes ASTER library).
 
 * Accept "mask" and "bg" keywords in `imshow` and `get_rgb` (affects color
   scaling).
@@ -189,7 +189,7 @@ Bug Fixes
 * Fixed a bug that potentially produced incorrect results in `map_class_ids`
   (issue #53).
 
-* Removed unecessary import that cause python3 compatibility error in
+* Removed unnecessary import that cause python3 compatibility error in
   `aviris.open` (issue #54).
 
 * Removed `has_key` call breaking python3 compatibility (issue #56).
@@ -439,7 +439,7 @@ Performance Improvements
 ------------------------
 
 * PerceptronClassifier is roughly an order of magnitude faster due to better
-  use of numpy. Inputs are now scaled and weights are initialized withing the
+  use of numpy. Inputs are now scaled and weights are initialized within the
   data limits, which usually results in fewer iterations for convergence.
 
 ================================================================================
@@ -490,7 +490,7 @@ New Features
 Performance Improvements
 ------------------------
 
-* Significant speedup for sevaral algorithms using ndarray inputs:
+* Significant speedup for several algorithms using ndarray inputs:
 
   * 14x speedup for GMCL
 
@@ -528,7 +528,7 @@ Changes
 Bug Fixes
 ---------
 
-* Exception occured when *bands* argument was not provided to `imshow`.
+* Exception occurred when *bands* argument was not provided to `imshow`.
 
 * *stretch_all* parameter for `get_rgb` was sometimes being used when set to
   False or 0 (because it was not None).
@@ -572,7 +572,7 @@ Changes
 * If present, image band info is saved when `envi.save_image` is called.
 
 * Allow calling :func:`~spectral.oi.envi.create_image` using keyword args
-  instead of ENVI-specific header paramter names.
+  instead of ENVI-specific header parameter names.
 
 * `save_rgb` automatically determines the output file type, based on the
   filename extension.
@@ -748,7 +748,7 @@ Release date: 2012.07.15
 
 - Bug Fixes
 
-  - Fixed a bug in several deprecation warnings that caused infinte recursion.
+  - Fixed a bug in several deprecation warnings that caused infinite recursion.
 
   - Fixed mismatch in parameter names in kmeans_ndarray.
 
@@ -796,7 +796,7 @@ Release date: 2011.01.17
   - :meth:`spectral.kmeans` will accept a :exc:`KeyboardInterrupt` exception
     (i.e., CTRL-C) and return the results as of the previous iteration.
 
-  - Documention is now online via Sphinx.
+  - Documentation is now online via Sphinx.
 
 - Changes
 

spectral/algorithms/algorithms.py

@@ -147,7 +147,7 @@ def iterator_ij(mask, index=None):
 
     If `index` is not defined, iteration is performed over all non-zero
     elements.  If `index` is defined, iteration is performed over all
-    coordinates for whch `mask[i,j] == index`.
+    coordinates for which `mask[i,j] == index`.
     '''
 
     if mask.ndim != 2:
@@ -271,7 +271,7 @@ def cov_avg(image, mask, weighted=True):
         `mask` (integer-valued ndarray):
 
             Elements specify the classes associated with pixels in `image`.
-            All pixels associeted with non-zero elements of `mask` will be
+            All pixels associated with non-zero elements of `mask` will be
             used in the covariance calculation.
 
         `weighted` (bool, default True):
@@ -1174,7 +1174,7 @@ def ndvi(data, red, nir):
 
 def bdist(class1, class2):
     '''
-    Calulates the Bhattacharyya distance between two classes.
+    Calculates the Bhattacharyya distance between two classes.
 
     USAGE:  bd = bdist(class1, class2)
 
@@ -1201,7 +1201,7 @@ def bdist(class1, class2):
 
 def bdist_terms(a, b):
     '''
-    Calulate the linear and quadratic terms of the Bhattacharyya distance
+    Calculate the linear and quadratic terms of the Bhattacharyya distance
     between two classes.
 
     USAGE:  (linTerm, quadTerm) = bDistanceTerms(a, b)
@@ -1438,7 +1438,7 @@ def noise_from_diffs(X, direction='lowerright'):
 
         `X` (np.ndarray):
 
-            The data from which to estimage noise statistics. `X` should have
+            The data from which to estimate noise statistics. `X` should have
             shape `(nrows, ncols, nbands`).
 
         `direction` (str, default "lowerright"):
@@ -1495,7 +1495,7 @@ def __init__(self, signal, noise, napc):
 
             `napc` (:class:`~spectral.PrincipalComponents`):
 
-                Noise-Adjusted Pricipal Components
+                Noise-Adjusted Principal Components
         '''
         self.signal = signal
         self.noise = noise
@@ -1592,7 +1592,7 @@ def reduce(self, X, **kwargs):
 
                 Threshold signal-to-noise ratio (SNR) to retain.
 
-        Returns a verions of `X` with reduced dimensionality.
+        Returns a versions of `X` with reduced dimensionality.
 
         Note that calling this method is equivalent to calling the
         `get_reduction_transform` method with same keyword and applying the
@@ -1725,13 +1725,13 @@ def ppi(X, niters, threshold=0, centered=False, start=None, display=0,
 
             An optional array of initial purity indices. This can be used to
             continue computing PPI values after a previous call to `ppi` (i.e.,
-            set `start` equal to the return value from a previou call to `ppi`.
+            set `start` equal to the return value from a previous call to `ppi`.
             This should be an integer-valued array whose dimensions are equal
             to the first two dimensions of `X`.
 
         `display` (integer):
 
-            If set to a postive integer, a :class:`~spectral.graphics.spypylab.ImageView`
+            If set to a positive integer, a :class:`~spectral.graphics.spypylab.ImageView`
             window will be opened and dynamically display PPI values as the
             function iterates. The value specifies the number of PPI iterations
             between display updates. It is recommended to use a value around
@@ -1751,7 +1751,7 @@ def ppi(X, niters, threshold=0, centered=False, start=None, display=0,
         These keywords will be passed to the image display and only have an
         effect if the `display` argument is nonzero.
 
-    This function can be interruped with a KeyboardInterrupt (ctrl-C), in which
+    This function can be interrupted with a KeyboardInterrupt (ctrl-C), in which
     case, the most recent value of the PPI array will be returned. This can be
     used in conjunction with the `display` argument to view the progression of
     the PPI values until they appear stable, then terminate iteration using
@@ -1920,7 +1920,7 @@ def smacc(spectra, min_endmembers=None, max_residual_norm=float('Inf')):
         for k in range(len(Fs)):
             t = On * Fs[k][q[n]]
             # This is not so important for the algorithm itself.
-            # These values correpond to values where On == 0.0, and these
+            # These values correspond to values where On == 0.0, and these
             # will be zeroed out below. But to avoid divide-by-zero warning
             # we set small values instead of zero.
             t[t == 0.0] = 1e-10

spectral/algorithms/classifiers.py

@@ -284,7 +284,7 @@ class PerceptronClassifier(Perceptron, SupervisedClassifier):
     '''A multi-layer perceptron classifier with backpropagation learning.
 
     Multi-layer perceptrons often require many (i.e., thousands) of iterations
-    through the traning data to converge on a solution. Therefore, it is not
+    through the training data to converge on a solution. Therefore, it is not
     recommended to attempt training a network on full-dimensional hyperspectral
     data or even on a full set of image pixels. It is likely preferable to
     first train the network on a subset of the data, then retrain the network

spectral/algorithms/clustering.py

@@ -72,7 +72,7 @@ def kmeans(image, nclusters=10, max_iterations=20, **kwargs):
 
         `class_map` (:class:`numpy.ndarray`):
 
-            An `MxN` array whos values are the indices of the cluster for the
+            An `MxN` array who's values are the indices of the cluster for the
             corresponding element of `image`.
 
         `centers` (:class:`numpy.ndarray`):
@@ -250,7 +250,7 @@ def kmeans_ndarray(image, nclusters=10, max_iterations=20, **kwargs):
 
         `class_map` (:class:`numpy.ndarray`):
 
-            An `MxN` array whos values are the indices of the cluster for the
+            An `MxN` array who's values are the indices of the cluster for the
             corresponding element of `image`.
 
         `centers` (:class:`numpy.ndarray`):

spectral/algorithms/continuum.py

@@ -4,7 +4,7 @@
 Continuum is defined as convex hull of spectrum.
 Continuum is removed from spectra by dividing spectra by its continuum.
 That results in values between 0 and 1, where absorption bands are expressed as
-drops below 1. It is usefull for comparing and classification based on
+drops below 1. It is useful for comparing and classification based on
 absorption bands and indifferent to scale.
 
 References:
@@ -239,13 +239,13 @@ def continuum_points(spectrum, bands, mode='convex'):
             1d :class:`numpy.ndarray`, holding band values of spectra.
             Length of `bands` should be the same as `spectrum`.
             Note that bands should be sorted in ascending order (which is often
-            not the case with AVIRIS), otherwise unexpected results could occure.
+            not the case with AVIRIS), otherwise unexpected results could occur.
 
         `mode` (string, default 'convex'):
 
             Default mode is 'convex' which returns convex upper hull of the
             spectrum. Another supported mode is 'segmented' which builds
-            segmented upper hull. This is usefull to identify more detailed
+            segmented upper hull. This is useful to identify more detailed
             contour of the spectrum, but without strong absorption bands.
 
     Returns:
@@ -286,13 +286,13 @@ def spectral_continuum(spectra, bands, mode='convex', out=None):
             1d :class:`numpy.ndarray`, holding band values of spectra.
             Length of `bands` should be the same as last dimension of `spectra`.
             Note that bands should be sorted in ascending order (which is often
-            not the case with AVIRIS), otherwise unexpected results could occure.
+            not the case with AVIRIS), otherwise unexpected results could occur.
 
         `mode` (string, default 'convex'):
 
             Default mode is 'convex' which returns convex upper hull of the
             spectrum. Another supported mode is 'segmented' which builds
-            segmented upper hull. This is usefull to identify more detailed
+            segmented upper hull. This is useful to identify more detailed
             contour of the spectrum, but without strong absorption bands.
 
         `out` (:class:`numpy.ndarray`, default None):
@@ -313,7 +313,7 @@ def spectral_continuum(spectra, bands, mode='convex', out=None):
 def remove_continuum(spectra, bands, mode='convex', out=None):
     '''Returns spectra with continuum removed.
     Continuum is defined as convex hull of spectra. Continuum is removed from
-    spectra by deviding spectra by its continuum.
+    spectra by dividing spectra by its continuum.
 
     Arguments:
 
@@ -327,13 +327,13 @@ def remove_continuum(spectra, bands, mode='convex', out=None):
             1d :class:`numpy.ndarray`, holding band values of spectra.
             Length of `bands` should be the same as last dimension of `spectra`.
             Note that bands should be sorted in ascending order (which is often
-            not the case with AVIRIS), otherwise unexpected results could occure.
+            not the case with AVIRIS), otherwise unexpected results could occur.
 
         `mode` (string, default 'convex'):
 
             Default mode is 'convex' which removes convex upper hull of the
             spectrum. Another supported mode is 'segmented' which removes
-            segmented upper hull. This is usefull to identify two or more small
+            segmented upper hull. This is useful to identify two or more small
             features instead of one large feature.
 
         `out` (:class:`numpy.ndarray`, default None):

spectral/algorithms/detectors.py

@@ -124,7 +124,7 @@ def matched_filter(X, target, background=None, window=None, cov=None):
 
             Must have the form (`inner`, `outer`), where the two values
             specify the widths (in pixels) of inner and outer windows centered
-            about the pixel being evaulated. Both values must be odd integers.
+            about the pixel being evaluated. Both values must be odd integers.
             The background mean and covariance will be estimated from pixels
             in the outer window, excluding pixels within the inner window. For
             example, if (`inner`, `outer`) = (5, 21), then the number of
@@ -319,7 +319,7 @@ def rx(X, background=None, window=None, cov=None):
 
             Must have the form (`inner`, `outer`), where the two values
             specify the widths (in pixels) of inner and outer windows centered
-            about the pixel being evaulated. Both values must be odd integers.
+            about the pixel being evaluated. Both values must be odd integers.
             The background mean and covariance will be estimated from pixels
             in the outer window, excluding pixels within the inner window. For
             example, if (`inner`, `outer`) = (5, 21), then the number of
@@ -407,7 +407,7 @@ def __init__(self, target, background=None, **kwargs):
             `vectorize` (bool, default True):
 
                 Specifies whether the __call__ method should attempt to vectorize
-                operations. This typicall results in faster computation but will
+                operations. This typically results in faster computation but will
                 consume more memory.
         '''
         for k in kwargs:
@@ -564,7 +564,7 @@ def ace(X, target, background=None, window=None, cov=None, **kwargs):
                 The return value will be an ndarray with shape (R, C).
 
                 A length-D sequence (e.g., list or tuple) of length-B ndarrays.
-                In this case, the detector will be applied seperately to each of
+                In this case, the detector will be applied separately to each of
                 the `D` targets. This is equivalent to calling the function
                 sequentially for each target and stacking the results but is
                 much faster. The return value will be an ndarray with shape
@@ -580,7 +580,7 @@ def ace(X, target, background=None, window=None, cov=None, **kwargs):
 
             Must have the form (`inner`, `outer`), where the two values
             specify the widths (in pixels) of inner and outer windows centered
-            about the pixel being evaulated. Both values must be odd integers.
+            about the pixel being evaluated. Both values must be odd integers.
             The background mean and covariance will be estimated from pixels
             in the outer window, excluding pixels within the inner window. For
             example, if (`inner`, `outer`) = (5, 21), then the number of
@@ -615,7 +615,7 @@ def ace(X, target, background=None, window=None, cov=None, **kwargs):
         `vectorize` (bool, default True):
 
             Specifies whether the function should attempt to vectorize
-            operations. This typicall results in faster computation but will
+            operations. This typically results in faster computation but will
             consume more memory.
 
     Returns numpy.ndarray:

spectral/algorithms/perceptron.py

@@ -47,7 +47,7 @@ def __init__(self, shape, k=1.0, weights=None):
     def randomize_weights(self):
         '''Randomizes the layer weight matrix.
         The bias weight will be in the range [0, 1). The remaining weights will
-        correspond to a vector with unit length and uniform random orienation.
+        correspond to a vector with unit length and uniform random orientation.
         '''
         self.weights = 1. - 2. * np.random.rand(*self.shape)
         for row in self.weights:

spectral/algorithms/spatial.py

@@ -192,7 +192,7 @@ def map_window(func, image, window, rslice=(None,), cslice=(None,),
         `border` (string, default "shift"):
 
             Indicates how to handles windows near the edge of the window. If
-            the value is "shift", the window dimensions will alway be
+            the value is "shift", the window dimensions will always be
             `(width, height)` but near the image border the pixel being
             iterated will be offset from the center of the window. If set to
             "clip", window regions falling outside the image border will be
@@ -331,7 +331,7 @@ def map_outer_window_stats(func, image, inner, outer, dim_out=1, cov=None,
 
 
 class WindowedGaussianBackgroundMapper(object):
-    '''A class for procucing window statistics with an inner exclusion window.
+    '''A class for producing window statistics with an inner exclusion window.
     '''
     def __init__(self, inner, outer, function=None, cov=None, dim_out=None,
                  dtype=None):
@@ -341,11 +341,11 @@ def __init__(self, inner, outer, function=None, cov=None, dim_out=None,
 
             `inner` (integer or 2-tuple of integers):
 
-                Width and heigth of inner window, in pixels.
+                Width and height of inner window, in pixels.
 
             `outer` (integer or 2-tuple of integers):
 
-                Width and heigth of outer window, in pixels. Dimensions must
+                Width and height of outer window, in pixels. Dimensions must
                 be greater than inner window
 
             `function` (callable object):

spectral/algorithms/spymath.py

@@ -67,7 +67,7 @@ def get_histogram_cdf_points(data, cdf_vals, ignore=None, mask=None):
         `ignore` (numeric, default `None`):
 
             A scalar value that should be ignored when computing histogram
-            points (e.g., a value that indicates bad data). If this valus is
+            points (e.g., a value that indicates bad data). If this value is
             not specified, all data are used.
 
     Return value:

spectral/algorithms/transforms.py

@@ -39,7 +39,7 @@ def __init__(self, A, **kwargs):
 
                 An (J,K) array to be applied to length-K targets.
 
-        Keyword Argments:
+        Keyword Arguments:
 
             `pre` (scalar or length-K sequence):
 
@@ -58,7 +58,7 @@ def __init__(self, A, **kwargs):
         self._post = kwargs.get('post', None)
         A = np.array(A, copy=True)
         if A.ndim == 0:
-            # Do not know input/ouput dimensions
+            # Do not know input/output dimensions
             self._A = A
             (self.dim_out, self.dim_in) = (None, None)
         else: