Refactor "colour.first_order_colour_fit" definition.

colour-science · May 2, 2015 · a63d7f8 · a63d7f8
1 parent d5e8bf8
commit a63d7f8
Show file tree

Hide file tree

Showing 5 changed files with 11 additions and 254 deletions.
diff --git a/colour/algebra/__init__.py b/colour/algebra/__init__.py
@@ -11,7 +11,6 @@
                             SpragueInterpolator)
 from .matrix import is_identity
 from .random import random_triplet_generator
-from .regression import linear_regression
 
 __all__ = []
 __all__ += coordinates.__all__
@@ -21,4 +20,3 @@
             'SpragueInterpolator']
 __all__ += ['is_identity']
 __all__ += ['random_triplet_generator']
-__all__ += ['linear_regression']
diff --git a/colour/algebra/regression.py b/colour/algebra/regression.py
diff --git a/colour/algebra/tests/tests_regression.py b/colour/algebra/tests/tests_regression.py
diff --git a/colour/characterisation/fitting.py b/colour/characterisation/fitting.py
@@ -17,8 +17,6 @@
 
 import numpy as np
 
-from colour.algebra import linear_regression
-
 __author__ = 'Colour Developers'
 __copyright__ = 'Copyright (C) 2013 - 2015 - Colour Developers'
 __license__ = 'New BSD License - http://opensource.org/licenses/BSD-3-Clause'
@@ -31,8 +29,8 @@
 
 def first_order_colour_fit(m1, m2):
     """
-    Performs a first order colour fit from given :math:`m2` colour matrix to
-    :math:`m1` colour matrix. The resulting colour matrix is calculated using
+    Performs a first order colour fit from given :math:`m1` colour matrix to
+    :math:`m2` colour matrix. The resulting colour matrix is calculated using
     multiple linear regression.
 
     The purpose of that object is for example matching of two *ColorChecker*
@@ -41,9 +39,9 @@ def first_order_colour_fit(m1, m2):
     Parameters
     ----------
     m1 : array_like, (3, n)
-        Reference matrix the matrix :math:`m2` will be colour fitted against.
+        Test matrix :math:`m1` to fit onto matrix :math:`m2`.
     m2 : array_like, (3, n)
-        Matrix to fit.
+        Reference matrix the matrix :math:`m1` will be colour fitted against.
 
     Returns
     -------
@@ -103,18 +101,9 @@ def first_order_colour_fit(m1, m2):
     ...     [0.10283975, 0.10424680, 0.10384975],
     ...     [0.04742204, 0.04772203, 0.04914226]])
     >>> first_order_colour_fit(m1, m2)  # doctest: +ELLIPSIS
-    array([[ 1.4043128...,  0.0112806..., -0.2029710...],
-           [-0.0998911...,  1.5012214..., -0.1856479...],
-           [ 0.2248369..., -0.0767236...,  1.0496013...]])
+    array([[ 0.6982266...,  0.0307162...,  0.1621042...],
+           [ 0.0689349...,  0.6757961...,  0.1643038...],
+           [-0.0631495...,  0.0921247...,  0.9713415...]])
     """
 
-    m1 = np.asarray(m1)
-    m2 = np.asarray(m2)
-
-    x_coefficients = linear_regression(m1[..., 0], m2)
-    y_coefficients = linear_regression(m1[..., 1], m2)
-    z_coefficients = linear_regression(m1[..., 2], m2)
-
-    return np.array([x_coefficients[:3],
-                     y_coefficients[:3],
-                     z_coefficients[:3]]).reshape((3, 3))
+    return np.transpose(np.linalg.lstsq(m1, m2)[0])
diff --git a/colour/characterisation/tests/tests_fitting.py b/colour/characterisation/tests/tests_fitting.py
@@ -95,9 +95,9 @@ def test_first_order_colour_fit(self):
 
         np.testing.assert_almost_equal(
             first_order_colour_fit(M1, M2),
-            np.array([[1.40431285, 0.01128059, -0.20297103],
-                      [-0.09989111, 1.50122142, -0.18564796],
-                      [0.22483693, -0.07672362, 1.04960133]]),
+            np.array([[0.69822661, 0.03071629, 0.16210422],
+                      [0.06893499, 0.67579611, 0.16430385],
+                      [-0.06314955, 0.0921247, 0.97134152]]),
             decimal=7)
 
     @ignore_numpy_errors