Weights data type now also specifies window data type

the-lay · Jan 6, 2022 · 7d75102 · 7d75102
1 parent a731aab
commit 7d75102
Show file tree

Hide file tree

Showing 3 changed files with 16 additions and 10 deletions.
diff --git a/docs/index.html b/docs/index.html
@@ -2000,7 +2000,7 @@ <h6 id="return">Return</h6>
 <span class="sd">            data_dtype (np.dtype): Specify data type for data buffer that stores cumulative result.</span>
 <span class="sd">                Default is `np.float32`.</span>
 
-<span class="sd">            weights_dtype (np.dtype): Specify data type for weights buffer that stores cumulative weights.</span>
+<span class="sd">            weights_dtype (np.dtype): Specify data type for weights buffer that stores cumulative weights and window array.</span>
 <span class="sd">                If you don&#39;t need precision but would rather save memory you can use `np.float16`.</span>
 <span class="sd">                Likewise, on the opposite, you can use `np.float64`.</span>
 <span class="sd">                Default is `np.float32`.</span>
@@ -2039,7 +2039,7 @@ <h6 id="return">Return</h6>
 <span class="sd">            np.ndarray: n-dimensional window of the given shape and function</span>
 <span class="sd">        &quot;&quot;&quot;</span>
 
-        <span class="n">w</span> <span class="o">=</span> <span class="n">np</span><span class="o">.</span><span class="n">ones</span><span class="p">(</span><span class="n">shape</span><span class="p">)</span>
+        <span class="n">w</span> <span class="o">=</span> <span class="n">np</span><span class="o">.</span><span class="n">ones</span><span class="p">(</span><span class="n">shape</span><span class="p">,</span> <span class="n">dtype</span><span class="o">=</span><span class="bp">self</span><span class="o">.</span><span class="n">weights_dtype</span><span class="p">)</span>
         <span class="n">overlap</span> <span class="o">=</span> <span class="bp">self</span><span class="o">.</span><span class="n">tiler</span><span class="o">.</span><span class="n">_tile_overlap</span>
         <span class="k">for</span> <span class="n">axis</span><span class="p">,</span> <span class="n">length</span> <span class="ow">in</span> <span class="nb">enumerate</span><span class="p">(</span><span class="n">shape</span><span class="p">):</span>
             <span class="k">if</span> <span class="n">axis</span> <span class="o">==</span> <span class="bp">self</span><span class="o">.</span><span class="n">tiler</span><span class="o">.</span><span class="n">channel_dimension</span><span class="p">:</span>
@@ -2095,7 +2095,7 @@ <h6 id="return">Return</h6>
                 <span class="k">raise</span> <span class="ne">ValueError</span><span class="p">(</span>
                     <span class="sa">f</span><span class="s2">&quot;Window function must have the same shape as tile shape.&quot;</span>
                 <span class="p">)</span>
-            <span class="bp">self</span><span class="o">.</span><span class="n">window</span> <span class="o">=</span> <span class="n">window</span>
+            <span class="bp">self</span><span class="o">.</span><span class="n">window</span> <span class="o">=</span> <span class="n">window</span><span class="o">.</span><span class="n">astype</span><span class="p">(</span><span class="bp">self</span><span class="o">.</span><span class="n">weights_dtype</span><span class="p">)</span>
         <span class="k">else</span><span class="p">:</span>
             <span class="k">raise</span> <span class="ne">ValueError</span><span class="p">(</span>
                 <span class="sa">f</span><span class="s2">&quot;Unsupported type for window function (</span><span class="si">{</span><span class="nb">type</span><span class="p">(</span><span class="n">window</span><span class="p">)</span><span class="si">}</span><span class="s2">), expected str or np.ndarray.&quot;</span>
@@ -2363,7 +2363,7 @@ <h6 id="return">Return</h6>
 <span class="sd">            data_dtype (np.dtype): Specify data type for data buffer that stores cumulative result.</span>
 <span class="sd">                Default is `np.float32`.</span>
 
-<span class="sd">            weights_dtype (np.dtype): Specify data type for weights buffer that stores cumulative weights.</span>
+<span class="sd">            weights_dtype (np.dtype): Specify data type for weights buffer that stores cumulative weights and window array.</span>
 <span class="sd">                If you don&#39;t need precision but would rather save memory you can use `np.float16`.</span>
 <span class="sd">                Likewise, on the opposite, you can use `np.float64`.</span>
 <span class="sd">                Default is `np.float32`.</span>
@@ -2424,7 +2424,7 @@ <h6 id="args">Args</h6>
 <code>self.data_visits</code>. Can be disabled to save some memory. Default is <code>True</code>.</li>
 <li><strong>data_dtype (np.dtype):</strong>  Specify data type for data buffer that stores cumulative result.
 Default is <code>np.float32</code>.</li>
-<li><strong>weights_dtype (np.dtype):</strong>  Specify data type for weights buffer that stores cumulative weights.
+<li><strong>weights_dtype (np.dtype):</strong>  Specify data type for weights buffer that stores cumulative weights and window array.
 If you don't need precision but would rather save memory you can use <code>np.float16</code>.
 Likewise, on the opposite, you can use <code>np.float64</code>.
 Default is <code>np.float32</code>.</li>
@@ -2516,7 +2516,7 @@ <h6 id="args">Args</h6>
                 <span class="k">raise</span> <span class="ne">ValueError</span><span class="p">(</span>
                     <span class="sa">f</span><span class="s2">&quot;Window function must have the same shape as tile shape.&quot;</span>
                 <span class="p">)</span>
-            <span class="bp">self</span><span class="o">.</span><span class="n">window</span> <span class="o">=</span> <span class="n">window</span>
+            <span class="bp">self</span><span class="o">.</span><span class="n">window</span> <span class="o">=</span> <span class="n">window</span><span class="o">.</span><span class="n">astype</span><span class="p">(</span><span class="bp">self</span><span class="o">.</span><span class="n">weights_dtype</span><span class="p">)</span>
         <span class="k">else</span><span class="p">:</span>
             <span class="k">raise</span> <span class="ne">ValueError</span><span class="p">(</span>
                 <span class="sa">f</span><span class="s2">&quot;Unsupported type for window function (</span><span class="si">{</span><span class="nb">type</span><span class="p">(</span><span class="n">window</span><span class="p">)</span><span class="si">}</span><span class="s2">), expected str or np.ndarray.&quot;</span>

diff --git a/tests/test_merger.py b/tests/test_merger.py
@@ -29,11 +29,17 @@ def test_init(self):
         self.assert_(merger3.data_visits is not None)
 
         # Check data and weights dtypes
-        merger4 = Merger(tiler=tiler, data_dtype=np.float32, weights_dtype=np.float32)
+        merger4 = Merger(
+            tiler=tiler,
+            data_dtype=np.float32,
+            weights_dtype=np.float32,
+            window="boxcar",
+        )
         self.assertEqual(merger4.data.dtype, np.float32)
         self.assertEqual(merger4.data_dtype, np.float32)
         self.assertEqual(merger4.weights_sum.dtype, np.float32)
         self.assertEqual(merger4.weights_dtype, np.float32)
+        self.assertEqual(merger4.window.dtype, np.float32)
 
     def test_add(self):
         tiler = Tiler(data_shape=self.data.shape, tile_shape=(10,))

diff --git a/tiler/merger.py b/tiler/merger.py
@@ -94,7 +94,7 @@ def __init__(
             data_dtype (np.dtype): Specify data type for data buffer that stores cumulative result.
                 Default is `np.float32`.
 
-            weights_dtype (np.dtype): Specify data type for weights buffer that stores cumulative weights.
+            weights_dtype (np.dtype): Specify data type for weights buffer that stores cumulative weights and window array.
                 If you don't need precision but would rather save memory you can use `np.float16`.
                 Likewise, on the opposite, you can use `np.float64`.
                 Default is `np.float32`.
@@ -133,7 +133,7 @@ def _generate_window(self, window: str, shape: Union[Tuple, List]) -> np.ndarray
             np.ndarray: n-dimensional window of the given shape and function
         """
 
-        w = np.ones(shape)
+        w = np.ones(shape, dtype=self.weights_dtype)
         overlap = self.tiler._tile_overlap
         for axis, length in enumerate(shape):
             if axis == self.tiler.channel_dimension:
@@ -189,7 +189,7 @@ def set_window(self, window: Union[None, str, np.ndarray] = None) -> None:
                 raise ValueError(
                     f"Window function must have the same shape as tile shape."
                 )
-            self.window = window
+            self.window = window.astype(self.weights_dtype)
         else:
             raise ValueError(
                 f"Unsupported type for window function ({type(window)}), expected str or np.ndarray."