Target tracker amend (#37)

tests/imaging_target_tracker_tests.py

@@ -34,7 +34,13 @@ def test_with_certain_match(self):
            target_1, target_2 
         ]
 
-        desc = TargetDescription(1,3,3,7)
+        desc = TargetDescription(
+                    np.eye(13)[1],
+                    np.eye(35)[3],
+                    np.eye(8)[3],
+                    np.eye(8)[7],
+                )
+
         assert tracker.closest_match(desc) is target_1
 
     def test_with_near_match(self):
@@ -67,6 +73,11 @@ def test_with_near_match(self):
            target_1, target_2 
         ]
 
-        desc = TargetDescription(2,2,2,7)
+        desc = TargetDescription(
+                    np.eye(13)[2],
+                    np.eye(35)[2],
+                    np.eye(8)[2],
+                    np.eye(8)[7],
+                )
         assert tracker.closest_match(desc) is target_2
 

uavf_2024/imaging/utils.py

@@ -6,11 +6,22 @@
 
 from .imaging_types import TargetDescription, Tile
 
+
+def normalize_distribution( b: TargetDescription, offset):
+    shape_norm_prob = (b.shape_probs + offset)/sum(b.shape_probs + offset)
+    letter_norm_prob = (b.letter_probs + offset)/sum(b.letter_probs + offset)
+    shape_col_norm_prob = (b.shape_col_probs + offset)/sum(b.shape_col_probs + offset)
+    letter_col_norm_prob = (b.letter_col_probs + offset)/sum(b.letter_col_probs + offset)
+
+    return TargetDescription(shape_norm_prob, letter_norm_prob, shape_col_norm_prob, letter_col_norm_prob)
+
+
 def calc_match_score(a: TargetDescription, b: TargetDescription):
         '''
         Returns a number between 0 and 1 representing how likely the two descriptions are the same target
         
         '''
+        b = normalize_distribution(b, 1e-4)
         shape_score = sum(a.shape_probs * b.shape_probs)
         letter_score = sum(a.letter_probs * b.letter_probs)
         shape_color_score = sum(a.shape_col_probs * b.shape_col_probs)