Added a cap on distance for matching and changed error handling on KSB

btk/metrics.py

@@ -98,12 +98,14 @@ def meas_ksb_ellipticity(image, additional_params):
     gal_image = galsim.Image(image[meas_band_num, :, :])
     gal_image.scale = pixel_scale
     shear_est = "KSB"
-    try:
-        res = galsim.hsm.EstimateShear(gal_image, psf_image, shear_est=shear_est, strict=True)
-        result = [res.corrected_g1, res.corrected_g2, res.observed_shape.e]
-    except RuntimeError as e:
-        print(e)
-        result = [-10.0, -10.0, -10.0]
+
+    res = galsim.hsm.EstimateShear(gal_image, psf_image, shear_est=shear_est, strict=False)
+    result = [res.corrected_g1, res.corrected_g2, res.observed_shape.e]
+    if res.error_message != "":
+        print(
+            f"Shear measurement error : '{res.error_message }'. \
+            This error may happen for faint galaxies or inaccurate detections."
+        )
     return result
 
 
@@ -121,7 +123,9 @@ def distance_center(true_gal, detected_gal):
     )
 
 
-def get_detection_match(true_table, detected_table, f_distance=distance_center):
+def get_detection_match(
+    true_table, detected_table, f_distance=distance_center, distance_threshold_match=5
+):
     r"""Uses the Hungarian algorithm to find optimal matching between detections and true objects.
 
     The optimal matching is computed based on the following optimization problem:
@@ -145,6 +149,8 @@ def get_detection_match(true_table, detected_table, f_distance=distance_center):
         f_distance (func): Function used to compute the distance between true and detected
             galaxies. Takes as arguments the entries corresponding to the two galaxies.
             By default the distance is the euclidean distance from center to center.
+        distance_threshold_match (float): Maximum distance for matching a detected and a
+                true galaxy.
 
     Returns:
         match_table (astropy.table.Table): Table where each row corresponds to each true
@@ -181,8 +187,9 @@ def get_detection_match(true_table, detected_table, f_distance=distance_center):
     match_indx = [-1] * len(true_table)
     dist_m = [0.0] * len(true_table)
     for i, indx in enumerate(true_indx):
-        match_indx[indx] = detected_indx[i]
-        dist_m[indx] = dist[indx][detected_indx[i]]
+        if dist[indx][detected_indx[i]] <= distance_threshold_match:
+            match_indx[indx] = detected_indx[i]
+            dist_m[indx] = dist[indx][detected_indx[i]]
 
     match_table["match_detected_id"] = match_indx
     match_table["dist"] = dist_m
@@ -534,6 +541,7 @@ def compute_metrics(  # noqa: C901
     channels_last=False,
     save_path=None,
     f_distance=distance_center,
+    distance_threshold_match=5,
 ):
     """Computes all requested metrics given information in a single batch from measure_generator.
 
@@ -572,6 +580,8 @@ def compute_metrics(  # noqa: C901
         f_distance (func): Function used to compute the distance between true and detected
             galaxies. Takes as arguments the entries corresponding to the two galaxies.
             By default the distance is the euclidean distance from center to center.
+        distance_threshold_match (float): Maximum distance for matching a detected and a
+                true galaxy.
 
     Returns:
         results (dict) : Contains all the computed metrics. Entries are :
@@ -590,7 +600,9 @@ def compute_metrics(  # noqa: C901
             deblended_images = [np.moveaxis(im, -1, 1) for im in deblended_images]
     results = {}
     matches = [
-        get_detection_match(blend_list[i], detection_catalogs[i], f_distance)
+        get_detection_match(
+            blend_list[i], detection_catalogs[i], f_distance, distance_threshold_match
+        )
         for i in range(len(blend_list))
     ]
     results["matches"] = matches
@@ -678,6 +690,7 @@ def __init__(
         noise_threshold_factor=3,
         save_path=None,
         f_distance=distance_center,
+        distance_threshold_match=5,
     ):
         """Initialize metrics generator.
 
@@ -697,10 +710,12 @@ def __init__(
                 correspond to a threshold of 3 sigmas (with sigma the standard deviation of
                 the noise)
             save_path (str): Path to directory where results will be saved. If left
-                    as None, results will not be saved.
+                as None, results will not be saved.
             f_distance (func): Function used to compute the distance between true and detected
                 galaxies. Takes as arguments the entries corresponding to the two galaxies.
                 By default the distance is the euclidean distance from center to center.
+            distance_threshold_match (float): Maximum distance for matching a detected and a
+                true galaxy.
         """
         self.measure_generator: MeasureGenerator = measure_generator
         self.use_metrics = use_metrics
@@ -709,6 +724,7 @@ def __init__(
         self.noise_threshold_factor = noise_threshold_factor
         self.save_path = save_path
         self.f_distance = f_distance
+        self.distance_threshold_match = distance_threshold_match
 
     def __next__(self):
         """Returns metric results calculated on one batch."""
@@ -744,6 +760,7 @@ def __next__(self):
                 if self.save_path is not None
                 else None,
                 f_distance=self.f_distance,
+                distance_threshold_match=self.distance_threshold_match,
             )
             metrics_results[meas_func] = metrics_results_f