Merge pull request #22 from LSSTDESC/#16

#16

README.md

@@ -2,7 +2,8 @@
 [![Documentation Status](https://readthedocs.org/projects/blendingtoolkit/badge/?version=latest)](https://blendingtoolkit.readthedocs.io/en/latest/?badge=latest)
 
 # BlendingToolKit
-Tools to create blend catalogs, produce training samples and implement blending metrics.
+Framework for fast generation and analysis of galaxy blends catalogs. This toolkit is a convenient way of
+producing multi-band postage stamp images of blend scenes.
 
 Documentation can be found at https://blendingtoolkit.readthedocs.io/en/latest/
 

btk/compute_metrics.py

@@ -4,13 +4,13 @@
 
 
 class Metrics_params(object):
-    def __init__(self, meas_generator, param):
+    def __init__(self, meas_generator, sim_param):
         """Class describing functions to return results of
          detection/deblending/measurement algorithm in meas_generator. Each
          blend results yielded by the meas_generator for a batch.
     """
         self.meas_generator = meas_generator
-        self.param = param
+        self.sim_param = sim_param
 
     def get_detections(self):
         """
@@ -31,9 +31,9 @@ def get_detections(self):
             and 'dy' respectively, in pixels from bottom left corner as (0, 0).
         """
         # Astropy table with entries corresponding to true sources
-        true_tables = [astropy.table.Table()] * self.param.batch_size
+        true_tables = [astropy.table.Table()] * self.config.batch_size
         # Astropy table with entries corresponding to detections
-        detected_tables = [astropy.table.Table()] * self.param.batch_size
+        detected_tables = [astropy.table.Table()] * self.config.batch_size
         return true_tables, detected_tables
 
     def get_segmentation(self):
@@ -150,19 +150,19 @@ def get_detection_match(true_table, detected_table):
     norm_dist = dist/norm_size[:, np.newaxis]
     detected_table['dSigma_min'] = np.min(norm_dist, axis=0)
     detected_table['d_min'] = np.min(dist, axis=0)
-    detection_threshold1 = 0.5
+    detection_threshold1 = 5
     condlist1 = [
-        np.min(norm_dist, axis=0) <= detection_threshold1,
-        np.min(norm_dist, axis=0) > detection_threshold1]
-    choicelist1 = [np.argmin(norm_dist, axis=0), -1]
+        np.min(dist, axis=0) <= detection_threshold1,
+        np.min(dist, axis=0) > detection_threshold1]
+    choicelist1 = [np.argmin(dist, axis=0), -1]
     match_id1 = np.select(condlist1, choicelist1)
     detected_table['match_true_id1'] = match_id1
     detected_table['match_galtileid1'] = true_table['galtileid'][match_id1]
-    detection_threshold2 = 5
+    detection_threshold2 = 0.5
     condlist2 = [
-        np.min(dist, axis=0) <= detection_threshold2,
-        np.min(dist, axis=0) > detection_threshold2]
-    choicelist2 = [np.argmin(dist, axis=0), -1]
+        np.min(norm_dist, axis=0) <= detection_threshold2,
+        np.min(norm_dist, axis=0) > detection_threshold2]
+    choicelist2 = [np.argmin(norm_dist, axis=0), -1]
     match_id2 = np.select(condlist2, choicelist2)
     detected_table['match_true_id2'] = match_id2
     detected_table['match_galtileid2'] = true_table['galtileid'][match_id2]
@@ -317,13 +317,13 @@ def run(Metrics_params, test_size=1000, dSigma_detection=True):
         batch_detection_result = Metrics_params.get_detections()
         if (
             len(batch_detection_result[0]) != len(batch_detection_result[1]) or
-            len(batch_detection_result[0]) != Metrics_params.param.batch_size
+            len(batch_detection_result[0]) != Metrics_params.sim_param.batch_size
            ):
             raise ValueError("Metrics_params.get_detections output must be "
                              "two lists of astropy table of length batch size."
                              f" Found {len(batch_detection_result[0])}, "
                              f"{len(batch_detection_result[1])}, "
-                             f"{ Metrics_params.param.batch_size}")
+                             f"{ Metrics_params.sim_param.batch_size}")
         true_table, detected_table, detection_summary = evaluate_detection(
             batch_detection_result[0], batch_detection_result[1],
             batch_index=i)

btk/config.py

@@ -16,6 +16,7 @@ class Simulation_params(object):
             in the sky background during a full exposure.
         verbose: If true, prints description at multiple steps.
     """
+
     def __init__(self, catalog_name, max_number=2,
                  batch_size=8, stamp_size=24,
                  survey_name="LSST",

btk/draw_blends.py

@@ -78,7 +78,8 @@ def draw_isolated(Args, galaxy, iso_obs):
         no_margin=False,
         verbose_render=False)
     iso_render_engine.render_galaxy(
-        galaxy, no_partials=True, calculate_bias=False, no_analysis=True)
+        galaxy, variations_x=None, variations_s=None, variations_g=None,
+        no_fisher=True, calculate_bias=False, no_analysis=True)
     return iso_obs
 
 

btk/plot_utils.py

@@ -26,7 +26,7 @@ def get_rgb(image, min_val=None, max_val=None):
     if max_val is None:
         max_val = image.max(axis=-1).max(axis=-1)
     new_image = np.transpose(image, axes=(1, 2, 0))
-    new_image = (new_image - min_val) / (max_val - min_val)*255
+    new_image = (new_image - min_val) / (max_val - min_val) * 255
     new_image[new_image < 0] = 0
     new_image[new_image > 255] = 255
     return new_image.astype(np.uint8)
@@ -47,8 +47,8 @@ def get_rgb_image(image, normalize_with_image=None):
         uint8 array [height, width, bands] of the input image.
     """
     try:
-        import scarlet
-        if normalize_with_image:
+        import scarlet.display
+        if normalize_with_image is not None:
             norm = scarlet.display.Asinh(img=normalize_with_image, Q=20)
         else:
             norm = None
@@ -94,7 +94,7 @@ def plot_blends(blend_images, blend_list, detected_centers=None,
         raise ValueError(f"band_indices must be a list with 3 entries, not \
             {band_indices}")
     if detected_centers is None:
-        detected_centers = [[]]*batch_size
+        detected_centers = [[]] * batch_size
     if (len(detected_centers) != batch_size or
             blend_images.shape[0] != batch_size):
         raise ValueError(f"Length of detected_centers and length of blend_list\
@@ -163,15 +163,17 @@ def plot_with_isolated(blend_images, isolated_images, blend_list,
         raise ValueError(f"band_indices must be a list with 3 entries, not \
             {band_indices}")
     if detected_centers is None:
-        detected_centers = [[]]*b_size
+        detected_centers = [[]] * b_size
     if (len(detected_centers) != b_size or len(isolated_images) != b_size or
             blend_images.shape[0] != b_size):
         raise ValueError(f"Length of detected_centers and length of blend_list\
             must be equal to first dimension of blend_images, found \
             {len(detected_centers), len(blend_list), len(blend_images)}")
     for i in range(len(blend_list)):
         images = np.transpose(blend_images[i], axes=(2, 0, 1))
-        blend_img_rgb = get_rgb_image(images[band_indices])
+        blend_img_rgb = get_rgb_image(
+            images[band_indices],
+            normalize_with_image=images[band_indices])
         plt.figure(figsize=(2, 2))
         plt.imshow(blend_img_rgb)
         plt.title(f"{len(blend_list[i])} objects")
@@ -184,7 +186,7 @@ def plot_with_isolated(blend_images, isolated_images, blend_list,
         plt.show()
         iso_blend = isolated_images[i]
         num = iso_blend.shape[0]
-        plt.figure(figsize=(2*num, 2))
+        plt.figure(figsize=(2 * num, 2))
         for j in range(num):
             iso_images = np.transpose(iso_blend[j], axes=(2, 0, 1))
             iso_img_rgb = get_rgb_image(
@@ -256,10 +258,10 @@ def plot_metrics_summary(summary, num, ax=None):
     ax.imshow(results_table, origin='left', cmap=plt.cm.Blues)
     ax.set_xlabel("# true objects")
     # Don't print zero'th column
-    ax.set_xlim([0.5, num+0.5])
+    ax.set_xlim([0.5, num + 0.5])
     ax.set_ylabel("# correctly detected objects")
-    ax.set_xticks(np.arange(1, num+1, 1.0))
-    ax.set_yticks(np.arange(0, num+2, 1.0))
+    ax.set_xticks(np.arange(1, num + 1, 1.0))
+    ax.set_yticks(np.arange(0, num + 2, 1.0))
     for (j, i), label in np.ndenumerate(results_table):
         if i == 0:
             # Don't print efficiency for zero'th column
@@ -270,7 +272,7 @@ def plot_metrics_summary(summary, num, ax=None):
         ax.text(i, j, f"{label:.1f}%",
                 ha='center', va='center', color=color)
         if i == j:
-            rect = patches.Rectangle((i-0.5, j-0.5), 1, 1, linewidth=2,
+            rect = patches.Rectangle((i - 0.5, j - 0.5), 1, 1, linewidth=2,
                                      edgecolor='mediumpurple',
                                      facecolor='none')
             ax.add_patch(rect)