major upd : asynchronous intake

src/deep_neurographs/feature_extraction.py

@@ -164,42 +164,14 @@ def generate_img_profiles(neurograph, path, anisotropy=[1.0, 1.0, 1.0]):
     )
     img = utils.normalize_img(img)
     for edge in neurograph.mutable_edges:
-        if neurograph.optimize_alignment or neurograph.optimize_path:
-            xyz = to_img_coords(neurograph, edge)
-            path = geometry_utils.sample_path(xyz, N_PROFILE_POINTS)
-        else:
-            i, j = tuple(edge)
-            xyz_i = utils.world_to_img(neurograph, i)
-            xyz_j = utils.world_to_img(neurograph, j)
-            path = geometry_utils.make_line(xyz_i, xyz_j, N_PROFILE_POINTS)
+        xyz_i, xyz_j = neurograph.get_edge_attr("xyz", edge)
+        xyz_i = utils.world_to_img(neurograph, xyz_i)
+        xyz_j = utils.world_to_img(neurograph, xyz_j)
+        path = geometry_utils.make_line(xyz_i, xyz_j, N_PROFILE_POINTS)
         features[edge] = geometry_utils.get_profile(img, path, window=WINDOW)
     return features
 
 
-def to_img_coords(neurograph, edge):
-    """
-    Converts xyz coordinate of each vertex in "edge" from real world to image
-    coordinates.
-
-    Parameters
-    ----------
-    neurograph : NeuroGraph
-        NeuroGraph generated from a directory of swcs generated from a
-        predicted segmentation.
-    edge : frozenset
-        The edge from "neurograph" to perform coordinate transformation.
-
-    Returns
-    -------
-    img_coords : numpy.ndarray
-        Image coordinates of each vertex from "edge".
-
-    """
-    xyz_list = neurograph.edges[edge]["xyz"]
-    img_coords = [utils.world_to_img(neurograph, xyz) for xyz in xyz_list]
-    return np.array(img_coords)
-
-
 def generate_mutable_skel_features(neurograph):
     features = dict()
     for edge in neurograph.mutable_edges:

src/deep_neurographs/geometry_utils.py

@@ -9,9 +9,9 @@
 
 
 # Directional Vectors
-def get_directional(neurograph, i, proposal_tangent, window=5):
+def get_directional(neurograph, i, proposal_tangent, window=5, n_svd_points=10):
     directionals = []
-    d = neurograph.optimize_depth
+    d = n_svd_points
     for branch in neurograph.get_branches(i):
         if branch.shape[0] >= window + d:
             xyz = deepcopy(branch[d:, :])

src/deep_neurographs/graph_utils.py

@@ -8,6 +8,15 @@
 Routines that extract the irreducible components of a graph.
 
 --define what an irreducible is
+    leafs : set
+        Nodes with degreee 1.
+    junctions : set
+        Nodes with degree > 2.
+    edges : dict
+        Set of edges connecting nodes in leafs and junctions. The keys are
+        pairs of nodes connected by an edge and values are a dictionary of
+        attributes.
+
 --define what a branch is
 
 """
@@ -41,14 +50,10 @@ def get_irreducibles(swc_dict, prune=True, depth=16, smooth=True):
 
     Returns
     -------
-    leafs : set
-        Nodes with degreee 1.
-    junctions : set
-        Nodes with degree > 2.
-    edges : dict
-        Set of edges connecting nodes in leafs and junctions. The keys are
-        pairs of nodes connected by an edge and values are a dictionary of
-        attributes.
+    dict
+        Irreducibles stored in a dictionary where key-values are type of
+        irreducible (i.e. leaf, junction, or edge) and corresponding set of
+        all irreducibles from the graph of that type.
 
     """
     # Initializations
@@ -80,7 +85,7 @@ def get_irreducibles(swc_dict, prune=True, depth=16, smooth=True):
             nbs = append_value(nbs, root, j)
             nbs = append_value(nbs, j, root)
             root = None
-    return leafs, junctions, edges
+    return {"leafs": leafs, "junctions": junctions, "edges": edges}
 
 
 def get_irreducible_nodes(graph):