refactor(registration): Added rigid registration of multimodal images…

… based on mean phase images and HOG features. Also added an examplary notebook

.pre-commit-config.yaml

@@ -0,0 +1,35 @@
+# See https://pre-commit.com for more information
+# See https://pre-commit.com/hooks.html for more hooks
+minimum_pre_commit_version: 1.20.0
+repos:
+-   repo: https://github.com/pre-commit/pre-commit-hooks
+    rev: v2.4.0
+    hooks:
+    -   id: trailing-whitespace
+    -   id: end-of-file-fixer
+    -   id: check-yaml
+    -   id: check-added-large-files
+    -   id: requirements-txt-fixer
+    -   id: fix-encoding-pragma
+    -   id: check-added-large-files
+    -   id: check-docstring-first
+-   repo: https://github.com/asottile/blacken-docs
+    rev: v1.3.0
+    hooks:
+    -   id: blacken-docs
+-   repo: https://github.com/pre-commit/pygrep-hooks
+    rev: v1.4.2
+    hooks:
+    -   id: python-use-type-annotations
+-   repo: https://github.com/myint/docformatter
+    rev: v1.3.1
+    hooks:
+    -   id: docformatter
+-   repo: https://github.com/pre-commit/mirrors-isort
+    rev: v4.3.21
+    hooks:
+    -   id: isort
+-   repo: https://github.com/psf/black
+    rev: 19.10b0
+    hooks:
+    -   id: black

docs/examples.rst

@@ -0,0 +1,10 @@
+========
+Examples
+========
+
+.. toctree::
+   :maxdepth: 2
+
+   examples/rigid_registration
+
+

eyepy/register/feature.py

@@ -0,0 +1,63 @@
+import numpy as np
+from skimage.feature import hog
+
+
+def hog_extract(
+    img,
+    orientations=12,
+    pixels_per_cell=(5, 5),
+    cells_per_block=(20, 20),
+    rotate_hist=False,
+    sub_sample_factor=2,
+):
+    """
+    """
+
+    hog_features = hog(
+        img,
+        orientations=orientations,
+        pixels_per_cell=pixels_per_cell,
+        cells_per_block=cells_per_block,
+        multichannel=False,
+        feature_vector=False,
+    )
+
+    ss_hog_features = hog_features[::sub_sample_factor, ::sub_sample_factor, ...]
+
+    if rotate_hist:
+        ss_hog_features = rotate_hog(ss_hog_features)
+
+    # flatten blocks
+    ss_hog_features = ss_hog_features.reshape(ss_hog_features.shape[:-3] + (-1,))
+    keys = []
+    features = []
+    for key in np.ndindex(ss_hog_features.shape[:-1]):
+        feature = ss_hog_features[key]
+        keys.append(key)
+        features.append(feature)
+
+    features = np.array(features)
+    keys = np.array(keys)
+
+    # Rescale features to original image positions
+    N0, N1 = pixels_per_cell
+    M0, M1 = cells_per_block
+    keys[:, 0] = keys[:, 0] * sub_sample_factor * N0 + (N0 * M0 / 2)
+    keys[:, 1] = keys[:, 1] * sub_sample_factor * N1 + (N1 * M1 / 2)
+
+    return keys, features
+
+
+def rotate_hog(hog_features):
+    """
+
+
+    """
+    orientations = hog_features.shape[-1]
+    M = hog_features.shape[-2]
+    rotated_hog = np.zeros(hog_features.shape[:2] + (orientations, M, M, orientations))
+    for block in np.ndindex(hog_features.shape[:2]):
+        for rot in range(orientations):
+            rotated_hog[block][rot] = np.roll(hog_features[block], rot, axis=-1)
+
+    return rotated_hog