Merge pull request #129 from jlaura/master (d1565d58) · Commits · aflab / astrogeology / Autocnet

autocnet/graph/node.py

+1 −23

Changes for autocnet/graph/node.py: 1 added line, 23 removed lines.
Original line number	Diff line number	Diff line
		@@ -226,29 +226,7 @@ class Node(dict, MutableMapping):
		kwargs passed to autocnet.feature_extractor.extract_features

		"""
		keypoint_objs, self.descriptors = fe.extract_features(array, **kwargs)
		if self.descriptors.dtype != np.float32:
		self.descriptors = self.descriptors.astype(np.float32)

		# OpenCV returned keypoint objects
		if isinstance(keypoint_objs, list):
		keypoints = np.empty((len(keypoint_objs), 7), dtype=np.float32)
		for i, kpt in enumerate(keypoint_objs):
		octave = kpt.octave & 8
		layer = (kpt.octave >> 8) & 255
		if octave < 128:
		octave = octave
		else:
		octave = (-128 \| octave)
		keypoints[i] = kpt.pt[0], kpt.pt[1], kpt.response, kpt.size, kpt.angle, octave, layer # y, x
		self._keypoints = pd.DataFrame(keypoints, columns=['x', 'y', 'response', 'size',
		'angle', 'octave', 'layer'])

		# VLFeat returned keypoint objects
		elif isinstance(keypoint_objs, np.ndarray):
		# Swap columns for value style access, vl_feat returns y, x
		keypoint_objs[:, 0], keypoint_objs[:, 1] = keypoint_objs[:, 1], keypoint_objs[:, 0].copy()
		self._keypoints = pd.DataFrame(keypoint_objs, columns=['x', 'y', 'size', 'angle'])
		self._keypoints, self.descriptors = fe.extract_features(array, **kwargs)

		def load_features(self, in_path):
		"""

autocnet/matcher/feature_extractor.py

+42 −7

Changes for autocnet/matcher/feature_extractor.py: 42 added lines, 7 removed lines.
Original line number	Diff line number	Diff line
		import cv2
		import numpy as np
		import pandas as pd

		from scipy.misc import bytescale

		try:
		import cyvlfeat as vl
		@@ -8,7 +12,7 @@ except:
		pass


		def extract_features(array, method='orb', extractor_parameters=None):
		def extract_features(array, method='orb', extractor_parameters={}):
		"""
		This method finds and extracts features from an image using the given dictionary of keyword arguments.
		The input image is represented as NumPy array and the output features are represented as keypoint IDs
		@@ -19,16 +23,20 @@ def extract_features(array, method='orb', extractor_parameters=None):
		array : ndarray
		a NumPy array that represents an image

		detector : {'orb', 'sift', 'fast', 'surf'}
		The detector method to be used
		method : {'orb', 'sift', 'fast', 'surf', 'vl_sift'}
		The detector method to be used. Note that vl_sift requires that
		vlfeat and cyvlfeat dependencies be installed.

		extractor_parameters : dict
		A dictionary containing OpenCV SIFT parameters names and values.

		Returns
		-------
		: tuple
		in the form ([list of OpenCV KeyPoints], [NumPy array of descriptors as geometric vectors])
		keypoints : DataFrame
		data frame of coordinates ('x', 'y', 'size', 'angle', and other available information)

		descriptors : ndarray
		Of descriptors
		"""

		detectors = {'fast': cv2.FastFeatureDetector_create,
		@@ -39,7 +47,34 @@ def extract_features(array, method='orb', extractor_parameters=None):
		detectors['vl_sift'] = vl.sift.sift

		if 'vl_' in method:
		return detectors[method](array, compute_descriptor=True, float_descriptors=True, **extractor_parameters)
		keypoint_objs, descriptors = detectors[method](array,
		compute_descriptor=True,
		float_descriptors=True,
		**extractor_parameters)
		# Swap columns for value style access, vl_feat returns y, x
		keypoint_objs[:, 0], keypoint_objs[:, 1] = keypoint_objs[:, 1], keypoint_objs[:, 0].copy()
		keypoints = pd.DataFrame(keypoint_objs, columns=['x', 'y', 'size', 'angle'])

		else:
		# OpenCV requires the input images to be 8-bit
		if not array.dtype == 'int8':
		array = bytescale(array)
		detector = detectors[method](**extractor_parameters)
		return detector.detectAndCompute(array, None)
		keypoint_objs, descriptors = detector.detectAndCompute(array, None)

		keypoints = np.empty((len(keypoint_objs), 7), dtype=np.float32)
		for i, kpt in enumerate(keypoint_objs):
		octave = kpt.octave & 8
		layer = (kpt.octave >> 8) & 255
		if octave < 128:
		octave = octave
		else:
		octave = (-128 \| octave)
		keypoints[i] = kpt.pt[0], kpt.pt[1], kpt.response, kpt.size, kpt.angle, octave, layer # y, x
		keypoints = pd.DataFrame(keypoints, columns=['x', 'y', 'response', 'size',
		'angle', 'octave', 'layer'])

		if descriptors.dtype != np.float32:
		descriptors = descriptors.astype(np.float32)

		return keypoints, descriptors
		No newline at end of file

autocnet/matcher/tests/test_feature_extractor.py

+4 −6

Changes for autocnet/matcher/tests/test_feature_extractor.py: 4 added lines, 6 removed lines.
Original line number	Diff line number	Diff line
		import os
		import numpy as np
		import pandas as pd
		import unittest
		from autocnet.examples import get_path
		import cv2
		@@ -25,17 +26,14 @@ class TestFeatureExtractor(unittest.TestCase):
		"sigma": 1.6}

		def test_extract_features(self):
		features = feature_extractor.extract_features(self.data_array,
		features, descriptors = feature_extractor.extract_features(self.data_array,
		method='sift',
		extractor_parameters=self.parameters)
		self.assertEquals(len(features), 2)
		self.assertIn(len(features[0]), range(8, 12))
		self.assertIsInstance(features[0][0], type(cv2.KeyPoint()))
		self.assertIsInstance(features[1][0], np.ndarray)
		self.assertEquals(len(features), 10)

		def test_extract_vlfeat(self):
		kps, descriptors = feature_extractor.extract_features(self.data_array,
		method='vl_sift',
		extractor_parameters={})
		self.assertIsInstance(kps, np.ndarray)
		self.assertIsInstance(kps, pd.DataFrame)
		self.assertEqual(descriptors.dtype, np.float32)