Feature Mapping using OpenCV's ORB detector and Brute-force Matching

feature_mapping_orb_opencv.py

import cv2

image1 = cv2.imread("1.jpg")
# Loaded first image in full color
image2 = cv2.imread("2.jpg")
# Loaded second image in full color

image1_gray = cv2.cvtColor(image1, cv2.COLOR_BGR2GRAY)
# Created gray version of first image
image2_gray = cv2.cvtColor(image2, cv2.COLOR_BGR2GRAY)
# Created gray version of second image

orb_detector = cv2.ORB_create()
#created our ORB(Oriented FAST and Rotated BRIEF) detector

key_points1, descriptors1 = orb_detector.detectAndCompute(image1_gray, None)
# obtained key points and descriptions for first gray image
key_points2, descriptors2 = orb_detector.detectAndCompute(image2_gray, None)
# obtained key points and descriptions for second gray image

brute_force_matcher = cv2.BFMatcher(cv2.NORM_HAMMING, crossCheck= True)
# Created OpenCV brute force matcher instance with a average hamming distance value of 6 (cv2.NORM_HAMMING = 6)

matches = brute_force_matcher.match(descriptors1, descriptors2)
# Computed the matches across the descriptors of the two gay images

matches = sorted(matches, key = lambda x:x.distance)
# Sorted the matches in order of their hamming distance

new_image = cv2.drawMatches(image1, key_points1, image2, key_points2, matches[:10], outImg=None, flags=2)
"""
Created a new image by
    - combining the two original colored images with the corresponding key_points of their gray versions
      i.e image1, key_points1, image2, key_points2 . Note that you can combine with the gray images instead too
    - display only the top 10 connect points, can be greater or lesser values
      i.e  matches[:10]
    - set a None value for 'outImg'
"""

# The code below displays the new image
# Image can be displayed using Matplotlib's PyPlot as well.
cv2.imshow("Window1", new_image)
cv2.waitKey(0)
cv2.destroyAllWindows()

See the example images for the code

You should see result like the one below.