'''
https://github.com/opencv/opencv/blob/master/samples/python/find_obj.py
https://www.pyimagesearch.com/2016/01/11/opencv-panorama-stitching/
'''
import numpy as np
import cv2 as cv
FLANN_INDEX_LSH = 6
def anorm2(a):
return (a*a).sum(-1)
def anorm(a):
return np.sqrt( anorm2(a) )
def matchKeypoints(keyPoints1, keyPoints2, descriptors1, descriptors2):
flann_params= dict(algorithm = FLANN_INDEX_LSH,
table_number = 6, # 12
key_size = 12, # 20
multi_probe_level = 1) #2
matcher = cv.FlannBasedMatcher(flann_params, {}) # bug : need to pass empty dict (#1329)
raw_matches = matcher.knnMatch(descriptors1, descriptors2, k = 2) #2
matches = []
for m in raw_matches:
if len(m) == 2 and m[0].distance < m[1].distance * 0.79:
matches.append((m[0].trainIdx, m[0].queryIdx))
if len(matches) >= 4:
keyPoints1 = np.float32([keyPoints1[i] for (_, i) in matches])
keyPoints2 = np.float32([keyPoints2[i] for (i, _) in matches])
H, status = cv.findHomography(keyPoints1, keyPoints2, cv.RANSAC,4.0)
print('%d / %d inliers/matched' % (np.sum(status), len(status)))
else:
H, status = None, None
print('%d matches found, not enough for homography estimation' % len(p1))
return matches, H, status
def drawMatches(image1, image2, keyPoints1, keyPoints2, matches, status):
h1, w1 = image1.shape[:2]
h2, w2 = image2.shape[:2]
img_matching_result = np.zeros((max(h1, h2), w1 + w2, 3), dtype="uint8")
img_matching_result[0:h2, 0:w2] = image2
img_matching_result[0:h1, w2:] = image1
for ((trainIdx, queryIdx), s) in zip(matches, status):
if s == 1:
keyPoint2 = (int(keyPoints2[trainIdx][0]), int(keyPoints2[trainIdx][1]))
keyPoint1 = (int(keyPoints1[queryIdx][0]) + w2, int(keyPoints1[queryIdx][1]))
cv.line(img_matching_result, keyPoint1, keyPoint2, (0, 255, 0), 1)
return img_matching_result
def main():
img1 = cv.imread('.\\images\\B.jpg')
img2 = cv.imread('.\\images\\A.jpg')
gray1 = cv.cvtColor(img1, cv.COLOR_BGR2GRAY)
gray2 = cv.cvtColor(img2, cv.COLOR_BGR2GRAY)
detector = cv.BRISK_create()
keyPoints1, descriptors1 = detector.detectAndCompute(gray1, None)
keyPoints2, descriptors2 = detector.detectAndCompute(gray2, None)
print('img1 - %d features, img2 - %d features' % (len(keyPoints1), len(keyPoints2)))
keyPoints1 = np.float32([keypoint.pt for keypoint in keyPoints1])
keyPoints2 = np.float32([keypoint.pt for keypoint in keyPoints2])
matches, H, status = matchKeypoints(keyPoints1, keyPoints2, descriptors1, descriptors2)
img_matching_result = drawMatches(img1, img2, keyPoints1, keyPoints2, matches, status)
result = cv.warpPerspective(img1, H,
(img1.shape[1] + img2.shape[1], img1.shape[0]))
result[0:img2.shape[0], 0:img2.shape[1]] = img2
cv.imshow('result', result)
cv.imshow('matching result', img_matching_result)
cv.waitKey()
print('Done')
if __name__ == '__main__':
print(__doc__)
main()
cv.destroyAllWindows() |
