First of all, sorry for the title, I didn't know how to name it.
Now, I am using openCV to compare 2 images.
The code has been taken from here: https://docs.opencv.org/3.0-beta/doc/py_tutorials/py_feature2d/py_matcher/py_matcher.html#flann-based-matcher
So in my code I have:
orb = cv2.ORB_create()
kp1, des1 = orb.detectAndCompute(image, None)
This is for the descriptors/keypoints to be calculated.
For the Flann matcher I have the following:
FLANN_INDEX_LSH = 6
index_params = dict(algorithm=FLANN_INDEX_LSH, trees=5)
search_params = dict(checks=50)
flann = cv2.FlannBasedMatcher(index_params, search_params)
And this is the same code for the second image:
kp2, des2 = orb.detectAndCompute(image2, None)
matches = flann.knnMatch(des1, des2, k=2)
And I also saw on some sources that this is how to achieve Lowe's ratio test:
good = []
for m_n in matches:
if len(m_n) != 2:
continue
(m, n) = m_n
if m.distance < acc * n.distance:
good.append(m)
Later I concat both images doing this:
matches = [item for sublist in matches for item in sublist]
#This is needed because matches is a list of lists.
img3 = cv2.drawMatches(image, kp1, target, kp2, matches[:20], None, flags=2)
img4 = cv2.drawMatches(image, kp1, target, kp2, good[:20], None, flags=2)
And I get this image as result:
The upper half is what matches has, and the other is good.
Could you explain me what is the difference between both please?
I have looked here and there and I still don't know what I am doing to build good, and is sad just copy pasting some code without knowing what is it.
