OpenCV draw non-matching points

问题

I have followed OpenCV Feature Detection and Description tutorial and used SIFT and other algorithms in OpenCV to find matching feature points between 2 images. From what i understood, these algorithms can find the similar regions between 2 images. But i am interested in identifying the different or dis-similar regions.

How can i draw all the NON-MATCHING feature points on both the images? Further more, can i draw boundaries around these non-matching points to be able to show which regions in the 2 images are different?

I am using Python code on Windows 7 and build from latest OpenCV source.

回答1:

Draw all the NON-MATCHING feature points on both the images:

This task is pretty straight forward once you know the structure of the Matcher objects resulting from the match of two descriptors (matches = bf.match(des1,des2)). The two Matcher objects' properties relevant to this problem are the following:

DMatch.trainIdx: Index of the descriptor (or keypoint from the train image) in train descriptors
DMatch.queryIdx: Index of the descriptor (or keypoint from the query image) in query descriptors

Then, knowing this information and as @uzair_syed said, this is just a simple list operations task.

Draw boundaries around the non-matching points:

To achieve this, I would do something like this:

Create a black mask with a white pixel for each non-matching points
Depending on the density of the non-matching point's cluster, dilate the mask with a big kernel (i.e. 15 x 15 px).
Erode the mask with the same kernel's size.
Finally, apply the findContours function on the mask to get the boundaries of the non-matching points.

For more information, you can have a look at this question and its answer.

Hope it gets you on the right track!

回答2:

It turned out to be simple list operations task. Here is my Python code

# code copied from
# http://docs.opencv.org/3.0-beta/doc/py_tutorials/py_feature2d/py_feature_homography/py_feature_homography.html

import numpy as np
import cv2
from matplotlib import pyplot as plt
from scipy.spatial.distance import euclidean

MIN_MATCH_COUNT = 10

img1 = cv2.imread('Src.png',0)  # queryImage
img2 = cv2.imread('Dest.png',0) # trainImage

# Initiate SIFT detector
sift = cv2.xfeatures2d.SIFT_create()

# find the keypoints and descriptors with SIFT
kp1, des1 = sift.detectAndCompute(img1,None)
kp2, des2 = sift.detectAndCompute(img2,None)

FLANN_INDEX_KDTREE = 0
index_params = dict(algorithm = FLANN_INDEX_KDTREE, trees = 5)
search_params = dict(checks = 50)

flann = cv2.FlannBasedMatcher(index_params, search_params)

matches = flann.knnMatch(des1,des2,k=2)

# store all the good matches as per Lowe's ratio test.
good = []
for m,n in matches:
    if m.distance < 0.7*n.distance:
        good.append(m)

if len(good)>MIN_MATCH_COUNT:
    src_pts = np.float32([ kp1[m.queryIdx].pt for m in good ]).reshape(-1,1,2)
    dst_pts = np.float32([ kp2[m.trainIdx].pt for m in good ]).reshape(-1,1,2)

    kp1_matched=([ kp1[m.queryIdx] for m in good ])
    kp2_matched=([ kp2[m.trainIdx] for m in good ])

    kp1_miss_matched=[kp for kp in kp1 if kp not in kp1_matched]
    kp2_miss_matched=[kp for kp in kp2 if kp not in kp2_matched]

    # draw only miss matched or not matched keypoints location
    img1_miss_matched_kp = cv2.drawKeypoints(img1,kp1_miss_matched, None,color=(255,0,0), flags=0)
    plt.imshow(img1_miss_matched_kp),plt.show()

    img2_miss_matched_kp = cv2.drawKeypoints(img2,kp2_miss_matched, None,color=(255,0,0), flags=0)
    plt.imshow(img2_miss_matched_kp),plt.show()

    M, mask = cv2.findHomography(src_pts, dst_pts, cv2.RANSAC,5.0)
    matchesMask = mask.ravel().tolist()

    h,w = img1.shape
    pts = np.float32([ [0,0],[0,h-1],[w-1,h-1],[w-1,0] ]).reshape(-1,1,2)
    dst = cv2.perspectiveTransform(pts,M)

else:
    print "Not enough matches are found - %d/%d" % (len(good),MIN_MATCH_COUNT)
    matchesMask = None

来源：https://stackoverflow.com/questions/43775842/opencv-draw-non-matching-points

标签

python

OpenCV

sift