detect a object by feature matching using FLANN

asked 2017-10-08 14:46:54 -0500

saif92 gravatar image

I have 3 images generated from a CAD file. The images are different angles of various components. I have following python code that finds similair points between two images.

import numpy as np
import cv2
from matplotlib import pyplot as plt

MIN_MATCH_COUNT = 4

img1 = cv2.imread('mutter_g_seite.png', 0)          # queryImage
img2 = cv2.imread('Bauteile/6.jpg', 0) # trainImage
img1 = cv2.resize(img1, (800,600))
img2 = cv2.resize(img2, (800,600))

# Edge detection
img1 = cv2.Canny(img1, 50, 50)
img2 = cv2.Canny(img2, 50, 50)                                            

# Laplacian
#img1 = cv2.Laplacian(img1, cv2.CV_64F)
#img2 = cv2.Laplacian(img2, cv2.CV_64F)

# 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 parameters
FLANN_INDEX_KDTREE = 0
index_params = dict(algorithm = FLANN_INDEX_KDTREE, trees = 5)
search_params = dict(checks=100)   # or pass empty dictionary

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:
        print(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)

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

    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)

    img2 = cv2.polylines(img2,[np.int32(dst)],True,255,3, cv2.LINE_AA)

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

draw_params = dict(matchColor = (0,255,0),
                   singlePointColor = (255,0,0),
                   matchesMask = matchesMask,
                   flags = 0)

img3 = cv2.drawMatchesKnn(img1,kp1,img2,kp2,matches,None,**draw_params)

plt.imshow(img3)
plt.show()

Queryimage 1

image description

Queryimage 2

image description

Queryimage 3

image description

Train Image

image description

So as you can see I have 3 query images where I want to detect if my train image contains the object. I have set a min_match of points to define if the object is on my image. I have read that Lowe's ratio and calculating the homograph help to get better results. Thats what I have been trying to do but I have encountered following issues:

  1. I can't find good result with my current approach.
  2. I'm comparing only 1 query image. Is it possible to compare all 3 at once and get there result?
  3. If I reduce the min_match to 2 (in my examples I get 2 good matches with query image 2) I try to build the homograph but I get following error:

119.74974060058594 127.00901641845702 135.53598022460938 168.57069396972656 [0, 0] OpenCV Error: Assertion failed (scn + 1 == m.cols) in perspectiveTransform, file /tmp/opencv-20170825-90583-1pdhamg/opencv-3.3.0/modules/core/src/matmul.cpp, line 2299 Traceback (most recent call last): File "flann-feature-matching.py", line 53, in <module> dst = cv2 ...

(more)
edit retag flag offensive close merge delete