摘要：匹配器匹配非常簡(jiǎn)單，首先在第一幅圖像中選取一個(gè)關(guān)鍵點(diǎn)然后依次與第二幅圖像的每個(gè)關(guān)鍵點(diǎn)進(jìn)行描述符距離測(cè)試，最后返回距離最近的關(guān)鍵點(diǎn)對(duì)于匹配器，首先我們必須使用創(chuàng)建對(duì)象。

Feature Matching

Brute-Force匹配非常簡(jiǎn)單，首先在第一幅圖像中選取一個(gè)關(guān)鍵點(diǎn)然后依次與第二幅圖像的每個(gè)關(guān)鍵點(diǎn)進(jìn)行（描述符）距離測(cè)試，最后返回距離最近的關(guān)鍵點(diǎn).

對(duì)于BF匹配器，首先我們必須使用cv2.BFMatcher（）創(chuàng)建BFMatcher對(duì)象。它需要兩個(gè)可選的參數(shù).

第一個(gè)是normType,它指定要使用的距離測(cè)量,默認(rèn)情況下，它是cv2.NORM_L2.它適用于SIFT，SURF等（cv2.NORM_L1也在那里）.對(duì)于基于二進(jìn)制字符串的描述符，如ORB，BRIEF，BRISK等，應(yīng)使用cv2.NORM_HAMMING,使用漢明距離作為度量,如果ORB使用WTA_K == 3or4，則應(yīng)使用cv2.NORM_HAMMING2.

crossCheck：默認(rèn)值為False。如果設(shè)置為T(mén)rue，匹配條件就會(huì)更加嚴(yán)格，只有到A中的第i個(gè)特征點(diǎn)與B中的第j個(gè)特征點(diǎn)距離最近，并且B中的第j個(gè)特征點(diǎn)到A中的第i個(gè)特征點(diǎn)也是最近時(shí)才會(huì)返回最佳匹配，即這兩個(gè)特征點(diǎn)要互相匹配才行.

兩個(gè)重要的方法是BFMatcher.match（）和BFMatcher.knnMatch（）, 第一個(gè)返回最佳匹配, 第二種方法返回k個(gè)最佳匹配，其中k由用戶(hù)指定.

使用cv2.drawMatches()來(lái)繪制匹配的點(diǎn)，它會(huì)將兩幅圖像先水平排列，然后在最佳匹配的點(diǎn)之間繪制直線(xiàn)。如果前面使用的BFMatcher.knnMatch()，現(xiàn)在可以使用函數(shù)cv2.drawMatchsKnn為每個(gè)關(guān)鍵點(diǎn)和它的個(gè)最佳匹配點(diǎn)繪制匹配線(xiàn)。如果要選擇性繪制就要給函數(shù)傳入一個(gè)掩模.

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

img1 = cv2.imread("img.jpg",0)          # queryImage
img2 = cv2.imread("img1.jpg",0) # trainImage

# Initiate ORB detector
orb = cv2.ORB_create()

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

# create BFMatcher object
bf = cv2.BFMatcher(cv2.NORM_HAMMING, crossCheck=True)

# Match descriptors.
matches = bf.match(des1,des2)

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

# Draw first 10 matches.
img3 = cv2.drawMatches(img1,kp1,img2,kp2,matches[:10],None, flags=2)

plt.imshow(img3),plt.show()

matches = bf.match（des1，des2）行的結(jié)果是DMatch對(duì)象的列表。 此DMatch對(duì)象具有以下屬性：
?DMatch.distance - 描述符之間的距離。 越低越好。
?DMatch.trainIdx - 訓(xùn)練描述符中描述符的索引
?DMatch.queryIdx - 查詢(xún)描述符中描述符的索引
?DMatch.imgIdx - 訓(xùn)練圖像的索引

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

img1 = cv2.imread("img.jpg",0)          # queryImage
img2 = cv2.imread("img1.jpg",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)

# BFMatcher with default params
bf = cv2.BFMatcher()
matches = bf.knnMatch(des1,des2, k=2)

# Apply ratio test
good = []
for m,n in matches:
    if m.distance < 0.75*n.distance:
        good.append([m])

# cv.drawMatchesKnn expects list of lists as matches.
img3 = cv2.drawMatchesKnn(img1,kp1,img2,kp2,good,None,flags=2)

plt.imshow(img3),plt.show()

FLANN 代表 Fast Library for Approximate Nearest Neighbors. 它包含一組算法，這些算法針對(duì)大型數(shù)據(jù)集中的快速最近鄰搜索和高維特征進(jìn)行了優(yōu)化.對(duì)于大型數(shù)據(jù)集，它比BFMatcher工作得更快.

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

img1 = cv2.imread("img.jpg",0)          # queryImage
img2 = cv2.imread("img1.jpg",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 parameters
FLANN_INDEX_KDTREE = 1
index_params = dict(algorithm = FLANN_INDEX_KDTREE, trees = 5)
search_params = dict(checks=50)   # or pass empty dictionary

flann = cv2.FlannBasedMatcher(index_params,search_params)

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

# Need to draw only good matches, so create a mask
matchesMask = [[0,0] for i in range(len(matches))]

# ratio test as per Lowe"s paper
for i,(m,n) in enumerate(matches):
    if m.distance < 0.7*n.distance:
        matchesMask[i]=[1,0]

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()