opencv-python 提取sift特征并匹配的實例

# -*- coding: utf-8 -*-
import cv2
import numpy as np
from find_obj import filter_matches,explore_match
from matplotlib import pyplot as plt
 
def getSift():
  '''
  得到并查看sift特征
  '''
  img_path2 = '../../data/home.jpg'
  #讀取圖像
  img = cv2.imread(img_path2)
  #轉(zhuǎn)換為灰度圖
  gray= cv2.cvtColor(img,cv2.COLOR_BGR2GRAY)
  #創(chuàng)建sift的類
  sift = cv2.SIFT()
  #在圖像中找到關(guān)鍵點 也可以一步計算#kp, des = sift.detectAndCompute
  kp = sift.detect(gray,None)
  print type(kp),type(kp[0])
  #Keypoint數(shù)據(jù)類型分析 http://www.cnblogs.com/cj695/p/4041399.html
  print kp[0].pt
  #計算每個點的sift
  des = sift.compute(gray,kp)
  print type(kp),type(des)
  #des[0]為關(guān)鍵點的list，des[1]為特征向量的矩陣
  print type(des[0]), type(des[1])
  print des[0],des[1]
  #可以看出共有885個sift特征，每個特征為128維
  print des[1].shape
  #在灰度圖中畫出這些點
  img=cv2.drawKeypoints(gray,kp)
  #cv2.imwrite('sift_keypoints.jpg',img)
  plt.imshow(img),plt.show()
 
def matchSift():
  '''
  匹配sift特征
  '''
  img1 = cv2.imread('../../data/box.png', 0) # queryImage
  img2 = cv2.imread('../../data/box_in_scene.png', 0) # trainImage
  sift = cv2.SIFT()
  kp1, des1 = sift.detectAndCompute(img1, None)
  kp2, des2 = sift.detectAndCompute(img2, None)
  # 蠻力匹配算法,有兩個參數(shù)，距離度量(L2(default),L1)，是否交叉匹配(默認false)
  bf = cv2.BFMatcher()
  #返回k個最佳匹配
  matches = bf.knnMatch(des1, des2, k=2)
  # cv2.drawMatchesKnn expects list of lists as matches.
  #opencv2.4.13沒有drawMatchesKnn函數(shù)，需要將opencv2.4.13\sources\samples\python2下的common.py和find_obj文件放入當(dāng)前目錄，并導(dǎo)入
  p1, p2, kp_pairs = filter_matches(kp1, kp2, matches)
  explore_match('find_obj', img1, img2, kp_pairs) # cv2 shows image
  cv2.waitKey()
  cv2.destroyAllWindows()
 
def matchSift3():
  '''
  匹配sift特征
  '''
  img1 = cv2.imread('../../data/box.png', 0) # queryImage
  img2 = cv2.imread('../../data/box_in_scene.png', 0) # trainImage
  sift = cv2.SIFT()
  kp1, des1 = sift.detectAndCompute(img1, None)
  kp2, des2 = sift.detectAndCompute(img2, None)
  # 蠻力匹配算法,有兩個參數(shù)，距離度量(L2(default),L1)，是否交叉匹配(默認false)
  bf = cv2.BFMatcher()
  #返回k個最佳匹配
  matches = bf.knnMatch(des1, des2, k=2)
  # cv2.drawMatchesKnn expects list of lists as matches.
  #opencv3.0有drawMatchesKnn函數(shù)
  # Apply ratio test
  # 比值測試，首先獲取與A 距離最近的點B（最近）和C（次近），只有當(dāng)B/C
  # 小于閾值時（0.75）才被認為是匹配，因為假設(shè)匹配是一一對應(yīng)的，真正的匹配的理想距離為0
  good = []
  for m, n in matches:
    if m.distance < 0.75 * n.distance:
      good.append([m])
  img3 = cv2.drawMatchesKnn(img1, kp1, img2, kp2, good[:10], None, flags=2)
  cv2.drawm
  plt.imshow(img3), plt.show()
 
matchSift()