OpenCV3 Python语言实现 笔记4

特征检测 特征提取 特征匹配

关键点检测+描述符
SIFT: DoG + SIFT
SURF: Hessian + SURF
ORB:  FAST + BRIEF

暴力匹配

FLANN匹配法

一、SIFT
Different of Gaussian(DoG) 对同一图像使用不同高斯滤波器
DoG得到感兴趣区域(关键点)，通过SIFT进行说明

import cv2import sysimport numpy as npimgpath = sys.argv[1]img = cv2.imread(imgpath)gray = cv2.cvtColor(img,cv2.COLOR_BGR2GRAY)sift = cv2.xfeatures2d.SIFT_create()#keypoints, descriptor = sift.detectAndCompute(gray,None)#返回关键点和描述符img = cv2.drawKeypoints(image=img, outImg=img, keypoints=keypoints,                        flags=cv2.DRAW_MATCHES_FLAGS_DRAW_RICH_KEYPOINT,                        color=(51,163,236))##cv2.DRAW_MATCHES_FLAGS_DRAW_RICH_KEYPOINT每个关键点绘制圆和方向cv2.imshow('sift_keypoints',img)while True:    if cv2.waitKey(1000/12) & 0xff == ord("c"):        breakcv2.destoryAllWindow()

关键点keypoints的属性：
1.pt: 点xy坐标
2.size: 特征直径
3.angle: 特征方向
4.response: 特征强度 评估特征好坏
5.octave: 特征所在金字塔层级
6.class_id: 关键点ID

二、SURF

surf = cv2.xfeatures2d.SURF_create(8000)#阈值越小特征点越多keypoints, descriptor = surf.detectAndCompute(gray,None)#返回关键点和描述符img = cv2.drawKeypoints(image=img, outImg=img, keypoints=keypoints,                        flags=4,                        color=(51,163,236))#

三、ORB特征检测与特征匹配

import numpy as npimport cv2from matplotlib import pyplot as pltcv2.ocl.setUseOpenCL(False)# query and test imagesimg1 = cv2.imread('../images/manowar_logo.png',0)img2 = cv2.imread('../images/manowar_single.jpg',0)# create the ORB detectororb = cv2.ORB_create()kp1, des1 = orb.detectAndCompute(img1,None)kp2, des2 = orb.detectAndCompute(img2,None)# brute force matchingbf = cv2.BFMatcher(cv2.NORM_HAMMING, crossCheck=True)#暴力匹配matches = bf.match(des1,des2)##matches = bf.knnMatch(des1,des2,k=2)#knn匹配# Sort by distance.matches = sorted(matches, key = lambda x:x.distance)img3 = cv2.drawMatches(img1,kp1,img2,kp2, matches[:25], img2,flags=2)#img3 = cv2.drawMatchesKnn(img1,kp1,img2,kp2, matches[:25], img2,flags=2)#knnplt.imshow(img3),plt.show()

四、FLANN特征匹配

flann = cv2.FlannBasedMatcher(indexParams,searchParams)
FLANN匹配器两个参数 python 字典形式 c++ 结构体形式
indexParams:LinearIndex、KTreeIndex、KMeansIndex、CompositeIndex、AutotuneIndexS
searchParams:checks=50 指定索引树遍历次数

import numpy as npimport cv2from matplotlib import pyplot as pltqueryImage = cv2.imread('../images/bathory_album.jpg',0)trainingImage = cv2.imread('../images/bathory_vinyls.jpg',0)# create SIFT and detect/computesift = cv2.xfeatures2d.SIFT_create()kp1, des1 = sift.detectAndCompute(queryImage,None)kp2, des2 = sift.detectAndCompute(trainingImage,None)# FLANN matcher parameters# FLANN_INDEX_KDTREE = 0indexParams = dict(algorithm = 0, trees = 5)searchParams = dict(checks=50)   # or pass empty dictionaryflann = cv2.FlannBasedMatcher(indexParams,searchParams)matches = flann.knnMatch(des1,des2,k=2)# prepare an empty mask to draw good matchesmatchesMask = [[0,0] for i in xrange(len(matches))]# David G. Lowe's ratio test, populate the maskfor i,(m,n) in enumerate(matches):#enumerate枚举列举 例 enumerate(['a',(2,3)])                                                         0 'a'                                                        1 (2,3)    if m.distance < 0.7*n.distance:#匹配过滤 丢弃距离大于0.7的值        matchesMask[i]=[1,0]drawParams = dict(matchColor = (0,255,0),                  singlePointColor = (255,0,0),                  matchesMask = matchesMask,                  flags = 0)resultImage = cv2.drawMatchesKnn(queryImage,kp1,trainingImage,kp2,matches,None,**drawParams)#收集参数 **变量存放的方式是字典 *元组plt.imshow(resultImage,),plt.show()

五、FLANN单应性匹配

单应性：是一个条件，该条件表示两幅图像当中一幅出现投影畸变时，彼此还能匹配

import numpy as npimport cv2from matplotlib import pyplot as pltMIN_MATCH_COUNT = 10#计算单应性至少需要4个匹配img1 = cv2.imread('images/bb.jpg',0)img2 = cv2.imread('images/color2_small.jpg',0)# Initiate SIFT detectorsift = cv2.xfeatures2d.SIFT_create()# find the keypoints and descriptors with SIFTkp1, des1 = sift.detectAndCompute(img1,None)kp2, des2 = sift.detectAndCompute(img2,None)FLANN_INDEX_KDTREE = 0index_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:    #m.queryIdx测试图像描述符des和对应kp的下标 m.trainIdx样本图像。。    src_pts = np.float32([ kp1[m.queryIdx].pt for m in good ]).reshape(-1,1,2)#三维 -1:自动计算多少个一行两列    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()#ravel()多维降为一维    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)#投影映射    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 = Nonedraw_params = dict(matchColor = (0,255,0), # draw matches in green color                   singlePointColor = None,                   matchesMask = matchesMask, # draw only inliers                   flags = 2)img3 = cv2.drawMatches(img1,kp1,img2,kp2,good,None,**draw_params)plt.imshow(img3, 'gray'),plt.show()

六、基于纹身取证的应用程序

1.将图像描述符保存为npy文件

import osimport cv2import numpy as npfrom os import walkfrom os.path import joinimport sysdef create_descriptors(folder):  files = []  for (dirpath, dirnames, filenames) in walk(folder):    files.extend(filenames)  for f in files:    save_descriptor(folder, f, cv2.xfeatures2d.SIFT_create())def save_descriptor(folder, image_path, feature_detector):  print "reading %s" % image_path  if image_path.endswith("npy"):    return  img = cv2.imread(join(folder, image_path), 0)  keypoints, descriptors = feature_detector.detectAndCompute(img, None)  descriptor_file = image_path.replace("jpg", "npy")##  np.save(join(folder, descriptor_file), descriptors)##dir = sys.argv[1]create_descriptors(dir)

2.扫描匹配

from os.path import joinfrom os import walkimport numpy as npimport cv2from sys import argv# create an array of filenamesfolder = argv[1]query = cv2.imread(join(folder, "tattoo_seed.jpg"), 0)# create files, images, descriptors globalsfiles = []images = []descriptors = []for (dirpath, dirnames, filenames) in walk(folder):  files.extend(filenames)  for f in files:    if f.endswith("npy") and f != "tattoo_seed.npy":      descriptors.append(f)  print descriptors# create the sift detectorsift = cv2.xfeatures2d.SIFT_create()query_kp, query_ds = sift.detectAndCompute(query, None)# create FLANN matcherFLANN_INDEX_KDTREE = 0index_params = dict(algorithm = FLANN_INDEX_KDTREE, trees = 5)search_params = dict(checks = 50)flann = cv2.FlannBasedMatcher(index_params, search_params)# minimum number of matchesMIN_MATCH_COUNT = 10potential_culprits = {}print ">> Initiating picture scan..."for d in descriptors:  print "--------- analyzing %s for matches ------------" % d  matches = flann.knnMatch(query_ds, np.load(join(folder, d)), k =2)  good = []  for m,n in matches:      if m.distance < 0.7*n.distance:          good.append(m)  if len(good) > MIN_MATCH_COUNT:        print "%s is a match! (%d)" % (d, len(good))  else:    print "%s is not a match" % d  potential_culprits[d] = len(good)max_matches = Nonepotential_suspect = Nonefor culprit, matches in potential_culprits.iteritems():  if max_matches == None or matches > max_matches:    max_matches = matches    potential_suspect = culpritprint "potential suspect is %s" % potential_suspect.replace("npy", "").upper()