第六章 图像检索以及基于图像描述符的搜索
通过提取特征进行图像的匹配与搜索
1 特征检测算法
常见的特征和提取算法:
Harris 检测角点
Sift 检测斑点(blob) 有专利保护
Surf 检测斑点 有专利保护
Fast 检测角点
Brief 检测斑点
Orb 带方向的fast算法和具有旋转不变性的brief算法
特征的定义
#!/usr/bin/env python # -*- coding: utf-8 -*- # @Time : 2016/12/5 12:30 # @Author : Retacn # @Site : 检测图像的角点 # @File : cornerHarris.py # @Software: PyCharm __author__ = "retacn" __copyright__ = "property of mankind." __license__ = "CN" __version__ = "0.0.1" __maintainer__ = "retacn" __email__ = "zhenhuayue@sina.com" __status__ = "Development" import cv2 import numpy as np # 读入图像 img = cv2.imread('../test1.jpg') # 转换颜色空间 gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY) gray = np.float32(gray) # 检测图像角点 dst = cv2.cornerHarris(gray, 2, 23, # sobel算子的中孔,3-31之间的奇数 0.04) # 将检测到有角点标记为红色 img[dst > 0.01 * dst.max()] = [0, 0, 255] while (True): cv2.imshow("corners", img) if cv2.waitKey(33) & 0xFF == ord('q'): break cv2.destroyAllWindows()
使用dog和sift进行特征提取和描述
示例代码如下:
import cv2 import sys import numpy as py # 读入图像 # imgpath=sys.argv[1] imgpath = '../test1.jpg' img = cv2.imread(imgpath) # 更换颜色空间 gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY) # 创建sift对象,计算灰度图像,会使用dog检测角点 sift = cv2.xfeatures2d.SIFT_create() keypoints, descriptor = sift.detectAndCompute(gray, None) # print(keypoints) # 关键点有以下几个属性 # angle 表示特征的方向 # class_id 关键点的id # octave 特征所在金字塔的等级 # pt 图像中关键点的坐标 # response 表示关键点的强度 # size 表示特征的直径 img = cv2.drawKeypoints(image=img, outImage=img, keypoints=keypoints, color=(51, 163, 236), flags=cv2.DRAW_MATCHES_FLAGS_DRAW_RICH_KEYPOINTS) # 显示图像 cv2.imshow('sift_keypoints', img) while (True): if cv2.waitKey(int(1000 / 12)) & 0xFF == ord('q'): break cv2.destroyAllWindows()
使用心有快速hessian算法和SURF来提取特征
示例代码发如下:
#!/usr/bin/env python # -*- coding: utf-8 -*- # @Time : 2016/12/10 17:30 # @Author : Retacn # @Site : sift用于检测斑点 # @File : sift.py # @Software: PyCharm __author__ = "retacn" __copyright__ = "property of mankind." __license__ = "CN" __version__ = "0.0.1" __maintainer__ = "retacn" __email__ = "zhenhuayue@sina.com" __status__ = "Development" import cv2 import sys import numpy as py # 读入图像 # imgpath=sys.argv[1] # alg=sys.argv[2] # threshold=sys.argv[3] imgpath = '../test1.jpg' img = cv2.imread(imgpath) # alg = 'SURF' alg = 'SIFT' # threshold = '8000' # 阈值越小特征点越多 threshold = '4000' def fd(algorithm): if algorithm == 'SIFT': return cv2.xfeatures2d.SIFT_create() if algorithm == 'SURF': # return cv2.xfeatures2d.SURF_create(float(threshold) if len(sys.argv) == 4 else 4000) return cv2.xfeatures2d.SURF_create(float(threshold)) # 更换颜色空间 gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY) # 创建sift对象,计算灰度图像,会使用dog检测角点 fd_alg = fd(alg) keypoints, descriptor = fd_alg.detectAndCompute(gray, None) # print(keypoints) # 关键点有以下几个属性 # angle 表示特征的方向 # class_id 关键点的id # octave 特征所在金字塔的等级 # pt 图像中关键点的坐标 # response 表示关键点的强度 # size 表示特征的直径 img = cv2.drawKeypoints(image=img, outImage=img, keypoints=keypoints, color=(51, 163, 236), flags=cv2.DRAW_MATCHES_FLAGS_DRAW_RICH_KEYPOINTS) # 显示图像 cv2.imshow('keypoints', img) while (True): if cv2.waitKey(int(1000 / 12)) & 0xFF == ord('q'): break cv2.destroyAllWindows()
基于ORB的特征检测和特征匹配
ORB是基于
FAST(featuresfrom accelerated segment test)关键点检测技术
在像素周围绘制一个圆,包含16个像素
BRIEF(binaryrobust independent elementary features) 描述符
暴力（brute-force）匹配法
比较两个描述符,并产生匹配结果
ORB特征匹配
示例代码如下:
import numpy as np import cv2 from matplotlib import pyplot as plt cv2.ocl.setUseOpenCL(False) # 读入灰度图像 img1 = cv2.imread('../test2_part.jpg', cv2.IMREAD_GRAYSCALE) img2 = cv2.imread('../test2.jpg', cv2.IMREAD_GRAYSCALE) # 创建orb特征检测器和描述符 orb = cv2.ORB_create() kp1, des1 = orb.detectAndCompute(img1, None) kp2, des2 = orb.detectAndCompute(img2, None) # 暴力匹配 bf = cv2.BFMatcher(cv2.NORM_HAMMING, crossCheck=True) matches = bf.match(des1, des2) matches = sorted(matches, key=lambda x: x.distance) # 显示图像 img3 = cv2.drawMatches(img1, kp1, img2, kp2, matches[:40], img2, flags=2) plt.imshow(img3), plt.show()
报如下错误:
cv2.error: D:BuildOpenCVopencv-3.1.0modulespythonsrc2cv2.cpp:163:error: (-215) The data should normally be NULL! in functionNumpyAllocator::allocate
解决办法,添加如下代码 :
cv2.ocl.setUseOpenCL(False)
k最邻近配匹
import numpy as np import cv2 from matplotlib import pyplot as plt cv2.ocl.setUseOpenCL(False) # 读入灰度图像 img1 = cv2.imread('../test2_part.jpg', cv2.IMREAD_GRAYSCALE) img2 = cv2.imread('../test2.jpg', cv2.IMREAD_GRAYSCALE) # 创建orb特征检测器和描述符 orb = cv2.ORB_create() kp1, des1 = orb.detectAndCompute(img1, None) kp2, des2 = orb.detectAndCompute(img2, None) # knn匹配,返回k个匹配 bf = cv2.BFMatcher(cv2.NORM_L1, crossCheck=False) matches = bf.knnMatch(des1, des2, k=2) # 显示图像 img3 = cv2.drawMatchesKnn(img1, kp1, img2, kp2, matches, img2, flags=2) plt.imshow(img3), plt.show()
Flann匹配法
Fast library for approximate nearestneighbors 近似最近邻的快速库
import numpy as np import cv2 from matplotlib import pyplot as plt # 读入图像 queryImage = cv2.imread('../test2_part.jpg', cv2.IMREAD_GRAYSCALE) trainingImage = cv2.imread('../test2.jpg', cv2.IMREAD_GRAYSCALE) # 创建sift对象 sift = cv2.xfeatures2d.SIFT_create() kp1, des1 = sift.detectAndCompute(queryImage, None) kp2, des2 = sift.detectAndCompute(trainingImage, None) FLANN_INDEX_KDTREE = 0 # 创建字典参数 indexParams = dict(algorithm=FLANN_INDEX_KDTREE, trees=5) # 处理索引 searchParams = dict(checks=50) # 创建对象,用来指定索引树的遍历次数 flann = cv2.FlannBasedMatcher(indexParams, searchParams) matches = flann.knnMatch(des1, des2, k=2) matchesMask = [[0, 0] for i in range(len(matches))] for i, (m, n) in enumerate(matches): if m.distance < 0.7 * n.distance: 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) plt.imshow(resultImage), plt.show()
运行结果如下:
Flann单应性匹配
示例代码如下:
#!/usr/bin/env python # -*- coding: utf-8 -*- # @Time : 2016/12/11 12:02 # @Author : Retacn # @Site : flann的单应性匹配 # @File : flann_homography.py # @Software: PyCharm __author__ = "retacn" __copyright__ = "property of mankind." __license__ = "CN" __version__ = "0.0.1" __maintainer__ = "retacn" __email__ = "zhenhuayue@sina.com" __status__ = "Development" import numpy as np import cv2 from matplotlib import pyplot as plt MIN_MATCH_COUNT = 10 # 读入图像 img1 = cv2.imread('../test3_part.jpg', cv2.IMREAD_GRAYSCALE) img2 = cv2.imread('../test3.jpg', cv2.IMREAD_GRAYSCALE) # 创建sift对象 sift = cv2.xfeatures2d.SIFT_create() # 查询特征点和描述符 kp1, des1 = sift.detectAndCompute(img1, None) kp2, des2 = sift.detectAndCompute(img2, None) FLANN_INDEX_KDTREE = 0 # 创建字典参数 indexParams = dict(algorithm=FLANN_INDEX_KDTREE, trees=5) # 处理索引 searchParams = dict(checks=50) # 创建对象,用来指定索引树的遍历次数 flann = cv2.FlannBasedMatcher(indexParams, searchParams) matches = flann.knnMatch(des1, des2, k=2) good = [] for m, n in matches: if m.distance < 0.7 * n.distance: good.append(m) if len(good) > MIN_MATCH_COUNT: # 在原始图像和训练图像中查询特征点 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) # 单应性 M, mask = cv2.findHomography(src_pts, dst_pts, cv2.RANSAC, 5.0) matchesMask = mask.ravel().tolist() # 对第二张图片计算相对于原始图像的投影畸变,并绘制边框 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 # 显示图像 draw_params = dict(matchColor=(0, 255, 0), # 绿线 singlePointColor=None, matchesMask=matchesMask, flags=2) img3 = cv2.drawMatches(img1, kp1, img2, kp2, good, None, **draw_params) plt.imshow(img3, 'gray'), plt.show()
运行结果如下:
基于纹身取证的应用程序示例
A 将图像描述符保存到文件中
#!/usr/bin/env python # -*- coding: utf-8 -*- # @Time : 2016/12/11 13:52 # @Author : Retacn # @Site : 将图像描述符保存到文件中 # @File : generate_descriptors.py # @Software: PyCharm __author__ = "retacn" __copyright__ = "property of mankind." __license__ = "CN" __version__ = "0.0.1" __maintainer__ = "retacn" __email__ = "zhenhuayue@sina.com" __status__ = "Development" import cv2 import numpy as np from os import walk from os.path import join import sys # 创建描述符 def 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()) # 保存描述符 def save_descriptor(folder, image_path, feature_detector): print("reading %s" % image_path) if image_path.endswith("npy") or image_path.endswith("avi"): return img = cv2.imread(join(folder, image_path), cv2.IMREAD_GRAYSCALE) keypoints, descriptors = feature_detector.detectAndCompute(img, None) descriptor_file = image_path.replace("jpg", "npy") np.save(join(folder, descriptor_file), descriptors) # 从执行参数中取得文件目录 dir = sys.argv[1] create_descriptors(dir)
B 扫描匹配
#!/usr/bin/env python # -*- coding: utf-8 -*- # @Time : 2016/12/11 14:05 # @Author : Retacn # @Site : 扫描匹配 # @File : scan4matches.py # @Software: PyCharm __author__ = "retacn" __copyright__ = "property of mankind." __license__ = "CN" __version__ = "0.0.1" __maintainer__ = "retacn" __email__ = "zhenhuayue@sina.com" __status__ = "Development" from os.path import join from os import walk import numpy as np import cv2 from sys import argv from matplotlib import pyplot as plt # 创建文件名数组 folder = argv[1] query = cv2.imread(join(folder, 'part.jpg'), cv2.IMREAD_GRAYSCALE) # 创建全局的文件,图片,描述符 files = [] images = [] descriptors = [] for (dirpath, dirnames, filenames) in walk(folder): files.extend(filenames) for f in files: if f.endswith('npy') and f != 'part.npy': descriptors.append(f) print(descriptors) # 创建sift检测器 sift = cv2.xfeatures2d.SIFT_create() query_kp, query_ds = sift.detectAndCompute(query, None) # 创建flann匹配 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) # 最小匹配数 MIN_MATCH_COUNT = 10 potential_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 = None potential_suspect = None for culprit, matches in potential_culprits.items(): if max_matches == None or matches > max_matches: max_matches = matches potential_suspect = culprit print("potential suspect is %s" % potential_suspect.replace("npy", "").upper())