使用单应性和筛选算法在两个图像之间进行差异检测

Question

我想检测两个图像img1和img2之间的差异，并获得类似desired output的输出。

我已经尝试使用筛选算法来检测两个图像的特征，然后提取出良好的匹配项。

 sift = cv.xfeatures2d.SIFT_create()
 kp1, des1 = sift.detectAndCompute(img1,None)
 kp2, des2 = sift.detectAndCompute(img2,None)
 FLANN_INDEX_KDTREE = 5
 index_params = dict(algorithm = FLANN_INDEX_KDTREE, trees = 5)
 search_params = dict(checks = 2)
 flann = cv.FlannBasedMatcher(index_params, search_params)
 matches = flann.knnMatch(des1,des2,k=2)
 good = []
for m,n in matches:
    if m.distance < 0.8*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 = cv.findHomography(src_pts, dst_pts, cv.RANSAC,5.0)
      matchesMask = mask.ravel().tolist()
      h,w= img1.shape
      pts = np.float32([ [0,0],[0,h-1],[w-1,h-1],[w-1,0] ]).reshape(-1,1,2)
      dst = cv.perspectiveTransform(pts,M)
      img2 = cv.polylines(img2,[np.int32(dst)],True,255,3, cv.LINE_AA)

我得到了结果result

有什么想法要从该算法中获得区别，而不是好的匹配吗？

Answer 1

您可以进行透视变形以使一个图像与另一个图像对齐。然后计算结构相似性指标度量（ssim）作为差异。然后对其进行阈值处理，获得轮廓，最后获得轮廓的边界框，并将其绘制在输入图像上。这是使用Python / OpenCV / Skimage的方法。注意，我尚未安装SIFT，因此将在其位置使用ORB。

输入1：

输入2：

import cv2
import numpy as np
from skimage.metrics import structural_similarity


MAX_FEATURES = 500
GOOD_MATCH_PERCENT = 0.15


def alignImages(im1, im2):

  # im2 is reference and im1 in to be warped to match im2

  # Convert images to grayscale
  im1Gray = cv2.cvtColor(im1, cv2.COLOR_BGR2GRAY)
  im2Gray = cv2.cvtColor(im2, cv2.COLOR_BGR2GRAY)

  # Detect ORB features and compute descriptors.
  orb = cv2.ORB_create(MAX_FEATURES)
  keypoints1, descriptors1 = orb.detectAndCompute(im1Gray, None)
  keypoints2, descriptors2 = orb.detectAndCompute(im2Gray, None)

  # Match features.
  matcher = cv2.DescriptorMatcher_create(cv2.DESCRIPTOR_MATCHER_BRUTEFORCE_HAMMING)
  matches = matcher.match(descriptors1, descriptors2, None)

  # Sort matches by score
  matches.sort(key=lambda x: x.distance, reverse=False)

  # Remove not so good matches
  numGoodMatches = int(len(matches) * GOOD_MATCH_PERCENT)
  matches = matches[:numGoodMatches]

  # Draw top matches
  imMatches = cv2.drawMatches(im1, keypoints1, im2, keypoints2, matches, None)
  cv2.imwrite("pipes_matches.png", imMatches)

  # Extract location of good matches
  points1 = np.zeros((len(matches), 2), dtype=np.float32)
  points2 = np.zeros((len(matches), 2), dtype=np.float32)

  for i, match in enumerate(matches):
    points1[i, :] = keypoints1[match.queryIdx].pt
    points2[i, :] = keypoints2[match.trainIdx].pt

  # Find homography
  h, mask = cv2.findHomography(points1, points2, cv2.RANSAC)

  # Use homography
  height, width, channels = im2.shape
  im1Reg = cv2.warpPerspective(im1, h, (width, height))

  return im1Reg, h


if __name__ == '__main__':

  # Read reference image
  refFilename = "pipes1.jpg"
  print("Reading reference image : ", refFilename)
  imReference = cv2.imread(refFilename, cv2.IMREAD_COLOR)

  # Read image to be aligned
  imFilename = "pipes2.jpg"
  print("Reading image to align : ", imFilename);  
  im = cv2.imread(imFilename, cv2.IMREAD_COLOR)

  # Aligned image will be stored in imReg. 
  # The estimated homography will be stored in h. 
  imReg, h = alignImages(im, imReference)

  # Write aligned image to disk. 
  outFilename = "pipes2_aligned.jpg"
  print("Saving aligned image : ", outFilename); 
  cv2.imwrite(outFilename, imReg)

  # Print estimated homography
  print("Estimated homography : \n",  h)

  # Convert images to grayscale, blur, get (inverted) ssim differences and threshold
  refGray = cv2.cvtColor(imReference, cv2.COLOR_BGR2GRAY)
  imGray = cv2.cvtColor(imReg, cv2.COLOR_BGR2GRAY)
  refBlur = cv2.GaussianBlur(refGray, (3,3), 0)
  imBlur = cv2.GaussianBlur(imGray, (3,3), 0)    
  (score, diff) = structural_similarity(imBlur, refBlur, full=True)
  diff = ((1-diff) * 255).astype("uint8")
  thresh = cv2.threshold(diff,210,255,cv2.THRESH_BINARY)[1]

  # Filter using contour area and draw contour bounding boxes on input image
  cnts = cv2.findContours(thresh, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)
  cnts = cnts[0] if len(cnts) == 2 else cnts[1]
  result = imReference.copy()
  for c in cnts:
      area = cv2.contourArea(c)
      if area > 150:
          x,y,w,h = cv2.boundingRect(c)
          cv2.rectangle(result, (x, y), (x+w, y+h), (0, 0, 255), 2)


  # show images
  cv2.imshow('reference', imReference)
  cv2.imshow('image', im)
  cv2.imshow('image_aligned', imReg)
  cv2.imshow('diff', diff)
  cv2.imshow('thresh', thresh)
  cv2.imshow('result', result)
  cv2.waitKey()

  # save images
  cv2.imwrite('pipes_diff.png', diff)
  cv2.imwrite('pipes_thresh.png', thresh)
  cv2.imwrite('pipes_result.png', result)

符合图片：

图像2（分别）与图像1对齐：

SSIM差异图片：

阈值模拟图像：

结果：

根据需要调整各种阈值和面积过滤器参数。