详解如何使用opencv实现图片相似度检测

  import cv2

  import numpy as np

  def calculate_complexity_similarity(img1_path, img2_path):

  # 加载两张图片

  img1 = cv2.imread(img1_path)

  img2 = cv2.imread(img2_path)

  # 将图片转换为灰度图像

  gray_img1 = cv2.cvtColor(img1, cv2.COLOR_BGR2GRAY)

  gray_img2 = cv2.cvtColor(img2, cv2.COLOR_BGR2GRAY)

  # 创建ORB特征检测器

  orb = cv2.ORB_create()

  # 检测特征点和描述符

  kp1, des1 = orb.detectAndCompute(gray_img1, None)

  kp2, des2 = orb.detectAndCompute(gray_img2, None)

  # 将描述符类型转换为CV_8U

  des1 = np.uint8(des1)

  des2 = np.uint8(des2)

  # 创建暴力匹配器

  bf = cv2.BFMatcher(cv2.NORM_HAMMING, crossCheck=True)

  # 进行特征匹配

  matches = bf.match(des1, des2)

  similarity = 0.0

  # 根据特征点匹配结果计算相似度

  if len(matches) > 0:

  similarity = sum([match.distance for match in matches]) / len(matches)

  print('图片相似度为:', similarity)

  else:

  print('未找到匹配的特征点')

  return similarity

  def calculate_histogram_similarity(img1_path, img2_path):

  # 读取两张图片

  img1 = cv2.imread(img1_path, cv2.IMREAD_GRAYSCALE)

  img2 = cv2.imread(img2_path, cv2.IMREAD_GRAYSCALE)

  # 计算直方图

  hist1 = cv2.calcHist([img1], [0], None, [256], [0, 256])

  hist2 = cv2.calcHist([img2], [0], None, [256], [0, 256])

  # 归一化直方图

  cv2.normalize(hist1, hist1, 0, 1, cv2.NORM_MINMAX)

  cv2.normalize(hist2, hist2, 0, 1, cv2.NORM_MINMAX)

  # 比较直方图

  similarity = cv2.compareHist(hist1, hist2, cv2.HISTCMP_CORREL)

  if similarity < 0.2:

  return similarity

  # 将第二个图片切割为4个区域并分别计算相似度

  h, w = img2.shape[:2]

  img2_1 = img2[:h // 2, :w // 2]

  img2_2 = img2[:h // 2, w // 2:]

  img2_3 = img2[h // 2:, :w // 2]

  img2_4 = img2[h // 2:, w // 2:]

  sim_list = []

  # 创建ORB特征检测器

  orb = cv2.ORB_create()

  kp1, des1 = orb.detectAndCompute(img1, None)

  # 计算四个区域的相似度

  for img in [img2_1, img2_2, img2_3, img2_4]:

  kp2, des2 = orb.detectAndCompute(img, None)

  if des1 is None or des2 is None:

  print('未找到足够的特征点')

  sim_list.append(0.1)

  else:

  des1 = np.uint8(des1)

  des2 = np.uint8(des2)

  bf = cv2.BFMatcher(cv2.NORM_HAMMING, crossCheck=True)

  matches = bf.match(des1, des2)

  orb_similarity = 0.0

  if len(matches) > 0:

  orb_similarity = sum([match.distance for match in matches]) / len(matches)

  sim_list.append(orb_similarity)

  max_sim = max(sim_list)

  print('图片相似度为:', max_sim)

  return max_sim

  if __name__ == '__main__':

  img1_path = 'pic/straw1.png'

  img2_path = 'pic/straw2.png'

  # 加载两张图片

  img1 = cv2.imread(img1_path)

  img2 = cv2.imread(img2_path)

  # 计算ORB特征相似度

  orb_similarity = calculate_complexity_similarity(img1_path, img2_path)

  # 计算直方图相似度

  hist_similarity = calculate_histogram_similarity(img1_path, img2_path)

  # 将第二张图片切割为4个区域

  h, w = img2.shape[:2]

  img2_1 = img2[:h // 2, :w // 2]

  img2_2 = img2[:h // 2, w // 2:]

  img2_3 = img2[h // 2:, :w // 2]

  img2_4 = img2[h // 2:, w // 2:]

  # 在图片上绘制绿色框和相似度

  cv2.rectangle(img2, (0, 0), (w // 2, h // 2), (0, 255, 0), 2)

  cv2.putText(img2, f" {orb_similarity:.2f}", (10, 30), cv2.FONT_HERSHEY_SIMPLEX, 1, (0, 255, 0), 2)

  cv2.rectangle(img2, (w // 2, 0), (w, h // 2), (0, 255, 0), 2)

  cv2.putText(img2, f" {hist_similarity:.2f}", (w // 2 + 10, 30), cv2.FONT_HERSHEY_SIMPLEX, 1, (0, 255, 0), 2)

  cv2.rectangle(img2, (0, h // 2), (w // 2, h), (0, 255, 0), 2)

  cv2.putText(img2, f" {orb_similarity:.2f}", (10, h // 2 + 30), cv2.FONT_HERSHEY_SIMPLEX, 1, (0, 255, 0), 2)

  cv2.rectangle(img2, (w // 2, h // 2), (w, h), (0, 255, 0), 2)

  cv2.putText(img2, f" {hist_similarity:.2f}", (w // 2 + 10, h // 2 + 30), cv2.FONT_HERSHEY_SIMPLEX, 1, (0, 255, 0), 2)

  # 显示结果图片

  cv2.namedWindow('Image', cv2.WINDOW_NORMAL)

  cv2.resizeWindow('Image', 800, 600)

  cv2.imshow('Image', cv2.resize(img2, (0, 0), fx=0.5, fy=0.5))

  cv2.waitKey(0)

  cv2.destroyAllWindows()