Python 使用QR码作为图像中的定位点进行透视扭曲 问题摘要

我有下图，其中有两个二维码始终位于固定位置，如下所示。它还具有从用户处获取的预定义感兴趣区域，如图1中的绿色矩形所示。现在我想做的是，当我看到一个以一定角度拍摄的图像（image2）时，我想使用QR码作为参考锚点进行透视扭曲，以便刻度盘中的读数适合红色边界框。下面给出了图像的示例。下面还给出了用于绘制用户定义的感兴趣区域的原始图像

我试过的 步骤1：用户选择ROI并保存输出

import cv2 
import numpy as np  
import pyzbar.pyzbar as pyzbar

 is_drawing = False
 ix = -1
 iy = -1

 image_two=cv2.imread('cal_img.jpeg')

 dim = (1536, 1024)
 #resize image
 resized = cv2.resize(image_two, dim, interpolation = cv2.INTER_AREA) 



 def draw_rectangle(event, x, y, flags, param):
    global is_drawing, ix, iy

   if event == cv2.EVENT_LBUTTONDOWN:
    is_drawing = True
    ix = x
    iy = y
elif event == cv2.EVENT_LBUTTONUP:
    is_drawing = False
    cv2.rectangle(img=resized, pt1=(ix, iy), pt2=(x, y), color=(0, 255, 0), thickness=-1)
    ix = -1
    iy = -1
elif event == cv2.EVENT_MOUSEMOVE:
    if is_drawing:
        cv2.rectangle(img=resized, pt1=(ix, iy), pt2=(x, y), color=(0, 255, 0), thickness=-1)


        cv2.namedWindow("image")
        cv2.setMouseCallback("image", draw_rectangle)

      while True:
        cv2.imshow("image", resized)

# wait 5 seconds then wait for ESC key
      if cv2.waitKey(5) & 0xFF == 27:
         break
         cv2.imwrite('image.jpg', resized)
         cv2.destroyAllWindows()

步骤2：找到校准ROI的像素坐标

         coordinates = []  # list of the green rectangle coordinates
         green_color = [0, 255, 0]
         last_x_green, last_y_green = 0, 0  # store the last x and y green positions

         # reads the image in the color mode
         image_four = cv2.imread('image.jpg', 1)

        rows, cols, _ = image_four.shape  

        for x in range(rows):
            for y in range(cols):
                px = list(image_four[x, y])
                if px == green_color:
        # find the first coordinate of the green rectangle (top left corner)
        if len(coordinates) == 0:
            coordinates.append((y, x))  # top left corner of the green rectangle

        last_x_green, last_y_green = x, y

        coordinates.append((last_y_green, last_x_green))

        top_left = coordinates[0]  # (x1, y1)
        bottom_left = (coordinates[0][0], coordinates[1][1])  # (x1, y2)
        top_right = (coordinates[1][0], coordinates[0][1])  # (x2, y1)
        bottom_right = coordinates[1]

        print('The coordinates of the green rectangle, from left to right and from top to bottom, are:')
        print(f'Top Left: {top_left}, Top Right: {top_right}, Bottom Left: {bottom_left}, Bottom Right: {bottom_right}')

第三步：阅读二维码

      decodedObjects =pyzbar.decode(image_one)

      print('The Decoded QR codeis:',decodedObjects[0].data)
      print('The Location of the Top Left QR Code is:',decodedObjects[0].rect)
      print('The Location of the Bottom Right QR Code is:',decodedObjects[1].rect)

问题

---

由于二维码在两幅图像中的位置保持不变，我无法确定如何在第二幅图像中获得ROI的四个角（绿色框），以进行透视变换。

绘制线，定义二维码的角并延伸这些线。然后得到外部直线组的交点。这将定义一个旋转的矩形。请注意，如果您有4个QR码（或者只使用十字或更简单的符号也可以），这将更加容易，因为透视变换需要4组共轭坐标