cv2.threshold不适用于白色背景，我们可以反转极性来创建黑色背景。
我们可以使用thresh = 254，并得到白色的输出。
更有用的解决方案是在使用cv2.threshold之前反转极性（反转白色）。
假设gray_image的dtype是np.uint8，那么白色等于255，黑色等于0。
反转极性：inv_gray_image = 255 - gray_image的数据。
代码示例：

import cv2
import numpy as np

gray_image = cv2.imread('gray_image.jpg', cv2.IMREAD_GRAYSCALE)  # Read image in grayscale format

# Invert black and white before applying the threshold
inv_gray_image = 255 - gray_image

thresh = 1  # 1 is just an example - we may use thresh = 100...

_, thresholded = cv2.threshold(inv_gray_image, thresh, 255, cv2.THRESH_BINARY)

cv2.imwrite('thresholded.png', thresholded)  # Save image for testing

字符串
thresholded：

的数据
为了找到面积，我们可以使用cv2.findContours，并找到最大轮廓的边界矩形。
范例：

import cv2
import numpy as np

gray_image = cv2.imread('gray_image.jpg', cv2.IMREAD_GRAYSCALE)  # Read image in grayscale format

# Invert black and white before applying the threshold
inv_gray_image = 255 - gray_image  # Valid if gray_image.dtype is uint8
# inv_gray_image = np.iinfo(gray_image.dtype).max - gray_image

thresh = 1

_, thresholded = cv2.threshold(inv_gray_image, thresh, 255, cv2.THRESH_BINARY)

contours = cv2.findContours(thresholded, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)[0]  # Find contours
c = max(contours, key=cv2.contourArea)  # Find largest contour

x, y, w, h = cv2.boundingRect(c)  # Get the bounding rectangle
bgr_image = cv2.cvtColor(gray_image, cv2.COLOR_GRAY2BGR)  # Convert to BGR for drawing colored rectangle
cv2.rectangle(bgr_image, (x, y), (x + w, y + h), (0, 255, 0), 3)  # Draw green rectangle.
cv2.imwrite('bgr_image.png', bgr_image)  # Save image for testing

型
输出量：

的