如何使用 OpenCV 或其他图像库在图像上打印文本而不重叠？

Question

我正在使用 Google OCR API 从手写文本图像中提取文本。但我想将这些文本打印在另一个空白图像中，它们的坐标分别在原始图像上。

def draw_location(ver, image, word_text):
    font = cv2.FONT_HERSHEY_SIMPLEX
    x = (int)((ver[0].x+ver[1].x) / 2 )
    y = (int)((ver[0].y+ver[3].y) / 2 )
    cv2.putText(image, word_text, (x,y), font, 2, (250, 10, 50), 1, cv2.LINE_AA)
 

Answer 1

你可以使用这个逻辑。

import cv2
import numpy as np

your_text = ['I', 'am', 'extracting', 'text', 'from', 'handwritten', 'text', 'images', 'using', 'Google', 'OCR', 'API.',
             'but', 'I', 'want', 'to', 'print', 'those', 'text', 'in', 'another', 'blank', 'image', 'with',
             'respective', 'their', 'coordinates', 'on', 'the', 'original', 'image.']

blank_image = np.zeros((400, 400))
height, width = blank_image.shape

FONT_SCALE = 0.6
FONT = cv2.FONT_HERSHEY_SIMPLEX
FONT_THICKNESS = 1
word_x, word_y = 20, 20  # Origin cordinate of text
height_spacing = 5  # space between 2 string


for word in your_text:
    (label_width, label_height), baseline = cv2.getTextSize(word + " ", FONT, FONT_SCALE, FONT_THICKNESS)

    if word_x + label_width > width:
        word_x = 0
        word_y += label_height + height_spacing

    cv2.putText(blank_image, word + " ", (word_x, word_y), FONT, FONT_SCALE, 255, FONT_THICKNESS)
    word_x += label_width

cv2.imshow("text image", blank_image)
cv2.waitKey(0)
cv2.destroyAllWindows()