使用Pytesseract / OpenCV绘制边界框

Question

我正在使用pytesseract（0.3.2）和openCV（4.1.2）识别图像中的数字。虽然image_to_string正常工作，但image_to_data和image_to_boxes不行。我需要能够在图像上绘制边界框，但这让我感到困惑。我尝试了不同的图像、旧版本的pytesseract等，但都没有解决问题。我正在使用Windows和Jupyter Notebooks。

import cv2 
import pytesseract

#erosion
def erode(image):
    kernel = np.ones((5,5),np.uint8)
    return cv2.erode(image, kernel, iterations = 1)

#grayscale
def get_grayscale(image):
    return cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)

#thresholding
def thresholding(image):
    #return cv2.adaptiveThreshold(image, 255, cv2.ADAPTIVE_THRESH_GAUSSIAN_C, cv2.THRESH_BINARY, 31, 2)
    return cv2.threshold(image, 200, 255, cv2.THRESH_BINARY + cv2.THRESH_OTSU)[1]

img = cv2.imread('my_image.jpg')
pytesseract.pytesseract.tesseract_cmd = r'C:\mypath\tesseract.exe'

gray = get_grayscale(img)
thresh = thresholding(gray)
erode = remove_noise(thresh)

custom_config = r'-c tessedit_char_whitelist=0123456789 --psm 6'
print(pytesseract.image_to_string(erode, config=custom_config))

cv2.imwrite("test.jpg", erode)

#these return nothing
print(pytesseract.image_to_boxes(Image.open('test.jpg')))
print(pytesseract.image_to_data(Image.open('test.jpg')))

- revod

Answer 1

Please try the following code:

from pytesseract import Output
import pytesseract
import cv2
 
image = cv2.imread("my_image.jpg")

#swap color channel ordering from BGR (OpenCV’s default) to RGB (compatible with Tesseract and pytesseract).
# By default OpenCV stores images in BGR format and since pytesseract assumes RGB format,
# we need to convert from BGR to RGB format/mode:
rgb = cv2.cvtColor(image, cv2.COLOR_BGR2RGB)
 
pytesseract.pytesseract.tesseract_cmd = r'C:\mypath\tesseract.exe'
custom_config = r'-c tessedit_char_whitelist=0123456789 --psm 6'
results = pytesseract.image_to_data(rgb, output_type=Output.DICT,lang='eng',config=custom_config)
boxresults = pytesseract.image_to_boxes(rgb,output_type=Output.DICT,lang='eng',config=custom_config)
print(results)
print(boxresults)

for i in range(0, len(results["text"])):
    # extract the bounding box coordinates of the text region from the current result
    tmp_tl_x = results["left"][i]
    tmp_tl_y = results["top"][i]
    tmp_br_x = tmp_tl_x + results["width"][i]
    tmp_br_y = tmp_tl_y + results["height"][i] 
    tmp_level = results["level"][i]
    conf = results["conf"][i]
    text = results["text"][i]
    
    if(tmp_level == 5):
        cv2.putText(image, text, (tmp_tl_x, tmp_tl_y - 10), cv2.FONT_HERSHEY_SIMPLEX,0.5, (0, 0, 255), 1)
        cv2.rectangle(image, (tmp_tl_x, tmp_tl_y), (tmp_br_x, tmp_br_y), (0, 0, 255), 1)
        
for j in range(0,len(boxresults["left"])):
    left = boxresults["left"][j]
    bottom = boxresults["bottom"][j]
    right = boxresults["right"][j]
    top = boxresults["top"][j] 
    cv2.rectangle(image, (left, top), (right, bottom), (255, 0, 0), 1)
       
    
cv2.imshow("image",image)
cv2.waitKey(0)

- livezingy

Answer 2

不使用image_to_boxes，另一种方法是使用cv2.findContours查找轮廓，使用cv2.boundingRect获得边界矩形坐标，并使用cv2.rectangle绘制边界框。

使用此示例输入图像

绘制方框

OCR结果

1234567890

代码

import cv2
import pytesseract
import numpy as np

pytesseract.pytesseract.tesseract_cmd = r"C:\Program Files\Tesseract-OCR\tesseract.exe"

# Load image, grayscale, Otsu's threshold
image = cv2.imread('1.png')
gray = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
thresh = cv2.threshold(gray, 0, 255, cv2.THRESH_BINARY_INV + cv2.THRESH_OTSU)[1]

# Draw bounding boxes
cnts = cv2.findContours(thresh, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)
cnts = cnts[0] if len(cnts) == 2 else cnts[1]
for c in cnts:
    x,y,w,h = cv2.boundingRect(c)
    cv2.rectangle(image, (x, y), (x + w, y + h), (36,255,12), 2)

# OCR
data = pytesseract.image_to_string(255 - thresh, lang='eng',config='--psm 6')
print(data)

cv2.imshow('thresh', thresh)
cv2.imshow('image', image)
cv2.waitKey()

- nathancy

---

有趣的是，10年后仍然会不断出现相同的问题。无论如何，您正在帮助人们做得非常好！ - karlphillip

我没有看到使用这种方法打印出预测字符及其对应框的方式。 - revod

我不明白，难道边界框不是你想要的吗？ - nathancy