如何提取图片中的数字？

Question

请将这个图片下载后，保存为 sample.png

from PIL import Image
import pytesseract
text=pytesseract.image_to_string(Image.open('sample.png'),lang='eng')  
print(text)

尝试另外一段代码

import cv2
import pytesseract
img = cv2.imread("/tmp/sample.png")
gry = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
thr = cv2.adaptiveThreshold(gry, 255, cv2.ADAPTIVE_THRESH_MEAN_C,
                            cv2.THRESH_BINARY_INV, 23, 100)
bnt = cv2.bitwise_not(thr)
txt = pytesseract.image_to_string(bnt, config="--psm 6")
res = ''.join(i for i in txt if i.isalnum())
print(res)

结果 ee ，差太远了。

没有任何结果，请问哪位可以提取这里面的字符？

Answer 1

运用了形态学操作中的腐蚀来去除经过处理之后网格所形成的白点，但是还是有一些误差，第三个字符'5' tesseract 会识别为'S'，正在想办法改进算法。

import cv2
import pytesseract
pytesseract.pytesseract.tesseract_cmd = r'C:\Program Files\Tesseract-OCR\tesseract.exe'
image = cv2.imread('sample.png')
gray_image = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
inverted_image = cv2.bitwise_not(gray_image)
binary_image = cv2.adaptiveThreshold(inverted_image, 255, 
                                     cv2.ADAPTIVE_THRESH_GAUSSIAN_C, 
                                     cv2.THRESH_BINARY, 11, 2)
denoised_image = cv2.medianBlur(binary_image, 3)
kernel = cv2.getStructuringElement(cv2.MORPH_RECT, (4, 4))
eroded_image = cv2.erode(denoised_image, kernel)
mask = (denoised_image == 255) & (eroded_image == 0)
denoised_image[mask] = 0
cv2.imwrite('preprocessed_image.png', denoised_image)
text = pytesseract.image_to_string(denoised_image, config='--psm 6')
print("result:", text.strip())

处理后的图片：
腐蚀前：
腐蚀后：

输出：

result: CRSP

---

注：
这段代码中运用了光学字符识别（OCR）引擎Tesseract，代码中的pytesseract.pytesseract.tesseract_cmd参数根据Tesseract安装位置更改。