Search code examples
pythonimage-processingtesseractpython-tesseract

Pytesseract skips "1" but not "10" in the same file

I am working with pytesseract and openCV to try to recognize a table of numbers. I have been working heavyly on the image to resize, resample and treshold its colors to make it easier for pytesseract to read. Below is the image I managed to generate.

My problem is that everytime a single "1" comes up in a row, pytesseract isn't able to recognize it...

This is the image I am trying to read (once I have applied all the mentionned processings) : The table of numbers

This is the relevant part of the code :

from PIL import Image
import pytesseract

img = cv2.imread('test.jpg', 0)
data = pytesseract.image_to_string(img)

And this is the output:

10

499

I also tried with --psm 10 and --psm 13 but the outputs are just gibberish like the following :

=
:x
Solution

  • Apply inverse binary threshold:

    enter image description here

    Set page-segmentation mode to 6

    1
10
499

    Code:

    import cv2
from pytesseract import image_to_string

image = cv2.imread('uHLww.png')
gray = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
thresh = cv2.threshold(gray, 0, 255, cv2.THRESH_BINARY_INV)[1]
text = image_to_string(thresh, config="--psm 6")
print(text)

    Second solution:

    You don't even have to apply thresholding, setting psm to 6 will give you the result.

    import cv2
from pytesseract import image_to_string

print(image_to_string(cv2.imread('uHLww.png'), config="--psm 6"))