Text Detection Using OpenCV and PyTesseract

Machine Learning

Recognizing text from an image is easier these days because of the libraries like OpenCV and pytesseract . In this article, I will show you how you can implement simple text recognizer using python.

I am using python3.6 . Some required libraries are :

1. OpenCV sudo pip install opencv
2. Numpy sudo pip install numpy
3. PyTesseract sudo pip install pytesseract; sudo apt-get install tesseract-ocr

Now let’s get to the coding part.

1. Create a working folder and name it as you like
2. Create a folder img inside working folder
3. Create a folder temp inside working folder
4. Put some test images in img folder
5. Create a file detector.py and write some code

import cv2
import numpy as np
import pytesseract as tes
from PIL import Image
import os

def get_string(path):
    img = cv2.imread(path)
    #gray conversion
    img = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)

    #removing noise
    kernel = np.ones((1, 1), np.uint8)
    img = cv2.dilate(img, kernel, iterations = 1)
    img = cv2.erode(img, kernel, iterations = 1)

    #write noise free image
    cv2.imwrite("./temp/noise_free.png", img)

    #apply threashold
    img = cv2.adaptiveThreshold(img, 255, cv2.ADAPTIVE_THRESH_GAUSSIAN_C, cv2.THRESH_BINARY, 31, 2)

    #write again
    cv2.imwrite("./temp/thres.png", img)

    #run tesseract
    result = tes.image_to_string(Image.open("./temp/thres.png"))

    #remove created files
    os.remove("./temp/thres.png")
    os.remove("./temp/noise_free.png")

    return result

print("starting recognizing...")
#change name.png to actual image file name
print(get_string("./img/name.png"))

That’s it. You are good to go. Now run it using

python3 detector.py

And you will see the detected text from image.

You May Also Like