Python OCR (Extracting URLs from images)

This tutorial aims to teach you how to use existing resources like tesseract, cv2, etc. to create a simple yet powerful OCR (optical character recognition system). We have added a new feature to the traditional OCR, i.e it can identify URLs in an image and also opens them in the web browser.

Tutorial written by Saksham Kapoor. Get in touch here.

What We Will Accomplish

Moreover, Your links will automatically open in your default browser

Excited? Let's gooo!

Prerequisites

Make sure you have python installed on your computer.
Knowledge of basic python syntax.

Step 1 - Install pytesseract OCR for windows from here.

During the installation, use the default destination path.
Copy the destination path as we'll need it soon.
After the installation is complete, search for environment variables in the windows search bar.
Click on Edit the environment variables search result.
Click on Environment Variables, then under the heading System Variables click Path and then click Edit.
Click New, then paste the path that we just copied.

Check if the installation was successful

Open windows powershell, type
```
    tesseract
```

If you see something like this, it was successfully installed

Usage:
C:\Program Files\Tesseract-OCR\tesseract.exe --help | --help-extra | --version
C:\Program Files\Tesseract-OCR\tesseract.exe --list-langs
C:\Program Files\Tesseract-OCR\tesseract.exe imagename outputbase [options...] [configfile...]

OCR options:
-l LANG[+LANG]        Specify language(s) used for OCR.
NOTE: These options must occur before any configfile.

Single options:
--help                Show this help message.
--help-extra          Show extra help for advanced users.
--version             Show version information.
--list-langs          List available languages for tesseract engine.

However, if not, please read the instructions carefully and try again.

Step 2 (Optional) - Tesseract Demo (Create Pdf from Image)

Download sample.jpg from this repo (You can use any image).
Put it in a new folder.
Open terminal and cd to the new folder
```
> cd Desktop/NewFolderName/
```
In the terminal type write the following command -
```
> tesseract sample.jpg output pdf
```
Now check the folder, you will have a new pdf named output.pdf generated from the sample.jpg

Step 3 - Install the following python packages-

pillow (helps us to deal with images in python)
pytesseract (creates a link between python and the tesseract OCR engine that we just installed)
opencv-python (OpenCV - Open Source Computer Vision Library, is an open source computer vision and machine learning software library. We'll use it to read images.)

    pip install pillow
    pip install pytesseract
    pip install opencv-python

Step 4 - Create New Python file

I have named it main.py

Step 5 - Import Required Python Modules

    from PIL import Image
    import pytesseract
    import cv2
    import re
    import os
    import webbrowser

Usage :

re : Used to define a regular expression (regex) which will help us to search a url in the image text.
webbrowser : Used to open pages on the browser.
os : We will use this is generate a unique id and also manage files in the directory.
cv2 : We will use this to read and convert the image to gray scale.

Step 6 - Basic Config

  # Set your tesseract.exe path here
  pytesseract.pytesseract.tesseract_cmd = r'C:\Program Files\Tesseract-OCR\tesseract.exe'

  # Specify Path/Name of image you want to read
  img_name = "sample.jpg"

Step 7 - Reading Text and URL from the Image

  # Reading the image
  img = cv2.imread(img_name)

  # Converting the image to grayscale
  gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)

  # Create temporary image which will be fed to the tesseract engine
  # file name has to be unique, therefore we have used OS
  temp_img = "{}.jpg".format(os.getpid())

  # Writing the image to the temporary image file
  cv2.imwrite(temp_img, gray)

  # Getting text from image
  text = pytesseract.image_to_string(Image.open(temp_img))

  # Finding the url using
  urls = re.findall('(?:(?:https?|ftp):\/\/)?[\w/\-?=%.]+\.[\w/\-?=%.]+', text)

  # Deleting the temporary file
  os.remove(temp_img)

Step 8 (Optional) - Storing Text and URLs in '.txt' files

  # Export text from image to a text file (Optional)
  text_file = open(f"{img_name}_text.txt", "w")
  text_file.write(text)
  text_file.close()

  # Export urls from image to a text file (Optional)
  urls_file = open(f"{img_name}_urls.txt", "w")
  for url in urls:
      urls_file.write(url + "\n")

Step 9 - Opening the urls in your default web browser

  # Open Urls in the default browser
  # If Multiple urls, it will open them in seperate tabs
  flag = 0
  for url in urls:
      if flag == 0:
          webbrowser.open(url)
      else:
          webbrowser.open_new_tab(url)
      flag = flag + 1

Step 10 - Test it out

A sample photo has been provided in this repo. Download/Clone the repo and run python script. It should just as expected.

Run Python Scripts on windows -

Open Script in any text editor, change img_name to the image of your choice.
Open Windows Powershell.
Change directory to the folder containing image and the script.

  cd ./Desktop/FolderName

Run script

  python main.py

Step 11 - Star and Share this tutorial :)

You can reach out to me at sakshamkapoor1729@gmail.com

saksham-kapoor/python-OCR-simplified