Implementing Optical Character Recognition (OCR) using Python and OpenCV

Implementing Optical Character Recognition (OCR) using Python and OpenCV

Learn how to implement Optical Character Recognition (OCR) using Python and OpenCV. This blog post covers the basic steps involved in OCR, including image preprocessing, text detection, and text recognition. By combining the power of OpenCV and pytesseract, you can extract text from images and scanned documents with ease. Explore the numerous applications of OCR in document digitization, data entry automation, and text-to-speech conversion. Start implementing OCR in your own projects using Python and OpenCV today!

Optical Character Recognition (OCR) is a technology that allows computers to recognize and extract text from images or scanned documents. It has numerous applications, such as digitizing printed documents, automating data entry, and enabling text-to-speech conversion. In this blog post, we will explore how to implement OCR using Python and OpenCV.

optical character recognition - OCR using Tesseract, OpenCV and Deep Learning

Getting Started with OCR

Before diving into the implementation details, let's first understand the basic steps involved in OCR:

1. Preprocessing the image: This step involves cleaning and enhancing the image to improve the accuracy of text extraction.
2. Text detection: In this step, we locate the regions of the image that contain text.
3. Text recognition: Once the text regions are identified, we apply OCR algorithms to recognize and extract the text.

Installing the Required Libraries

To get started, we need to install the necessary libraries. OpenCV is a popular computer vision library that provides various image processing functions. We can install it using pip:

pip install opencv-python        

In addition to OpenCV, we also need to install the pytesseract library, which is a Python wrapper for the Tesseract OCR engine:

pip install pytesseract        

Implementing OCR using Python and OpenCV

Now that we have installed the required libraries, let's dive into the implementation of OCR using Python and OpenCV.

Step 1: Preprocessing the Image

The first step in OCR is to preprocess the image. This involves converting the image to grayscale, applying thresholding to create a binary image, and performing noise removal.

optical character recognition - OCR using Tesseract, OpenCV and Deep Learning

import cv2

def preprocess_image(image):
    gray = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
    _, binary = cv2.threshold(gray, 0, 255, cv2.THRESH_BINARY_INV + cv2.THRESH_OTSU)
    denoised = cv2.fastNlMeansDenoising(binary, None, 10, 7, 21)
    return denoised

image = cv2.imread('input_image.jpg')
processed_image = preprocess_image(image)        

Step 2: Text Detection

After preprocessing the image, we can proceed with text detection. In this step, we use the EAST (Efficient and Accurate Scene Text) text detector, which is a deep learning model trained to detect text regions in images.

import cv2
import numpy as np

def detect_text(image):
    net = cv2.dnn.readNet('frozen_east_text_detection.pb')
    blob = cv2.dnn.blobFromImage(image, 1.0, (320, 320), (123.68, 116.78, 103.94), swapRB=True, crop=False)
    net.setInput(blob)
    scores, geometry = net.forward(["feature_fusion/Conv_7/Sigmoid", "feature_fusion/concat_3"])
    return scores, geometry

scores, geometry = detect_text(processed_image)        

Step 3: Text Recognition

Once the text regions are detected, we can proceed with text recognition using the pytesseract library. This library provides a simple interface to the Tesseract OCR engine.

import pytesseract

def recognize_text(image, scores, geometry):
    rows, cols, _ = image.shape
    confidences = []
    boxes = []

    for i in range(scores.shape[2]):
        confidence = scores[0, 0, i, 0]
        if confidence > 0.5:
            x1 = int(geometry[0, 0, i, 1] * cols)
            y1 = int(geometry[0, 0, i, 2] * rows)
            x2 = int(geometry[0, 0, i, 3] * cols)
            y2 = int(geometry[0, 0, i, 4] * rows)
            confidences.append(confidence)
            boxes.append((x1, y1, x2, y2))

    indices = cv2.dnn.NMSBoxes(boxes, confidences, 0.5, 0.4)
    results = []

    for i in indices:
        x1, y1, x2, y2 = boxes[i[0]]
        cropped_image = image[y1:y2, x1:x2]
        text = pytesseract.image_to_string(cropped_image, config='--psm 6')
        results.append((text, (x1, y1, x2, y2)))

    return results

recognized_text = recognize_text(processed_image, scores, geometry)        

Conclusion

In this blog post, we explored how to implement Optical Character Recognition (OCR) using Python and OpenCV. We covered the basic steps involved in OCR, including image preprocessing, text detection, and text recognition. By combining the power of OpenCV and pytesseract, we can extract text from images and scanned documents with ease.

optical character recognition - OCR using Tesseract, OpenCV and Deep Learning

OCR has numerous applications in various industries, such as document digitization, data entry automation, and text-to-speech conversion. It is a powerful technology that can save time and effort by automating manual tasks.

By following the code examples and steps provided in this blog post, you can start implementing OCR in your own projects using Python and OpenCV. Experiment with different image preprocessing techniques and OCR algorithms to achieve the best results for your specific use case.

Happy coding!

==================================================

For more IT Knowledge, visit https://itexamtools.com/

check Our IT blog - https://itexamsusa.blogspot.com/

check Our Medium IT articles - https://itcertifications.medium.com/

Join Our Facebook IT group - https://www.facebook.com/groups/itexamtools

check IT stuff on Pinterest - https://in.pinterest.com/itexamtools/

find Our IT stuff on twitter - https://twitter.com/texam_i