srijanmukherjee/image-processing-ocr-assignment

Performs image smoothening, converting to binary image, generating skeleton of the image for OCR purpose

Image processing for OCR

Basic image processing (smoothening, binarizing, skeletonizing) before using image for OCR

Instructions

# start virtual environment
python -m venv env
source ./env/bin/activate

# install dependencies
pip install -r requirements.txt

# run the code
python main.py <image>

# Example
python main.py --blur gaussian -k 7 -v dataset/English/E3.png 

# Run over all examples in the dataset
python experiment.py

NOTE:
Minimum python version requirement: 3.10

Usage

usage: main.py [-h] [-v] [-b {average,gaussian,median,bilateral}] [-k KERNEL] [-o OUTPUT] image

positional arguments:
  image                 path to image

options:
  -h, --help            show this help message and exit
  -v, --verbose         Enable verbose output
  -b {average,gaussian,median,bilateral}, --blur {average,gaussian,median,bilateral}
                        Choose blurring techinque
  -k KERNEL, --kernel KERNEL
                        set kernel size (must be odd), the kernel will be set as (size x size)
  -o OUTPUT, --output OUTPUT
                        output directory

What does it do?

• Apply image filtering technique to smoothen the image
  • Average blurring
  • Gaussian blurring
  • Median blurring
  • Bilateral blurring
• Convert into binary image
• Skeletonize the image
• Resize into 64x64 image

This example demonstrates some basic image processing techinques to be applied before we use the image for processing techniques like OCR.

References

• https://pyimagesearch.com/2021/04/28/opencv-smoothing-and-blurring
• https://scikit-image.org/docs/stable/auto_examples/edges/plot_skeleton.html