Image Textifier

• Row1 : Low resolution, Row2 : Medium resolution
• Left : source, Middle : Derivative, Right : Binarized

Create an image made with texts out of an image.
This small toy is to study python.

Dependency:

OpenCV (Probably any version will fit)

Example:

from ImageTextifier import ImageTextifier
import cv2

def simple_example():
  textifier = ITEX.ImageTextifier()
  result_text, result_image = textifier.textify(cv2.imread("img_path.jpg"))

• For more options, refer to full_function_example() in example.py

Generating dataset

1. Prepare alphabets, numbers, and characters
2. Make those as a set of binary images

Process (ImageTextifier.py)

1. Preprocess : Source(wid * hi) to gray, blur, canny
2. Begin comparing
   • Divide ROI mask into blocks(Higher blocks means higher resolution)
   • Compare each block with pre-maid text images, and choose one with the highest score
3. Return
   • texts
   • an image with texts on it

TODO

• Make the result look better

Things I've learned

• Python environment setup using Conda
• Basic OpenCV & NumPy
• Basic multiprocess module & Timer