Skip to content

This is a tesseract based OCR to read Embossed text on metals. It can also be use as a general OCR.

Notifications You must be signed in to change notification settings

DevashishPrasad/Embossed-Text-Reader

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

28 Commits
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Embossed-Text-Reader

This is a tesseract based OCR to read Embossed text on metals. It can also be use as a general OCR.

This is not generalized solution. The satements and parameters of some funtions will changed according to the texture, color, lighting effect and visibility of text of image.

Tesseract is an open-sourced OCR which is capable of reading text from papers, pdfs and other clean formats. Tesseract fails when tried to perform OCR on noisy and dirty images (for eg. Embossed or Engraved text). The code uses opencv image filtering techniques to filter the images as clean as possible and then feeds it to Tesseract.

Dependencies

Python
OpenCV
Tesseract
Numpy
imutils

There are three files -

  1. picture_from_webcam.py clicks a single picture from webcam and performs OCR on it
  2. read_from_disk.py loads a picture from an image file
  3. video_from_webcam.py feeds the realtime video from the webcam (doesn't perform as expected)

Example

Edged image (canny) with dialation

Picture not availble in your browser

Finding and deleting unessesary contours

Picture not availble in your browser

Inverted threshold image

Picture not availble in your browser

Results

Picture not availble in your browser

Sample results -

Picture not availble in your browser

Picture not availble in your browser

Picture not availble in your browser

Picture not availble in your browser

The code may fail sometimes but tuning parameters of canny and appropriate dialation will improve results

Data.zip contains all data I have used during implementation of this project

About

This is a tesseract based OCR to read Embossed text on metals. It can also be use as a general OCR.

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages