Hi everyone!
What happens
I’m working on a hacking challenge and I’m stuck. The challenge is to solve in one minute 45 of 50 (90%) captchas that are distorted.
What do you understand or find about that problem
I must manipulate the image to reduce the distortion generated by the captcha.
You make any workaround? What did you do?
I am using tesseract and pytesseract for OCR and imagemagick for image processing, all this with python.
(Optional) Why fails your workaround?
My python script works partially, I can download the images and make the post request to resend the processed information.
However my problem is in the text recognition, with Imagemagick I have been trying to correct the images, but I get only between 8 and 14% accuracy and I can’t increase it.
I have tried quite a few image corrections with imagemagick (Morphology of Shapes -- IM v6 Examples), but I can’t get it and do research about how to bypass captchas.
Evidences
Original captcha
A resize of image
Pixel smoothing
And this is the final result
I need help with
Any idea how I can improve the image processing or if there is some other method with which I can reverse the captcha.
Thanks in advance.