ABT

Today I passed 8.901, Astrophysics I, my final breadth requirement. That makes me officially ABD (All But Dissertation). Expect "Silencing and Recombination in Yeast Ribosomal DNA" at your local bookstore around December. A movie deal with Burger King tie-in should follow shortly thereafter.

CAPTCHAs and Tesseract

CAPTCHAs always seem to come up in discussions of OCR projects like Tesseract, so I decided to test Tesseract to see if it was actually the next big thing in spammer technology.

A commenter pointed me to a Berkeley effort to defeat CAPTCHAs that handled 92% of a selection of text-images from 2002 using a tailored OCR algorithm. I used PyTesser and difflib to run through the images and check the results.

Tesseract read the image correctly for 36 out of 191 images (19%), and was close (within one character) for 5 more. Here are a few of the harder images it was able to crack:





Tesseract had trouble most frequently on text that was more skewed, had lots of distracting dots, or white or black lines crisscrossing the words. With a small amount of image preprocessing (removing speckles and narrow lines), it might do much better on this old set. On modern CAPTCHAs, though, it's probably SOL.

PyTesser: OCR for Python

In two years using Python, I never once searched for "Python X" without finding PyX, someone's labor of love making X easy to understand and use. Many examples come to mind. That's what hooked me when I started out: any tool I wanted was at my fingertips in 5 minutes, and just worked.

Optical character recognition was just about the only exception. So, I got excited when Google released Tesseract OCR, a straightforward, relatively accurate OCR package written in C++. Tesseract didn't have Python bindings, but it didn't take much work with PIL and subprocess to make it act like it did.

Behold! PyTesser.

>>> from pytesser import *
>>> image = Image.open('fnord.tif') # Open image object using PIL
>>> print image_to_string(image) # Run tesseract executable on image
fnord
>>> print image_file_to_string('fnord.tif')
fnord