I need a script that will work on Linux, using an open source OCR like Tesseract, to decode a captcha, an example of which can be found here:
[url removed, login to view]
This will probably require the rotation of the image to align the characters horizontally and the separation of letters before they are individually OCRed.
25% accuracy would be OK, 50%+ would be great.
I have a framework in java and C # for decoding captchas. I've been working on this for over 10 years. I have decoded more than 100 captchas including the recaptcha.