Basically, your job is to make this code work:
QString [url removed, login to view](QImage);
The function is given an image and you return the text. You can also set a percent certainty variable but that is optional. QString & QImage are an amazingly easy-to-use library by Trolltech company which you'll find you can learn virtually instantly.
The background of the image may be a color gradient background possibly with mild textures, so you'll have to pre-process the image if necessary. The deadline is very important so you'll be expected to send updates at least once a day on most days of the week.
All images will be cropped ahead of time so each image is one line of text only.
I can offer at least some assistance in how to work with QStrings and QImages. They are very easy to use.
The function must work at least 99.9% of the time on a plain background and 99.5% of the time on a mildly textured background such as included in the sample image. A high accuracy rate on the text font marked "challenging" in the attached sample is not required though I'll guarantee a $25 bonus if you can get a 99% accuracy rate. I attached to this project. If the OCR object detects a character that seems like a character but is not readable, then some sort of method of communicating that back should be established. In the sample I included I set the window size to a fairly small setting, so thats the toughest I imagine it will get. However, if the font is larger it should still work of course!