display | more...
Producers of digital text often try to protect their "intellectual property" using encryption and a proprietary reader software (like Microsoft Reader: http://www.microsoft.com/reader). The reader program usually lacks the ability to copy & paste, or only allows copying & pasting small paragraphs. Of course such protections are very annoying and IMHO, readers should have the right to circumvent them. The problem will get worse the more popular ebooks become (which is largely dependent on display quality).

What is required is a general solution that turns screen text (which can usually only be captured into the form of bitmaps) into normal text. While existing OCR programs like OmniPage aren't totally helpless when confronted with text captured onscreen, they are also not really good at it. There will be a real market for a program that allows users to capture & recognize any kind of text from their screens, spanning multiple pages.

Since the letters displayed on screen are usually constant in their appearance (with the exception of ligatures), it should be pretty trivial to write an algorithm that, when supplied with the input bitmaps for all letters, checks their re-appearance.

There is a program called Kleptomania (http://www.structurise.com) that currently has something like a monopoly on screen OCR, but it really doesn't make capturing large quantities of screen text easy and relies completely on the rendering of fonts installed on the user's system. This is good for small chunks of text in standard fonts, but not for something like a complete book with an encrypted font hard-coded into the program.

Instead, the user should be required to puzzle together the alphabet with a screen loupe and a selection tool. Whenever a letter is added to the program's alphabet, it is marked red on screen indicating that it is recognized. Once you have added all letters, you can start the OCR process and, voila, out comes the text file. The alphabets could be shared among the users.

To allow capturing multiple pages, the program would require an option to send a keystroke or mouseclick to a window after each screen capture, possibly with a delay to accomodate slow computers.

Once such a solution would be in place, it would be impossible to stop readers from copying&pasting the text wherever they like.

Blind people who are, under the DMCA allowed to bypass DRM in order to get access to the content, already have this type of solution. The scanning and OCR program Kurzweil 1000 already comes with a feature ccalled the virtual printer. This feature installs fake printer drivers on the system. Whenever anything is printed from a program that allows printing, the Kurzweil 1000 program is launched. An image of the printed page is then put through OCR. As the image is much cleaner than a regular scan of a page, recognition is nearly perfect without any proofing. As most ebook publishers and online files allow printing at least once, this works really well.

Log in or register to write something here or to contact authors.