The focus of the paper is on making scanned books easier to read online. The main type of book they talked about are children's books. In these books, the images can be nearly as important as the text. How the text is displayed can also be important. The writers have come up with two solutions to this problem.
The first is called ClearText, and it is when the text in the page is taken out and rendered in the font set of the user's machine. This makes it possible to "resize the text without resizing the background." It also would help towards the translation of the text for readers of other languages. However, problems come up when dealing with the font size. If the font gets too large, then it could obscure the image. There is also the issue of what the font size should be when switching between pages. They determined that there should be a maximum font size allowed based on the page with the most text.

The second is called PopoutText. This leaves the text as is in the picture/page, but it makes a sort of cut-out of it that can be clicked on and popped out of the image with greater magnification. The advantage of this is that the text stays the same, and the imagery with the text, if there is anything, will remain. However, this does not lend itself for translations of the text.

The writers then compared these two methods with the current method of online books. This method shows the image, and the user must magnify the image to see all the text. Then they must scroll around the page to find all the wording, and sometimes this can be a challenge, depending on the size of their screen.

I really enjoyed reading this paper. I can see these sorts of things coming into standard use, especially as more and more material starts appearing on the internet. Personally, I don't like reading things online, and part of the reason for this is in fact the smaller resolution of the screen. The writers mention this as one of their reasons for creating these interfaces. It looks like they have a good base, and I hope to see this come into more standard use in the future.
Link to paper


