I'm finding this document quite useful: http://www.succeed-project.eu/sites/default/files/deliverables/Succeed_60055...
See description of ALTO pasted below, which is a followup to https://lists.wikimedia.org/pipermail/wikisource-l/2014-September/002081.htm... . We should find a way to convert the transcribed books' HTML to ALTO format. :)
Some libraries are apparently using http://www.primaresearch.org/tools/Aletheia which seems an augmented (but unfree?!) version of ScanTailor with some different purpose.
Nemo
Principles ALTO stores layout information and OCR recognized text of pages of any kind of printed documents like books, journals and newspapers. ALTO can detail technical metadata for describing the layout and content of physical resources (text, illustrations, graphics). ALTO describes a content page with different views: The Description section helps to describe some general settings and information of the ALTO file (measurement units, file name, etc.), and the production process itself (processing steps, software used, dates and actors, etc.) The Layout section contains what‟s on the page. A page is divided into several regions (print space; left, right, top and bottom margins). For each region, all objects are listed which have been detected inside: text blocks, illustrations, graphical elements, composed blocks. Each object previously identified is defined by generic attributes: width, height, text content (for the String element). Besides, the reading order of all the elements can be managed. Each ALTO file may also contain a style section where different styles (for paragraphs and fonts) are listed. Use cases ALTO is one of the most common formats used by libraries for converting text from images. It‟s used both to deliver digitized contents and to preserve these contents. In a delivery perspective, the ability of ALTO to store the text content coordinates in a page allows the overlay of image and text (multilayer PDF) and highlight search words in a query.