chengchang

Thanks for your comment.

The PDF format will ignore some structures of paper and `draw` contents on a fixed-layout flat document. For a char, e-readers cannot know it belongs a paragraph, a figure caption or a formula. Plus, publishers render original paper with the specific style. It is a problem. Some research concentrate to improve PDF format, extract infos for scholarly papers, like articles on DocEng - a compute science conference: http://dl.acm.org/event.cfm?id=RE135. Then I read the spec of PDF format...

Anyway, that's hurt.

- The output is a reflowed PDF. It keeps 'PDF quality'. - Any PDF reader/device could display new version paper technically. - Kindle has supported.

The majority of scientific and technical papers are delivered by the PDF file for portable presentation. However, it difficult to read the PDF paper on small screen devices, such as tablets and eBook readers.

So I wrote PDFlower to reflow PDF papers for small-screen reading.

Compared with plain-text extraction based tools, PDFlower reconstructs the object level - columns, formulas, tables and figures - layout using some smart heuristics, and then renders them for new page size.