Abstract
The Web, TV and print publishing are the top three advertisement media. With the technology breakthroughs in digital printing, scanning and data processing, print on demand emerges into a multi-billion dollar business that is due to revolutionize the traditional publishing industry. With the rapid growth of print-on-demand, a whole new advertising media is born, that is, print on demand books. We discuss some unique characteristic of this new media that enables user-targeted contextual advertising, which is more effective than traditional print media. In this paper, we first proposed a contextual-matching based advertising model for print on demand books, and then concentrate on using the page frame segmentation algorithm to solve the problem of where to insert the selected contextual ads in print on demand books. Our page segmentation algorithm shows significant improvement over existing methods in its robustness against noisy background.