A University of Maryland computer scientist has begun collaborating with researchers at a leading document management corporation to advance high-quality mobile scans of business, legal or personal documents.
David Doermann, a senior research scientist in the university's Institute for Advanced Computer Studies (UMIACS), and Raja Bala, a principal research scientist in the Xerox Research Center Webster facility near Rochester, N.Y., are combining their expertise in document imaging to design next-generation scanning tools that are easy to use and computationally efficient.
The research is funded in part by a $30,000 gift, renewable for three years, from the Xerox Foundation.
While today's optical character recognition technology is very good for cleanly scanned images, attempting to replicate such success with mobile devices is still a significant challenge, Doermann says. Many handheld users have problems visually framing the document, with their captured images often suffering from poor lighting, blurring and/or optical scaling.
Doermann and Bala plan to apply recent advances in computational photography—where a camera can take multiple images simultaneously, each having a different field of depth or light-setting that are immediately merged together—to overcome these problems in the document domain.
Traditional flatbed scanners often don't need the intricacy of computational photography, says Bala, because documents are normally properly placed on the scanner and are well illuminated.
But with many smartphones having somewhat limited camera quality and computing power—and their users having differing skill levels—the challenge for the UMD/Xerox team is to scale the sophistication of computational photography to a small device. Applications might include a landlord scanning a lease for a tenant, a graduate student e-delivering a research paper on time or a 9-year-old sending his grandparents a hand-drawn holiday greeting.
"Ultimately, we want a handheld product that can offer legal-quality reproductions, yet is also very reliable and affordable, and provides a positive user experience," says Doermann, who leads the Laboratory for Language and Media Processing in UMIACS.
For more information, visit http://umdrightnow.umd.edu/news/umd-xerox-seek-improve-mobile-document-scanning
April 18, 2013