Magazine article Information Today

Image Access and Retrieval: Defining Content

Magazine article Information Today

Image Access and Retrieval: Defining Content

Article excerpt

A library applications and data processing clinic focuses on digital images

The ability to search images by content is luring computer scientists and others to pursue research together with information scientists, librarians, and archivists. All are enticed by potential applications in today's libraries, which are actively involved in handling large digital image collections in nearly every art and science.

Indexing those images--labor-intensive, thus costly--often does not convey the full meaning. A possible answer to a fast-growing problem is content retrieval by such attributes as shape, color, texture, and composition.

To explore digital image technology and the many facets of its impact on collections and their institutions and to see what problems retrieval by content raises, an interdisciplinary group of specialists from the library, museum, computer science, electrical engineering, and business fields recently met at Digital Image Access and Retrieval, the 33rd Annual Clinic on Library Applications of Data Processing.

The conference was held at the University of Illinois at Urbana-Champaign (UIUC) in March and was sponsored by the Graduate School of Library and Information Science and the Beckman Institute of Advanced Science, UIUC. Presentations also covered the licensing of images, preservation of photos and text, video image access and retrieval, and more.

Keynote speaker Howard Besser, visiting associate professor at the University of Michigan's School of Information, compared technical capabilities in 1986 with today's and noted the extent to which we understood the problem of image storage and retrieval a decade ago in his detailed overview of image databases. In his look at a few of those capabilities a decade ago, he noted that scanners were expensive, rare, used mainly for instrumentation, and not well understood; hard disks were novel for PCs; and 30 MB of storage was considered a lot.

Some ideas about image processing were right, Besser said, but for reasons different from those we give now. Imaging tools were available, but people did not know what to do with them, and problems with associated text persisted. Along the way, a number of image-browsing tools became available with good linking of images, but, as Besser noted, they were "lousy" for searching and linking. The critical breakthrough, he said, was the World Wide Web.

We were partly right about--but didn't really understand--external operator tools, interoperability, integration, and the idea that tools should be provided for different operations.

Besser divided the work that we still need to do into two categories: what can be resolved outside the library and information science (LIS) field (bandwidth, storage, client-servers, scanners) and what can be resolved within LIS. The latter includes topics such as attributes that relate to metadata (data about data); technical image information (e.g., standards for file formats, compression, network communication protocols); data standards for data structure, content, and values; procedural guidelines (e.g., shared cataloging); intellectual property (e.g., what is "fair use," rights management); longevity (refreshing by moving to new media); and migration (e.g., making the data usable 50 to 100 years from now).

We need interfaces and navigation and query by image content, along with access through such means as virtual reality gloves and headsets, said Besser. We need more tools to move from having "a collection of library images" to having "a digital library." And we need well-written software.

When we put an image on the Web, we should worry about intellectual property, Besser said. Many institutions incorrectly assume they own images.

Besser warned that vendors will put high-volume images on the Internet and let the rest lie fallow--a phenomenon he termed "the cherry picking syndrome." Any time there is scarcity of bandwidth, the vendor will select the large-volume materials. …

Search by... Author
Show... All Results Primary Sources Peer-reviewed

Oops!

An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.