|
A picture is worth a thousand words -- or in this case, a thousand keywords -- especially when searching the Web for a display of men's shirts, bathroom wallpaper or Japanese prints. When keywords alone cannot locate that special "something" to fit a specific taste, users can turn to IBM's patented Query By Image Content or QBIC, a new search engine that sorts through database images according to colors, textures, shapes, sizes and their positions.
How QBIC Works
When words are not precise enough to describe a pattern, color or texture, users can turn to QBIC for exacting and efficient visual retrieval of images across many fields, from stock photo archives and textile manufacturing to business graphics, art and retail catalogs and certain medical applications. Used in combination with keywords to first narrow a field of interest, QBIC's image content database can then refine a search by accessing thousands of images that match a user's specific visual request. The resulting list is ranked from best to worst match.
For example, a user might choose keywords to access a particular on-line clothing catalog and then utilize QBIC to visually narrow his choices to a particular pattern or color of interest. By automatically indexing images in a pre-processing stage, QBIC then provides image examples or color wheel to complete the search. QBIC can also use keyframes to search for similar videos for the purpose of content management.
Future Applications
Image content is one of the most rapidly evolving facets of the World Wide Web, and has an enormous potential to deliver still and video images of almost anything imaginable. When combined with keywords, QBIC offers new and easier ways to search visually rich databases, locating images on the Web as accurately as keywords can locate text.
QBIC could serve as a new source of ideas and information for people in such visually-based industries as animation, enabling searches of large image/video databases for new ideas; in the textile industry, by providing designers with a visual databank of patterns; or to facilitate the study of art in libraries and museums. It could also enhance the use of large databases of graphical foils like PowerPoint and Freelance and help trademark searches by finding trademarks similar in shape and color.
An outgrowth of robotic machine vision inspection systems for microelectronics, QBIC is available as a search engine and is also integrated in DB2 Universal and IBM Digital Library.
For Researchers
Links to related topics
QBIC Home Page
In Art and Museums Demos
Query by Image Content (QBIC)
Internet World '95 Integration Exhibits
|
