Efficient query by image content for very large image databases

摘要: The QBIC (query by image content) project in the IBM Almaden Research Center San Jose, CA, is conducting a theoretical, experimental, and prototyping study of problem querying large still databases efficiently based on content. Since difficult, aim to discover general principles, but at same time identify target application(s) for which concrete pilot systems will be prototyped. A number algorithms have been developed that allow user search color, texture, shape. can focused either objects (areas previously outlined user) or whole images. argument all part particular image, user-selected patterns shape selected from 'pickers', any weighted combination these patterns. An example associated result given. results initial experimentation are encouraging. >