Extending image retrieval with group-oriented interface

ImageGrouper is a new user interface for digital image search and organization. It is group-oriented in that image search, annotation and organizing are done by grouping images on the workspace. This enables faster and flexible text annotation on a large number of images as well as effective image organization. In addition, this interface improves and extends content-based image retrieval (CBIR) as well. First, finding good combinations of query examples is essential for successful retrieval. The query-by-groups concept allows the user to try different combinations of query examples quickly and easily. Second, because of semantic gap between the user's concept and extracted visual features, it may be difficult to represent the user's request by only one class of query examples. We propose a new feature-weighting scheme where the user's request can be expressed in multiple positive classes.