rnnr \ - A * ) 19990719 021 Indexing Flower

A new approach to indexing a specialized database by utilizing the color and spatial domain knowledge available for the database is described. This approach is illustrated by using it to provide a solution to the problem of indexing images of flowers for searching a flower patents database by color. The flower region is isolated from the background by using an automatic iterative segmentation algorithm with domain knowledge-driven feedback. The color of the flower is defined by the color names present in the flower region and their relative proportions. The database can be queried by example and by color names. The system provides a perceptually correct retrieval with natural language queries by using a natural language color classification derived from the ISCC-NBS color system and the X Window color names. The effectiveness of the strategy on a test database is demonstrated.