The paper describes a project aiming at content-based retrieval of coats of arms to facilitate information finding. The retrieval system should be usable by experts and by casual users. In order to make use of the specific nature of arms about 20 different features have been developed to allow an expert to fine-tune an image search. For the casual user different features are grouped into feature lists that are called query models. These models correspond to image clusters in the database. The paper discusses the customization of features for the retrieval of coats of arms and their grouping into query models. Experiments and test results show a good performance of query models.
Proceedings of International Symposium on Multimedia Application and Distance Education, Baden-Baden, Germany, 1999