Content-based image retrieval using similarity

With the fast increase in digital image data and its broad applications, an efficient image retrieval system is demanded. This paper proposes a content-based image retrieval system using similarity that capitalizes on the complementary advantages of both current methods, which arc Retrieval by Image Example (RIE) and Retrieval by Semantic Content (RSC). A list of exemplar images is generated from a content-based image synthesis system, and then is used as examples in a GNU Image Finding Tool (GIFT) for image matching. Queries on battlefield images will be tested as experiments and the results are discussed for different two image matching algorithms. The analysis on the results show that image retrieval through semantic queries can be accomplished using the approach described in this paper.