An Efficient Retrieval of Annotated Images based on WordNet

The amount of multimedia information is rapidly increasing due to digital cameras and mobile telephones equipped with such devices. To interpret semantic of image, many researchers use keywords as textual annotation. However, current state of the art produces too many irrelevant keywords for images by annotator. They make bad retrieval result. In this paper, we measured the similarity between keywords using WordNet and classified to have semantically similar keywords through clustering. Finally, we calculated weight values of classified keywords so we used a weight value for efficient retrieval of annotated images.