A prototype implementation of metadata generation for image retrieval
H. Sesaki, Yasushi Kiyoki · 2004
We present a prototype implementation of a metadata generation method by exemplary textual indexing to digital images for image retrieval. The proposed method uses image processing for computing structural similarity between the prepared sample images with sample indexes and query images based on visual features, and selects the sample indexes of the most similar image as metadata to the requested query images. The use of image processing allows cost-efficient implementation for content-based metadata indexing to visual objects.