Interactive indexing into image databases

摘要: The general problem of object recognition is difficult and often requires a large amount computing resources, even for locating an within single image. How, then, can it be possible to build tool indexing into database of, say, thousands images, which works effectively in `interactive time' on affordable hardware? One important optimization take advantage interaction with the user find out what types variation are expected database, rely discriminate between similar-looking objects. Another create appropriate data structures off-line speed on-line searches. We building tool, called FINDIT, image from number images scenes may contain object. outlines that he wants specifies constraints transformations occur. program acts as filter quickly reduce candidates small enough perused by user. FINDIT chooses search algorithm depending selection user.© (1993) COPYRIGHT SPIE--The International Society Optical Engineering. Downloading abstract permitted personal use only.