Associative Retrieval Method for Image Database

摘要

The paper presents the principles, means of practical application, and examples of the associative retrieval method that the authors have been studying as a practical retrieval method for an image database in an image synthesizing system. As described in the paper, associative retrieval enables retrieval by relying on the scope and direction of association while, based on keyword descriptions of the meanings of the image, using a key image as the starting point of association. To demonstrate the validity of this method, the authors created an experimental system that achieves image retrieval using the image itself as a means of man-machine interaction. With this system, they obtained a processing speed fast enough for a practical application. It is necessary for the user to be able to adjust the weight of the keyword from experience. To simplify weighting, one needs to provide additional learning functions and establish a new image description method to achieve more advanced retrieval.