Retrieval of Images using Rich-region Descriptions

• Published in: Journal of visual languages and computing Vol. 11; no. 3; pp. 303 - 321
• Main Authors: CINQUE, L., LECCA, F., LEVIALDI, S., TANIMOTO, S.
• Format: Journal Article
• Language: English
• Published: Elsevier Ltd 01.06.2000
• Subjects: content query, image database, segmentation, regions, matching, similarity, distance, user interface, rich-region description; content query, image database, segmentation, regions, matching, similarity, distance, user interface, rich-region description
• ISSN: 1045-926X
• DOI: 10.1006/jvlc.2000.0161

Retrieval of images from databases using their visual features is a challenging and important problem. While the technical problem shares some aspects of image analysis with image understanding, the goal is not to obtain a correct interpretation of the image but to enhance the recall and precision of retrieval. The dominant visual features of an image depend on subjective interpretations and can vary from user to user. We present a technique to improve recall in region-based retrieval; the method is based upon a family of representations of images called ‘rich-region descriptions’. We show in a simple experiment how this kind of representation can improve the flexibility allowed to users in obtaining desired results. We also discuss issues related to the user interface for segmentation and query systems. In this subject, the paper extends a previous work.