MCDM'08 - paper no. 10


 

Back to MCDM'08 contents
 

Applying reference sets in content-based interactive image retrieval

Andrzej M.J. Skulimowski, Paweł Rotter

Abstract:

The search for graphical objects in multimedia databases is a challenging field of current research and an emerging area of application of multicriteria decision theory. It is characterized by co-existence of qualitative, quantitative, and graphical criteria and gradual approximation of preference structures during the search. Here, we propose a new approach to image search based on preference information in form of reference images provided by the user interacting with an intelligent search system. Such information can be used in image retrieval systems with relevance feedback for complex graphical objects such as leisure facilities, human faces etc. Reference sets can be combined with any other method of content-based image retrieval (CBIR), resulting in a refined search. Computational experiments have proven that the proposed approach is computationally efficient. Finally, we provide a real-life illustration of the methods proposed: an image-based hotel selection procedure.

Keywords:

Multimedia databases, content-based image retrieval, relevance feedback, multicriteria decision support, preference elicitation, reference sets

Reference index:

Andrzej M.J. Skulimowski, Paweł Rotter, (2009), Applying reference sets in content-based interactive image retrieval, Multiple Criteria Decision Making (4), pp. 185-202

Full text:

download