Combining location and feature information for multimedia retrieval

Abstract

In this paper, we propose a cross-media retrieval method for heterogeneous multimedia data by which the query examples and the returned results can be of different modalities, e.g., to query images by an example of audio clip. Taking multimedia location and content information into consideration, an affinity propagation based clustering approach is proposed to analyse and fuse the information carried by the co-existing multimedia objects so as to learn the semantic correlations among the heterogeneous multimedia data and perform cross-media retrieval. We also propose active learning methods of Relevance Feedback to make the search model more accurate.