Paper:
Indexing Visual Features Using a Hybrid Neural Network
Jesse S. Jin*, Henry C. Wang* and Tom Gedeon**
*Department of Computer Science, University of Sydney NSW 2006, Australia
**School of Information Technology, Murdoch University WA, Australia
Indexing and retrieving visual information is an important issue in multimedia development. It involves handling high dimensional vectors. Current tree-based high dimensional index structures, such as R-tree, SS+-tree, TV-tree, etc, have the similar low bound to the one-dimensional comparison-based search methods. It is far from being practical in the multimedia area. We propose a fast indexing method using surrogate coding. It possesses many good properties such as preserving similarity ranking and being fast in retrieval. It also preserves a clustered space and is easy to maintain.
This article is published under a Creative Commons Attribution-NoDerivatives 4.0 Internationa License.