As a step towards developing a high-performance image retrieval system, we propose a clustering method that efficiently classifies image objects having an unknown probability distribution, without requiring the determination of complicated parameters, through the use of a self-organizing map (SOM) and a method of image processing. To ensure that this clustering method is fast and highly reliable, we defined a hierarchical SOM and used it to construct the clustering method. Experiments using artificial image data confirmed the basic performance and adaptability of the SOM for clustering images. We also confirmed experimentally and theoretically that our clustering method using the hierarchical SOM is faster than one using a non-hierarchical SOM for the objects used in these experiments.
[1]
Teuvo Kohonen,et al.
Self-Organizing Maps
,
2010
.
[2]
Heekuck Oh,et al.
Neural Networks for Pattern Recognition
,
1993,
Adv. Comput..
[3]
Amarnath Gupta,et al.
Visual information retrieval
,
1997,
CACM.
[4]
Michael R. Anderberg,et al.
Cluster Analysis for Applications
,
1973
.
[5]
Shih-Fu Chang,et al.
VisualSEEk: a fully automated content-based image query system
,
1997,
MULTIMEDIA '96.
[6]
Teuvo Kohonen,et al.
The self-organizing map
,
1990,
Neurocomputing.