In this paper, we propose a feature-weighted mountain clustering method. The proposed method can work well when there are noisy feature variables and could be useful for obtaining initial estimat of cluster centers for other clustering algorithms. Results from color image segmentation illustrate the proposed method actually produces better segmentation than previous methods.
關聯:
INTERNATIONAL JOURNAL OF COMPUTATIONAL INTELLIGENCE SYSTEMS Volume: 4 Issue: 5 Pages: 1002-1011
