COLLABORATIVE CLUSTERING ALGORITHM WITH REDUCING DIMENTIONALITY FOR HYPERSPECTRA SATELLINE IMAGES

  • Đặng Trọng Hợp
  • Mai Đình Sinh
Keywords: Hyperspectral image; fuzzy clustering; collaborative clustering; feature reduction.

Abstract

Hyperspectral satelline images (HSI) have received popularityand shown
their usefulness in various earth observation applications in recent years.
Segmentation is the basic problems in HSI processing but it also is one of the
most difficult taks because HSI have hundreds of channels and high-performance
computing is crucial. In this paper, we proposed solution for HSI segmentation by
using collaborative clustering with reducing image dimentionality by random
projection based on Johnson Lindenstrauss lemma (C2JL algorithm) which also
preserves the relative distance between data points. The experiments which
were done on 2 HSI data sets with 5 validity indexes shows that proposed
methods give the better performance.

điểm /   đánh giá
Published
2022-03-31
Section
RESEARCH AND DEVELOPMENT