Hyperspectral image segmentation based on spatial model

This paper proposes the use of data-driven segmentation methods for spatial segmentation of hyperspectral images. In this context, hyperspectral images are represented with 3 PCA bands and they are divided into segments via Structured Forest Model which is prepared by supervised learning techniques....

Full description

Saved in:
Bibliographic Details
Published in2016 24th Signal Processing and Communication Application Conference (SIU) pp. 1249 - 1252
Main Authors Demirel, Berkan, Ozdil, Omer, Esin, Yunus Emre
Format Conference Proceeding
LanguageEnglish
Published IEEE 01.05.2016
Subjects
Conferences
Histograms
hyperspectral image processing
Hyperspectral imaging
Image segmentation
spatial segmentation
tree detection
Vegetation
Online AccessGet full text
DOI10.1109/SIU.2016.7495973

Cover

More Information
Summary:This paper proposes the use of data-driven segmentation methods for spatial segmentation of hyperspectral images. In this context, hyperspectral images are represented with 3 PCA bands and they are divided into segments via Structured Forest Model which is prepared by supervised learning techniques. These segments are used for tree detection in the hyperspectral datacubes. Experimental results show that spatial information can be used effectively in the solution of various problems in the field of hyperspectral imaging.
DOI:10.1109/SIU.2016.7495973