Loading…
Hyperspectral Image Classification using Extended Local Binary Patterns and Wavelet Transform Descriptors
The performance and accuracy of the hyperspectral image classifier depends on the specific features present in the image as well as the method employed to select the samples used for the learning stage. In this paper, we investigate the use of evolved texture descriptors in a spectral-spatial design...
Saved in:
Main Authors: | , , |
---|---|
Format: | Conference Proceeding |
Language: | English |
Subjects: | |
Online Access: | Request full text |
Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
Summary: | The performance and accuracy of the hyperspectral image classifier depends on the specific features present in the image as well as the method employed to select the samples used for the learning stage. In this paper, we investigate the use of evolved texture descriptors in a spectral-spatial design for the classification of hyperspectral images. We integrate the Extended Local Binary Pattern operator in a parallel hyperspectral image classification framework chain that uses wavelet descriptors for the spectral dimension. We investigate the benefits of the rich spatial information on the accuracy obtained based on the classification chain, and we show through experimental validation carried out on two open datasets (Salinas and Pavia University) that the performances are improved. The proposed framework with the corresponding methods performs well when evaluated on the two open datasets, by using a random, respectively controlled strategy for selecting the databases for training and testing. |
---|---|
ISSN: | 2475-7861 |
DOI: | 10.1109/ISETC56213.2022.10009976 |