Abstract
This paper proposes a three-dimensional Gabor feature extraction for pixel-based hyperspectral imagery classification using a memetic algorithm. The proposed algorithm named MGFE combines 3-D Gabor wavelet feature generation and feature selection together to capture the signal variances of hyperspectral imagery, thereby extracting the discriminative 3-D Gabor features for accurate classification. MGFE is characterized with a novel fitness evaluation function based on independent feature relevance and a pruning local search for eliminating redundant features. The experimental results on two real-world hyperspectral imagery datasets show that MGFE succeeds in obtaining significantly improved classification accuracy with parsimonious feature selection.
Original language | English |
---|---|
Title of host publication | Advances in Swarm Intelligence - Third International Conference, ICSI 2012, Proceedings |
Pages | 479-488 |
Number of pages | 10 |
Edition | PART 1 |
DOIs | |
Publication status | Published - 2012 |
Event | 3rd International Conference on Swarm Intelligence, ICSI 2012 - Shenzhen, China Duration: 17 Jun 2012 → 20 Jun 2012 |
Publication series
Name | Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) |
---|---|
Number | PART 1 |
Volume | 7331 LNCS |
ISSN (Print) | 0302-9743 |
ISSN (Electronic) | 1611-3349 |
Conference
Conference | 3rd International Conference on Swarm Intelligence, ICSI 2012 |
---|---|
Country/Territory | China |
City | Shenzhen |
Period | 17/06/12 → 20/06/12 |
Bibliographical note
Copyright:Copyright 2013 Elsevier B.V., All rights reserved.
Keywords
- Gabor Feature Extraction
- Hyperspectral Imagery Classification
- Memetic Algorithm
ASJC Scopus subject areas
- Theoretical Computer Science
- Computer Science(all)