This is a Preprint and has not been peer reviewed. This is version 3 of this Preprint.
Downloads
Authors
Abstract
This paper presents a comparative analysis of tra- ditional machine learning methods and Convolutional Neural Networks (CNNs) for hyperspectral image classification. Utilizing the Indian Pines dataset, we explore the efficacy of Principal Component Analysis (PCA) combined with a Support Vector Machine (SVM) classifier against a deep learning approach involving CNNs. Our methodology includes dimensionality re- duction via PCA, followed by SVM classification, and the design of a tailored CNN model for hyperspectral data. Performance metrics like accuracy, supported with confusion matrices and classification maps, are employed to evaluate and compare the models. Results indicate that CNNs, with their ability to capture spatial and spectral information, outperform traditional methods in classification accuracy and robustness.
DOI
https://doi.org/10.31223/X5811M
Subjects
Computer Engineering, Computer Sciences, Other Computer Sciences, Physical Sciences and Mathematics
Keywords
This paper presents a comparative analysis of tra- ditional machine learning methods and Convolutional Neural Networks (CNNs) for hyperspectral image classification. Utilizing the Indian Pines dataset, we explore the efficacy of Principal Component Analysis (PCA) combined with a Support Vector Machine (SVM) classifier against a deep learning approach involving CNNs. Our methodology includes dimensi, followed by SVM classification, and the design of a tailored CNN model for hyperspectral data. Performance metrics like accuracy, supported with confusion matrices and classification maps, are employed to evaluate and compare the models. Results indicate that CNNs, with their ability to capture spatial and spectral information, outperform traditional methods in classification accuracy and robustness., Hyperspectral Imaging, Convolutional Neural Network (CNN), Principal Component Analysis (PCA), Support Vector Machine (SVM) classifier
Dates
Published: 2024-08-13 14:02
Last Updated: 2024-08-14 16:42
Older Versions
License
CC BY Attribution 4.0 International
Additional Metadata
Data Availability (Reason not available):
https://purr.purdue.edu/publications/1947/1
There are no comments or no comments have been made public for this article.