Feature reduction using minimum noise fraction and principal component analysis transforms for improving the classification of hyperspectral image

Murinto, Murinto and Pujiastuti, Nur Rochmah Dyah (2017) Feature reduction using minimum noise fraction and principal component analysis transforms for improving the classification of hyperspectral image. [Artikel Dosen]

Text HASILCEK_Murinto_Feature reduction using minimum noise fraction and principal component analysis transforms for improving the classification of hyperspectral image.pdf Download (1MB)

Abstract

Dimensionality reduction is an important milestone in the preliminary process of high-dimensional data analysis.
Most of the research on hyperspectral image fields deal with data extraction techniques. Each feature extraction
technique is unique and has its advantages and disadvantages. However, using certain techniques may result in
significant data loss. To avoid such problems, this research employs a combination of reduction techniques. In
this paper, dimensionality reduction was conducted using principal component analysis (PCA) and minimum noise fraction (MNF). A combined principle component analysis and minimum noise fraction (PCA-MNF) method is proposed. Image classification using a minimum distance (MC) method was performed subsequent to the dimensionality reduction technique. We found that our proposed method increases the accuracy of image classification to 80.77% outperforming both PCA and MNF, which yield 40.37% and 77.21% accuracy, respectively.

Item Type:	Artikel Dosen
Keyword:	Classification, Feature Reduction, Hyperspectral image, MNF, PCA.
Subjects:	Q Science > Q Science (General)
Divisi / Prodi:	Faculty of Industrial Technology (Fakultas Teknologi Industri) > S1-Informatics Engineering (S1-Teknik Informatika)
Depositing User:	murinto murinto
Date Deposited:	18 Apr 2023 09:44
Last Modified:	23 Sep 2023 05:52
URI:	http://eprints.uad.ac.id/id/eprint/42969

Actions (login required)

View Item