Litcius/Paper detail

Hyperspectral Remote Sensing Images Classification Using Fully Convolutional Neural Network

Nyan Linn Tun, Alexander I. Gavrilov, Naing Min Tun, Do Minh Trieu, Htet Aung

202120 citationsDOI

Abstract

In remote sensing technology, hyperspectral images are widely used to research and monitor the earth's surface with high spatial and spectral resolution. Hyperspectral images contain more than three bands compared to traditional RGB Images. Deep learning techniques are very useful and achievable in the field of computer vision systems. This paper proposes a fully convolutional neural network for hyperspectral image classification (HSI). The idea of the proposed model is 2D convolutional layers learned the features maps based on the spectral-spatial information of the HSI data, and the fully connected layers of CNN performed the HSI classification. All experiences are performed over three remote sensing datasets, Indian Pines, Saline Scene, and University of Pavia. The proposed method has achieved high classification accuracies in the field of hyperspectral images classification.

Topics & Concepts

Hyperspectral imagingConvolutional neural networkComputer scienceArtificial intelligenceRemote sensingRGB color modelImage resolutionPattern recognition (psychology)Contextual image classificationField (mathematics)Computer visionDeep learningImage (mathematics)GeographyMathematicsPure mathematicsRemote-Sensing Image ClassificationRemote Sensing and Land UseRemote Sensing in Agriculture