| HYPERSPECTRAL IMAGE CLASSIFICATION BY EXPLOITING CONVOLUTIONAL NEURAL NETWORKS |
| Paper ID : 1348-SMPR |
| Authors: |
|
Benyamin Hosseiny1, Heidar Rastiveis *2, Somaye Daneshtalab3 1Department of Geospatial engineering, Faculty of Engineering, University of Tehran, Tehran, Iran 2Department of Photogrammetry and Remote Sensing,
School of Surveying and Geo-Spatial Engineering
College of Engineering, University of Tehran, 3Department of Photogrammetry and Remote Sensing, School of Surveying and Geo-Spatial Engineering College of Engineering, University of Tehran, |
| Abstract: |
| High spectral dimensionality of hyperspectral images makes them as a useful data resource for earth observation in many remote sensing applications. Recently, analysing these images using advanced machine learning techniques has recently become one of the hottest research topics in remote sensing communities. In this case, deep architectures such as convolutional neural network (CNN) can help to extract deep and robust features in hyperspectral images. The main goal of this paper is to propose a new architecture of convolutional neural network to extract deep features from hyperspectral datasets to achieve a better classification result. In the proposed method, after pre-processing step, data is fed to a convolutional neural network, in order to extract deep features. Extracted features are then imported in a multi-layer perceptron (MLP) network as our selected classifier. Experiments are performed on two famous hyperspectral datasets and compared with results of classic MLP network. Obtained results expressed more than 10% improvements compared to the classic MLP classification technique. |
| Keywords: |
| Hyperspectral, Remote sensing, Artificial Neural Networks, Classification, Convolutional Neural Networks, Deep learning |
| Status : Conditional Accept (Poster) |
