eduzhai > Applied Sciences > Engineering >

A Novel Spatial-Spectral Framework for the Classification of Hyperspectral Satellite Imagery

  • king
  • (0) Download
  • 20210507
  • Save

... pages left unread,continue reading

Document pages: 13 pages

Abstract: Hyper-spectral satellite imagery is now widely being used for accuratedisaster prediction and terrain feature classification. However, in suchclassification tasks, most of the present approaches use only the spectralinformation contained in the images. Therefore, in this paper, we present anovel framework that takes into account both the spectral and spatialinformation contained in the data for land cover classification. For thispurpose, we use the Gaussian Maximum Likelihood (GML) and Convolutional NeuralNetwork methods for the pixel-wise spectral classification and then, usingsegmentation maps generated by the Watershed algorithm, we incorporate thespatial contextual information into our model with a modified majority votetechnique. The experimental analyses on two benchmark datasets demonstrate thatour proposed methodology performs better than the earlier approaches byachieving an accuracy of 99.52 and 98.31 on the Pavia University and theIndian Pines datasets respectively. Additionally, our GML based approach, anon-deep learning algorithm, shows comparable performance to thestate-of-the-art deep learning techniques, which indicates the importance ofthe proposed approach for performing a computationally efficient classificationof hyper-spectral imagery.

Please select stars to rate!

         

0 comments Sign in to leave a comment.

    Data loading, please wait...
×