Deep Learning Based Classification Techniques for Hyperspectral Images in Real Time

  1. Suárez Garea, Jorge Alberto
unter der Leitung von:
  1. Francisco Argüello Pedreira Doktorvater
  2. Dora Blanco Heras Doktormutter

Universität der Verteidigung: Universidade de Santiago de Compostela

Fecha de defensa: 08 von Juli von 2021

Gericht:
  1. Jorge Azorín López Präsident/in
  2. José Ramón Ríos Viqueira Sekretär
  3. Begüm Demir Vocal
Fachbereiche:
  1. Departamento de Electrónica e Computación

Art: Dissertation

Zusammenfassung

Remote sensing can be defined as the acquisition of information from a given scene without coming into physical contact with it, through the use of sensors, mainly located on aerial platforms, which capture information in different ranges of the electromagnetic spectrum. The objective of this thesis is the development of efficient schemes, based on the use of deep learning neural networks, for the classification of remotely sensed multi and hyperspectral land cover images. Efficient schemes are those that are capable of obtaining good results in terms of classification accuracy and that can be computed in a reasonable amount of time depending on the task performed. Regarding computational platforms, multicore architectures and Graphics Processing Units (GPUs) will be considered.