Hiperspektrālu attēlu dimensiju skaita redukcijas metode
Loading...
Date
Authors
Advisor
Journal Title
Journal ISSN
Volume Title
Publisher
Latvijas Universitāte
Language
lav
Abstract
Bakalaura darbs veltīts daudzdimensionālu attēlu apstrādes un pikseļu klasifikācijas metožu izveidei. Bakalaura darbā sniegti klasifikāciju uzdevumu risinājumi, veidojot matemātisku modeli. Modeļu izveidei tiek izmantoti Eiklīda, Čebiševa, Mahalanobisa un galvenā virziena distances mēri. Dimensiju skaita redukcija ir veikta ar galveno komponentu metodi. Par kvalitātes rādītāju izvēlēta iegūtā klasifikatora kopējā precizitāte.
The bachelor work is related to hyperspectral image processing and implementation of the pixel classification method.. For making mathematical model classification task solution are given in the bachelor work. The model is based on Euclidean, Chebyshev, Mahalanobis and city block distance. Dimension reduction is done with principal component analysis. Overall accuracy measure is used as a quality indicator of the classification method.
The bachelor work is related to hyperspectral image processing and implementation of the pixel classification method.. For making mathematical model classification task solution are given in the bachelor work. The model is based on Euclidean, Chebyshev, Mahalanobis and city block distance. Dimension reduction is done with principal component analysis. Overall accuracy measure is used as a quality indicator of the classification method.