Institutional Repository [SANDBOX]
Technical University of Crete
EN  |  EL

Search

Browse

My Space

Real time spectral classification and mapping for insitu detection of plant pathologies

Takas Panagiotis

Full record


URI: http://purl.tuc.gr/dl/dias/920E3E79-577F-4965-8DB1-C202175F9924
Year 2023
Type of Item Diploma Work
License
Details
Bibliographic Citation Panagiotis Takas, "Real time spectral classification and mapping for insitu detection of plant pathologies", Diploma Work, School of Electrical and Computer Engineering, Technical University of Crete, Chania, Greece, 2023 https://doi.org/10.26233/heallink.tuc.95934
Appears in Collections

Summary

Spectroscopy is the study of interaction between matter and light to obtain structural and chemical information. Spectral imaging uses a plethora of narrow-band images across the electromagnetic spectrum to collect and analyze this kind of spectroscopic and imaging information at the same time. Spectroscopic information can be used to create chemical maps of the imaged area through capturing the spectrum of each individual pixel of the image and using it as data set for classification algorithms. For this purpose a hyperspectral camera has been acquired to collect data and detect abnormalities and pathologies in plants and then study if they are detectable from an early stage. The acquired data were classified using clustering algorithms which were developed using Qt Firmware for the GUI and C++ for the main source. In an effort to achieve real time mapping and optimal spectral classification, the usage of GPU’s computational capability was mandatory . The produced maps consist of pseudocolors where each pseudocolor corresponds to a cluster of pixels with spectral similarities. These similarities along with broadly used vegetation indices were studied in plants which have been submitted under stressful conditions (e.g. increased salinity) using our developed algorithms. The findings from the experiment highlight the potential for an early detection of plant’s destruction.

Available Files

Services

Statistics