Citas bibligráficas
Felipe, W., (2020). Análisis y procesamiento de imágenes hiperespectrales NIR para detección de contaminantes plásticos [Tesis, Universidad Nacional de Ingeniería]. http://hdl.handle.net/20.500.14076/21493
Felipe, W., Análisis y procesamiento de imágenes hiperespectrales NIR para detección de contaminantes plásticos [Tesis]. PE: Universidad Nacional de Ingeniería; 2020. http://hdl.handle.net/20.500.14076/21493
@mastersthesis{renati/711217,
title = "Análisis y procesamiento de imágenes hiperespectrales NIR para detección de contaminantes plásticos",
author = "Felipe Gaspar, Walter Sullivan",
publisher = "Universidad Nacional de Ingeniería",
year = "2020"
}
It is an accepted reality that thousands of tons of plastics are discarded into the environment every year and a huge part of them are consumed by birds, fish and other animals, getting alterations in their body and finally death. As a first step in the remediation of damaged ecosystems, it is important to develop a fast and reliable method for plastics identification. Plastics are organic materials that in most cases are produced from raw materials such as cellulose, carbon, natural gas, salt, and of course oil. Consequently, plastics structures are composed of C – H bonds that have significant absorption properties in the shortwave infrared region (SWIR) between 1000 and 2500 nm, suitable to use a hyperspectral camera that operates in a portion of this region (NIR) between 900 – 1700 nm, which was used to record these properties. Plastic samples were collected and images were taken in a laboratory. In order to simulate a polluted environment, plastics were photographed alone (as pure samples) and mixed with soil, seawater, wood and vegetation. The obtained hyperspectral images were analyzed to find their main spectral characteristics and correctly identify the plastics within all these mixtures. The processing models used were Principal component analysis (PCA), Spectral angle Mapper (SAM), Support Vector Machine (SVM) and K – Means Clustering. To validate the obtained reflectance spectra, the USGS spectral library was used, checking the presence of the absorption bands of the C – H bonds in the 1100 – 1225 nm, 1300 – 1420 nm and 1650 – 1800 nm regions. SVM proved to be the most useful method in the identification of plastics types and other materials, which they were mixed. The analysis with SAM showed good results to differentiate types of plastics under precise under precise setting of their reference parameters, although in mixtures with water and oil, the spectra of the submerged plastics were not relevant. PCA and K – Means clustering were not always able to distinguish the types of plastics in the mixtures since they had a high dependence on visual inspection for the choice of appropriate correlations and the number of clusters chosen respectively. In the environment, in addition to plastics there are also other materials such as vegetation and wood. The spectra of these materials have been compared with the spectra of plastics, finding that their characteristics are significantly different. Therefore, hyperspectral images are definitely an applicable and especially reliable method to detect plastics in the environment.
Este ítem está sujeto a una licencia Creative Commons Licencia Creative Commons