Citas bibligráficas
Flores, R., (2019). Análisis por imágenes hiperespectrales de mezclas de suelo con derivados de petróleo, en la región de 900 a 1700 nm [Tesis, Universidad Nacional de Ingeniería]. http://hdl.handle.net/20.500.14076/21210
Flores, R., Análisis por imágenes hiperespectrales de mezclas de suelo con derivados de petróleo, en la región de 900 a 1700 nm [Tesis]. PE: Universidad Nacional de Ingeniería; 2019. http://hdl.handle.net/20.500.14076/21210
@mastersthesis{renati/711109,
title = "Análisis por imágenes hiperespectrales de mezclas de suelo con derivados de petróleo, en la región de 900 a 1700 nm",
author = "Flores Camargo, Ricardo",
publisher = "Universidad Nacional de Ingeniería",
year = "2019"
}
In this investigation, soils contaminated with petroleum derivatives were studied using a laboratory hyperspectral imaging system with the objective of analyzing the spectral response in the region of 900 to 1700 nm. For this purpose, the main objectives are to prepare reference patterns of petroleum-derived soils, apply different methods of preprocessing and processing for hyperspectral images and differentiate contaminated soil from uncontaminated soil. The work is divided into three parts, (1) calibration of the laboratory hyperspectral imaging system, (2) preparation of reference standards samples and image acquisition, (3) preprocessing and processing of hyperspectral ima-ges (HSI). For the calibration of the laboratory hyperspectral imaging system the calibration of the reflectance of the hyperspectral chamber with a reference standard (Teflon) and the calibration of the laboratory HSI system where the focus-sample distance, speed of the sample should be controlled platform, frames per second (fps) and spatial resolution. To make the standard samples, 70 % of the soil (silty or clayey) is mixed with 30 %of crude oil, then reduce the concentration in a range of 3 % until reaching a minimum concentration of 3 % of crude of oil, this processing is repeated for standard samples with diesel. For the standard sample of sandy soil, the maximum concentration of the contaminant is 20 % (crude or diesel) and is reduced in a range of 2 % until reaching a minimum concentration of 2 % (crude or diesel). 60 reference standards were obtained which were compared to uncontaminated soils (silty, clayey and sandy). For the preprocessing of the HSI, normalization was carried out, the noise was eli-minated by means of smoothing, the multiplicative value and the baseline of the values obtained from the HSI were corrected, using the methods: maximum normalization, Savitzky-Golay, standard normal variable (SNV) and de-trend. For the HSI processing, principal component analysis (PCA) and thresholding were used to eliminate unwanted pixels or classes (background, sample holder, shadows or points of maximum reflectance) from the sample. After processing, the images were grouped by the type of contaminant to form mosaics and classify it by PCA and analyze the dispersion of each reference standard against its average value ( sigma) and standard deviation (sn); managing to conclude that the average value of a contaminated sample is greater than the average value of an uncontaminated sample, noting that there is less dispersion in the contaminated samples compared to the one that is not contaminated. In order to be able to relate the concentration to the second derivative of the SNV-D reflectance spectra, a linear model is made between the wavelength regions of 1087 to 1274 nm, obtaining the best results for silty soil with petroleum oil. With all the results obtained in the investigation, it was concluded that a system of laboratory hyperspectral images in the range of 900 to 1700 nm can be used as a rapid and non-invasive system to classify between contaminated and uncontaminated soil as well as to find the degree of contamination in soils contaminated by petroleum derivatives.
Este ítem está sujeto a una licencia Creative Commons Licencia Creative Commons