Bibliographic citations
Apaza, H., (2021). Análisis de imágenes hiperespectrales para la detección y cuantificación de trozos de plástico mezclados con granos y microplásticos mezclados con harina de almendra [Tesis, Universidad Nacional de Ingeniería]. http://hdl.handle.net/20.500.14076/26993
Apaza, H., Análisis de imágenes hiperespectrales para la detección y cuantificación de trozos de plástico mezclados con granos y microplásticos mezclados con harina de almendra [Tesis]. PE: Universidad Nacional de Ingeniería; 2021. http://hdl.handle.net/20.500.14076/26993
@mastersthesis{renati/1751765,
title = "Análisis de imágenes hiperespectrales para la detección y cuantificación de trozos de plástico mezclados con granos y microplásticos mezclados con harina de almendra",
author = "Apaza Mamani, Hebbert Alexander",
publisher = "Universidad Nacional de Ingeniería",
year = "2021"
}
Hyperspectral images of plastics mixed with almond, lentil, pea and rice grains as well as microplastics mixed with almond flour obtained in the laboratory using a hyperspectral camera RESONON model Pika NIR are presented. The identification and quantification of abundances of plastics within a group of grains present in a hyperspectral image was carried out using two ways or procedures. In the first procedure, the dimensionality of the workspace was reduced using the Minimum Noise Fraction (MNF) method, then going on to identify the endmembers using the Simplex Growing Algorithm (SGA) and once the elements pre¬sent in the images had been identified when compared with reflectance spectra obtained from the USGS spectral library (United States Geological Survey), the abundances were found th- rough a subpixel analysis using the linear mixture model (MML). In the second procedure to find the abundance of plastic present in a hyperspectral image, the dimensionality was redu- ced by applying the principal component analysis (PCA) method, then working on regions of interest (ROI) of the image that registered the least noise obtained by PCA. These regions of interest made it possible to obtain the reflectance spectra of the elements present in the image to find the transformation matrix for the analysis of the complete image. Prediction models of the percentages of grains and plastics present in the analyzed image were then applied using the PLS-DA model. For this part of the process, routines were carried out in R. Identification maps of microplastics and almond flour were also elaborated. Microplastics abundance identification maps and almond flour were also prepared. Likewise, the presence of microplastics mixed with almond flour present in a hyperspectral image was identified. For this only the the first procedure of those that were followed in the identification of plastics in images of mixtures of plastics with grains was used because it was not possible to find ROI’s in this image. The results show a very good identification of plastics present in grains as well as microplastics present in almond flour with an error of around 2 %.
This item is licensed under a Creative Commons License