Citas bibligráficas
Cortez, E., (2021). Modelos de visión por computadora para la determinación de características superficiales y categorías de calidad del Pallar (Phaseolus lunatus L.) Iqueño seco [Tesis, Universidad Privada Antenor Orrego]. https://hdl.handle.net/20.500.12759/7848
Cortez, E., Modelos de visión por computadora para la determinación de características superficiales y categorías de calidad del Pallar (Phaseolus lunatus L.) Iqueño seco [Tesis]. PE: Universidad Privada Antenor Orrego; 2021. https://hdl.handle.net/20.500.12759/7848
@misc{renati/379811,
title = "Modelos de visión por computadora para la determinación de características superficiales y categorías de calidad del Pallar (Phaseolus lunatus L.) Iqueño seco",
author = "Cortez Agreda, Ever Omar",
publisher = "Universidad Privada Antenor Orrego",
year = "2021"
}
In this research, a Computer Vision System (SVC) has been developed at the laboratory level that captures images with a color digital camera connected to a portable personal computer that, using specialized software, multivariate statistical algorithms and those available In MATLAB, it performs the processing, analysis and classification, in a non-invasive, objective and automatic way of the grains of the dry Pallar (Phaseolus lunatus L.) Iqueño. The classification in 2, 3 and 5 categories of quality of the pallar, the SVC performs it using simple discriminant models: linear (LDA) and quadratic (QDA). The linear (LDA) and quadratic (QDA) discriminant classification models, applied to 3169 images of dry Pallar Iqueño grains, grouped into 2, 3 and 5 quality categories, for all the 6 cases studied, showed performances with a General Efficiency greater than 80% for the linear and quadratic discriminant models, showing good performance. The best general performance was obtained by the LDA model, classified in 3 quality categories, with a General Efficiency of 85.6%, which also showed a Class Efficiency of 95.6%, 97.5% and 96.7% in the classification for each individual category: OPTIMAL, TOLERANT and DISCARD. Likewise, all the classification models studied proved to be robust by maintaining, without significant variation, the general efficiency values, which were above 75% in a balance of samples, between 10% and 80% for validation. cross of 2, 3 and 5 quality categories. The data of the descriptors of surface defects of the grain, obtained by computer vision, allowed to elaborate a letter on the identification and xvi characterization of Pallar Iqueño in three classes, as well as a primer with shades of whiteness, useful for those interested. The research results show that the discriminant linear and quadratic classification models in computer vision are effective in emulating expert human vision, in the objective evaluation and classification of the quality of samples from Pallar Iqueño seco
Este ítem está sujeto a una licencia Creative Commons Licencia Creative Commons