System detection and automatic classification of pollen grain applies technical digital imaging process

  • Jorge Arroyo Hernández Universidad Nacional
  • Carlos M. Travieso González Instituto Universitario para el Desarrollo Tecnológico y la Innovación en Comunicaciones
  • Jaime Ticay Rivas Instituto Universitario para el Desarrollo Tecnológico y la Innovación en Comunicaciones
  • Federico Mora Mora Universidad Nacional
  • Oscar Salas Huertas Universidad Nacional
  • Melvin Ramírez Bogantes Universidad Nacional
  • Luis Sánchez Chavez Universidad Nacional
Keywords: Pollen, Digital Image Processing, Palynology, Principal Components Analysis (PCA), Neural Networks

Abstract

This paper show the current state of a computer system that will allow the recognition and taxonomic classification of pollen grains of some of the most important tropical honey plants in Costa Rica using techniques of pre and post processing of digital images. The digital system uses filters on the images allowing it to detect and highlights its features and contour. Afterwards it is parametrized and finally a system of neuronal interconnections is used for the automatic recognition of pollen grains. The idea behind the implementation of a computer program is to move from a qualitative to a quantitative paradigm, using different mathematical tools and artificial intelligence in a way that can speed the process of recognition and classification of pollen grains. Using the PCA and the Sum at the outputs (CA) of 30 networks were able to obtain a success rate of 91,67 ± 3,13 which is highly promisisng for the purpose of the automatic classification system.

References

-
Published
2013-01-01
How to Cite
Arroyo Hernández, J., Travieso González, C., Ticay Rivas, J., Mora Mora, F., Salas Huertas, O., Ramírez Bogantes, M., & Chavez, L. (2013). System detection and automatic classification of pollen grain applies technical digital imaging process. Uniciencia, 27(1), 59-73. Retrieved from https://www.revistas.una.ac.cr/index.php/uniciencia/article/view/4943
Section
Original scientific papers (evaluated by academic peers)

Most read articles by the same author(s)