Water Classification Based on Underwater Monocular Image

  • Everson Fagundes de Toledo FURG
  • Edwilson Silva Vaz FURG
  • Paulo L. J. Drews FURG

Resumo


The oceans occupy a considerable part of our planet and are underexplored compared to their own significance. The bottom of the sea was reached, but there are several difficulties involved in trying to extract information from that location. In the spectrum of robotics and computer vision, underwater images are particularly challenges due to the vast range of existing aquatic environments, presenting diverse characteristics. A robot working underwater needs to understand the environment to which it is exposed. Thus, the proposed work aims to help the capabilities of its vision system addressing the problem of classifying underwater images according to their water type. The proposed approach takes into account the color channels and uses a classification tool widely accepted in the scientific world as a basis. The method is developed through the recognition of patterns observed in several underwater images of varying depths and types. We achieve good results when compared to state-of-the-art methods opening several opportunities for underwater image processing.
Palavras-chave: Measurement, Image recognition, Image color analysis, Planets, Oceans, Machine vision, Tools
Publicado
11/10/2021
TOLEDO, Everson Fagundes de; VAZ, Edwilson Silva; DREWS, Paulo L. J.. Water Classification Based on Underwater Monocular Image. In: SIMPÓSIO BRASILEIRO DE ROBÓTICA E SIMPÓSIO LATINO AMERICANO DE ROBÓTICA (SBR/LARS), 13. , 2021, Online. Anais [...]. Porto Alegre: Sociedade Brasileira de Computação, 2021 . p. 282-287.