Real time water removal for underwater photogrammetry with depth information
Date
2024-12-03
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Pontificia Universidad Católica del Perú
Abstract
En la actualidad, la restauración de imágenes submarinas sigue siendo un tópico
complejo en el ámbito de la imagenología óptica subacuática. Esto se debe a que los
efectos atmosféricos y de reflectividad producidos por la escena suelen generar una
degradación significativa a la fotografía y limitar el rango capturado debido a la
borrosidad producida en el ambiente. Con la finalidad de solucionar este problema, la
siguiente tesis implementa un algoritmo de retrodispersión que busca eliminar el efecto
de degradación generado por el agua. Este proceso utilizara una imagen extraída de una
cámara RGB y un mapa de profundidad extraído por software de fotogrametría como
puntos de entrada.
Sin embargo, dada la complejidad computacional que involucra ejecutar, el tiempo de
ejecución aumenta exponencialmente, especialmente para imágenes de alta resolución y
matrices generados de mapas de profundidad. Debido a esto, es indispensable diseñar un
algoritmo paralelo e implementado en una GPU para poder acelerar significativamente
el tiempo de procesamiento.
Nowadays, underwater image restoration remains as one complex issue in underwater optical imaging. The reason of this is because the atmospheric and reflective effects produced by the scene that can severely degrade the photograph and limits the captured range because of the blurriness produced in the environment. To solve this topic, this thesis makes an implementation of a backscatter algorithm that tries to remove the degradation effect produced by the water. This process will utilize a RGB photograph and a depth map extracted from a photogrammetry software as inputs. However, given the significant computational complexity involved, the execution time increases considerably, especially for high resolution images and depth map matrices. For that reason, it is necessary to design a parallel algorithm and implement it in a GPU to be able to significantly accelerate the processing time.
Nowadays, underwater image restoration remains as one complex issue in underwater optical imaging. The reason of this is because the atmospheric and reflective effects produced by the scene that can severely degrade the photograph and limits the captured range because of the blurriness produced in the environment. To solve this topic, this thesis makes an implementation of a backscatter algorithm that tries to remove the degradation effect produced by the water. This process will utilize a RGB photograph and a depth map extracted from a photogrammetry software as inputs. However, given the significant computational complexity involved, the execution time increases considerably, especially for high resolution images and depth map matrices. For that reason, it is necessary to design a parallel algorithm and implement it in a GPU to be able to significantly accelerate the processing time.
Description
Keywords
Procesamiento de imágenes digitales, Sistemas de imágenes, Visión por computadoras
Citation
Collections
Endorsement
Review
Supplemented By
Referenced By
Creative Commons license
Except where otherwised noted, this item's license is described as info:eu-repo/semantics/openAccess