Visibility in underwater robotics: Benchmarking and single image dehazing

Dealing with underwater visibility is one of the most important challenges in autonomous underwater robotics. The light transmission in the water medium degrades images making the interpretation of the scene difficult and consequently compromising the whole intervention. This thesis contributes by a...

Descripción completa

Detalles Bibliográficos
Autor: Pérez Soler, Javier
Tipo de recurso: tesis doctoral
Estado:Versión publicada
Fecha de publicación:2017
País:España
Institución:CBUC, CESCA
Repositorio:TDR. Tesis Doctorales en Red
OAI Identifier:oai:www.tdx.cat:10803/432778
Acceso en línea:http://hdl.handle.net/10803/432778
http://dx.doi.org/10.6035/14028.2017.178642
http://mediaserver.csuc.cat/tdx/documents/48/65/15/48651557423080424863754938277457379135/
Access Level:acceso abierto
Palabra clave:Robótica submarina
Benchmarking
Dehazing
Aprendizaje automático
Simulación de robots
Visibilidad submarina
Tecnologies de la Informació i les Comunicacions (TIC)
68
Descripción
Sumario:Dealing with underwater visibility is one of the most important challenges in autonomous underwater robotics. The light transmission in the water medium degrades images making the interpretation of the scene difficult and consequently compromising the whole intervention. This thesis contributes by analysing the impact of the underwater image degradation in commonly used vision algorithms through benchmarking. An online framework for underwater research that makes possible to analyse results under different conditions is presented. Finally, motivated by the results of experimentation with the developed framework, a deep learning solution is proposed capable of dehazing a degraded image in real time restoring the original colors of the image.