Skip to Main content Skip to Navigation

Prétraitements des images sous-marines basés sur la polarisation et le filtrage fréquentiel : application offshore

Abstract : Study of the underwater environment requires significant technological advances, particularly, in the development of autonomous underwater vehicles, and their perception sensors. This thesis is dealing with the development of a real time solution for underwater image quality improvement in order to promote the use of autonomous underwater vehicles. Developing a fast image processing algorithms are required due the limitations of these kinds of vehicles in terms of energy, computing capacity and storage. In this context, a fast and effective method of underwater image quality improvement has been proposed. On the one hand, this method uses a polarimetric imaging optical system to reduce the diffusion effects on the image acquisition.On the other hand, it is based on an optimized version of the dark channel prior (DCP) method that has received a great deal for image dehazing.In this thesis, we are also interested in the detection and the identification of markers used for the automatic docking of an underwater vehicle with a submerged station. The success of this task requires a good contrast in the area where the marker is located. To solve this problem, an object-oriented dehazing method is proposed to optimize the contrast of markers. The proposed strategy exploits the texture features derived by Gabor multichannel filtering for image segmentation. Once different objects of the image are separated, an optimized Dark Channel Prior dehazing method is applied to optimize the contrast of each individual object. The system has been tested on a large image dataset and the obtained results show that the object-oriented dehazing improves the markers identification in underwater environment.
Complete list of metadatas

Cited literature [154 references]  Display  Hide  Download
Contributor : Abes Star :  Contact
Submitted on : Friday, March 27, 2020 - 10:35:20 AM
Last modification on : Wednesday, October 14, 2020 - 9:16:06 AM
Long-term archiving on: : Sunday, June 28, 2020 - 1:31:36 PM


Version validated by the jury (STAR)


  • HAL Id : tel-02519911, version 2



Khadidja Ould Amer. Prétraitements des images sous-marines basés sur la polarisation et le filtrage fréquentiel : application offshore. Traitement du signal et de l'image [eess.SP]. Université de Bretagne occidentale - Brest, 2019. Français. ⟨NNT : 2019BRES0049⟩. ⟨tel-02519911v2⟩



Record views


Files downloads