Supervisor : Florence Tupin
Context : SAR images are acquired by satellite or aerial sensors in all-weather and all-time conditions. Being a coherent imaging system, these images are highly noisy and difficult to segment. The aim of this project is to segment water bodies on SAR images to prepare the next SWOT mission (NASA / CNES mission).
Objectives : The aim of this project is to study the performances of the Chan and Vese approach with a curve evolving using a functional minization taking into account both region homogeneity and curve geometry. In our case only two areas have to be considered one for the background and one for water surface. This approach has been first developed for additive gaussian noise and it will thus be necessary to adapt it to the SAR distributions (Gamma).
Useful links : paper and code implementation with demos are available on IPOL :
http://www.ipol.im/pub/art/2012/g-cv/
Programming langage : C (to be able to re-use the IPOL code).
Other papers and codes in link with the project:
http://www.mathworks.com/matlabcentral/fileexchange/29447-multiphase-level-set-image-segmentation
http://www.ipol.im/pub/art/2012/g-cv/
Programming langage : C (to be able to re-use the IPOL code).
Other papers and codes in link with the project:
http://www.mathworks.com/matlabcentral/fileexchange/29447-multiphase-level-set-image-segmentation