Automated Segmentation of SAS Images using the Mean - Standard Deviation Representation

Frederic Maussang (1) , Jocelyn Chanussot (1) , Alain Hétet (2)

1 LIS - Laboratoire des images et des signaux
2 DGA/DET/GESMA - Groupe d'Etudes Sous-Marines de l'Atlantique

A segmentation method of synthetic aperture sonar (SAS) images is presented, in order to highlight some characteristics (number, position, shape, ...) of underwater mines echoes. This segmentation method is based on statistical characteristics of the sonar images, highlighted by the mean – standard deviation plane. It is automated by using an entropy criterion.

Mots clés: segmentation; entropy; Synthetic Aperture Sonar; detection; mean; standard deviation representation

Frederic Maussang, Jocelyn Chanussot, Alain Hétet. Automated Segmentation of SAS Images using the Mean - Standard Deviation Representation. MTS/IEEE Oceans'03 conference, 2003, San Diego, United States. pp.2155 - 2160. ⟨hal-00086795⟩