Robust extraction algorithms for the IPI in sperm whale clicks : ethological application and population monitoring

• Main Author: Abeille, Régis
• Other Authors: Laboratoire des Sciences de l'Information et des Systèmes (LSIS), Aix Marseille Université (AMU)-Université de Toulon (UTLN)-Arts et Métiers Paristech ENSAM Aix-en-Provence-Centre National de la Recherche Scientifique (CNRS), Université de Toulon, Hervé Glotin
• Format: Doctoral or Postdoctoral Thesis
• Language: French
• Published: HAL CCSD 2013
• Subjects: IPI; Population monitoring; BIVIP; MOVIP; PELAGOS; Suivi de population; [INFO.INFO-OH]Computer Science [cs]/Other [cs.OH]; Physeter macrocephalus; Sperm whale
• Online Access: https://theses.hal.science/tel-01012047; https://theses.hal.science/tel-01012047/document; https://theses.hal.science/tel-01012047/file/These_Abeille.pdf

The sperm whale, Physeter macrocephalus, the largest odontocete has been exposed for years to whaling due to the presence of liquid wax located in its head (spermaceti). This species is now considered as vulnerable. Sperm whales are located in various oceans and seas around the world. These animals are able to reach the deepest depths, where they use an echolocation technique to hunt. They emit broadband clicks that are comprised of a multi-pulse structure resulting from intra-head reflexions(spermaceti). These clicks contain information about the length of the animal and on its orientation, which reside in the delays between the pulse structure within the clicks, known as the Inter-Pulse-Interval (IPI). Measurement of these IPIs are determinant in the in the global preservation and the monitoring of sperwhales population since it is highly difficult to obtain visual clues of these deep-diving mammals.Moreover, the increasing technological advances have enabled the obtention of larger amount of underwater recordings. Therefore, the use of computational methodologies to automatically analyze the sperm whales click sounds have become a necessity in order to carry out populations monitoring and marine ecosystems studies. The current literature offers a variety of methodologies to calculate the IPI. However, these methodologies suffer from a merging of the different pulse delays and thus leading to less accurate estimation of the IPI and, in turn, of the whale's size. Moreover, from this merging methods no additional information regarding the orientation of the sperm whale can be easily extracted. In this thesis work, a new methodology for accurate IPI estimation is presented without relying on the merging of the different pulse delays, but, instead, by selecting the unique candidate pulse through a combinatorial and statistical analysis resulting in a better precision in the final estimation of the sperm whale's size and providing additional information on the whale's orientation.Our methodologies are compared with ...