[O16] Target Detetction in Clutter for Sonar Imagery
Target detection in in sonar imagery has been widely researched in the past. Yet, the performances achieved are operationally unacceptable for all but the most benign seabed types. Because of the limited resolution of existing imaging sonars and the strong effect of speckle on target scattering, most algorithms have focused on the acoustic shadow casted by the target when illuminated by the sonar. These techniques fail when the seabed is complex (3D texture). Recent developments in sonar sensors have significantly improved the resolution (synthetic aperture sonars, SAS) and the introduction of interferometric systems enables the joint recovery of imagery and 3D. This project has the following key objectives:
1- Develop robust models of 3D textures in sonar imagery. This will be based on the use of ground truthed data from interferometric side scan and simulation tools developed in Heriot-Watt University
2- Develop context adaptive detection and classification algorithms. Once clutter models are understood, parameters estimation can be tackled and the models integrated into the detection & classification algorithms.
3- Review generative model based classification using SAS data.
Project Supervisor
I am a Professor at Heriot Watt University. My main areas of interest are image understanding, sensor fusion and underwater robotics. I am and active member of the Oceans Systems Laboratory and the Signal And Image Group. Finally I am a director ofSeeByte Ltd, a Spin-out of the Oceans Systems Laboratory commercialising some of the technologies developed in the Oceans Systems Laboratory. I hold an Engineering degree in Telecommunications with a specialisation in Image and signal processing from ENSTBr. I also have a M.Sc. in optics and signal processing and a Ph.D. in image processing.
