Multi-Beam/Multi-Aspect image processing for AUV application

Conventional side scan sonars are generally single beam systems with a small horizontal and a large vertical beam width. The side scan sonar image of a scene is generated by stringing together the signals from successive pings. In contrast, Multi-Beam/Multi-Aspect side scan sonars create an image of a large scene with each single ping. Here, the fact is exploited that a transducer insonifies a horizontally and vertically large area. The insonified region is then sampled with a multitude of horizontally steered receiving beams. The successive sonar images represent overlapping areas which can be fused to one image. Straightforward superposition of the single ping sonar images results in incoherent processing SNR gain for the fused image. By applying image processing methods and sophisticated fusion strategies, the SNR can be increased further. For implementation in autonomous underwater vehicles these algorithms have to run fully automatically. The basic procedures proposed for automatic Multi-Beam/Multi-Aspect image processing presented here are segmentation of the single ping sonar images and subsequent fusion of the single ping sonar images.