Multi-echo, multi-aspect sonar object recognition from echoes and acoustic images

Period of Performance: 05/29/2007 - 03/11/2010

$630K

Phase 2 SBIR

Recipient Firm

Chirp Corp.
8248 Sugarman Drive
La Jolla, CA 92037
Principal Investigator

Abstract

Objectives are to increase robustness in mine detection/classification against significant amounts of clutter, with probability of a correct decision vs. false alarm rate sufficiently good to be implemented in third generation search, classify and map UUV (unmanned underwater vehicle) systems. Echoes and images are used to enhance displays for operator review. The system will be less tedious for an operator than current methods, move more processing onto the UUV, and emphasize automated data processing. Other objectives are to mitigate technical risk in using broadband, multi-aspect processing, to implement a smooth transition between side scan and synthetic aperture sonar (SAS) as operating frequency is reduced, and to reduce false alarms via fusion of side scan sonar images, side scan echo features, SAS images, and multi-aspect echo features. Specially designed algorithms are used for signal/array processing, feature extraction, echo-image fusion, and sequential classification. For example, SAS imaging with a small aperture wide-beam, wideband, dolphin-like sonar depends upon two different algorithms for adaptive beam narrowing, one for echo-based motion compensation (registration) and the second for reducing SAS spotlight size. Another example is a method for image similarity measurement that can tolerate distortion, size, rotation, and location differences without loss of shape information.