The Multi-Sensor Towbody (MuST) incorporates a low frequency sonar for imaging buried unexploded ordnance (UXO) and a high frequency sidescan sonar for wide area surveys and imaging proud UXO. Both systems have their relative merits; the goal of this project is to use the combined set of observed target and environment features from both systems to improve target localization, map geo-rectification, multi-aspect data combination, and to facilitate identification of key target features using the combined set of observations. Key to mapping side-scan data to the ground-plane with enough precision for multi-aspect image fusion is an accurate bathymetry model. For this reason, as well as to provide accurate burial depth information (in combination with the low frequency sonar), three-dimensional shape information, and sediment parameters, a multibeam will be incorporated onto the MuST as an additional sensor.

This project will begin by developing a post-processing library for the sidescan and multibeam sonars that can be integrated into the BOSS (Buried Object Scanning Sonar) software framework that was developed in MR18-1051. Accurate array position and Inertial Navigation System lever-arm corrections will be determined for the sidescan and multibeams, the goal being accurate (sub-pixel) alignment between sonar observations. An algorithm combining low and high-frequency observations for feature-based navigation improvement and mosaic alignment will be developed and incorporated into the MuST post-processing software framework. Following procurement of a multibeam sonar and mechanical integration onto the MuST, data from integration tests in Lake Washington and Sequim Bay will be leveraged for improving feature-based navigation. Development of multi-aspect and multi-sensor data fusion and region-of-interest data extraction will begin with 2020 Sequim Bay survey data. Parsing, post-processing, and high fidelity region-of-interest extraction for each sonar system will be combined in a unified and compiled postprocessing framework for simple third-party usage alongside operation of the MuST.

This project will enable operators or classification algorithms to access accurately aligned multi-aspect, multi-frequency observations of objects for improved feature detection and UXO identification capability. Mosaics of the surveyed environment will be improved, facilitating target and region-of-interest detection. A unified software framework will be developed, enabling end users to take full advantage of the different sensors on the MuST, and for any given detection and region of interest the full array of acoustic observations from each sensor and an estimate of sediment parameters will be made available for UXO classification.