Low frequency, downward-looking synthetic aperture sonar systems have been developed to detect unexploded ordnance (UXO) by ensonifying a region with pulses of sound and sensing the echoes on one or more hydrophones. Echoes are reconstructed using delay-and-sum (DAS) algorithms into imagery that is a spatial map of the relative scattering strength within a resolution cell. UXO is difficult to detect in these images when they appear at similar intensity to the background. Existing sonar reconstruction algorithms like DAS, however, do not leverage all the available information in the acoustic returns. This research developed and demonstrated alternative acoustic image reconstruction algorithms for munitions response sonar system that exploit differences in the spatial coherence of the acoustic echoes to improve image quality. 

A delay multiply and sum (DMAS) image reconstruction algorithm was developed for low frequency, downward-looking synthetic aperture sonar. While similar approaches have been used in biomedical ultrasound imaging, these existing algorithms cannot be directly applied to sonar data. Instead, this research adapted these concepts for sonar imaging by considering the wave propagation of higher-order statistical properties of the acoustic field. DMAS reconstruction was applied to field data from the Sediment Volume Search Sonar, collected from a target field in Foster Joseph Sayers Reservoir in Howard, PA on November 8, 2019. Thirteen targets, spanning a range of size, shape, water depth and burial depth were used in the analysis. The DMAS reconstructed images were compared to conventional DAS reconstructed images using the same raw acoustic data. Improvement in the imagery was compared quantitatively using two metrics, contrast and the generalized contrast to noise ratio (gCNR), as well as qualitatively. 

Qualitatively, it is much easier to identify the target in the DMAS imagery because it appears more clearly against the background of the lakebed. Quantitatively, DMAS reconstruction resulted in increased contrast and gCNR compared to DAS. Both metrics relate to the detectability of objects in the image, where higher values indicate easier detection of targets. This result indicates that using DMAS reconstruction can improve the probability of detection of UXO during munitions response sonar surveys. 

Spatial coherence-based sonar image reconstruction, using a delay multiply and sum algorithm, can help to detect UXO in difficult-to-image scenarios by leveraging information that was previously ignored. This algorithm can be applied to existing munitions response acoustic sensors without change to their hardware or operation. The approach may be especially helpful to more clearly identify UXO when they are buried near the sediment-water interface and compete for contrast with strong scattering from the seabed. The resulting improvements make it easier to observe the shape and pose of the object against the background. Theoretical considerations suggest that DMAS reconstruction will outperform DAS and may be preferrable for a detection, classification, and decision-making pipeline. (Project Completion - 2026)