Use of the minimum variance beamformer in synthetic aperture sonar imaging

Abstract

In synthetic aperture sonar (SAS) imaging, successive pings are combined coherently to synthesize a long aperture. For each ping, the received data is delayed, weighted and summed coherently for each pixel in the image, before images from all pings are summed together. The process is known as conventional beamforming. In reason years, adaptive beamforming for active sonar imaging has been promoted. From the same data, improved azimuth resolution and less sidelobe artifacts are reported for these approaches compared to conventional beamforming. In multi-aspect SAS imaging and in situations where the necessary accuracy in navigation or knowledge of the imaging environment are not known, a reduced length synthetic aperture is used. In such situations benefits could be achieved using an adaptive beamformer. We have used the minimum variance beamformer to process the data for each ping before summing only a limited synthetic aperture (1/10 to 1/5 of a full-length aperture). Adaptive weights for pixels for one ping are estimated from data for several neighbour pings, and subarray averaging is used within each physical aperture. Results show that sidelobe artifacts, which spoil the conventional low-resolution SAS images, are greatly reduced for the adaptive images, and images of acceptable quality are produced.