Look Ma, No Hands! Next Generation Sonar Processing

  • Posted on: 13 March 2015
  • By: Matthew Zimmerman
Our new algorithms are much better at automatically detecting small targets against the seafloor (more videos).
entering a harbor with FLS
A FarSounder-1000 forward looking sonar approaching Newport Shipyard.
Look Ma, No Hands! - used under creative commons license: https://www.flickr.com/photos/29233640@N07/7762799686
Using a sonar shouldn't be as scary as riding a bicycle with no hands.
huge overlap from ping to ping
A large volume of coverage with a single ping translates to huge overlaps on each subsequent ping. FarSounder's algorithms leverage this to improve detections and stabilize images.
engineering display of a sophisticated bottom
A development display our engineers use to evaluate sophisticated sea bottom images.

One of the exciting characteristics of our 3D FLS products is that with each release of our software we unleash more and more of the sonar sensor hardware’s full potential. This means that our performance and capabilities will continue to get even better via software updates without requiring new sensor hardware. Over the past year, the FarSounder engineering team has been focusing on improving our automated sonar processing algorithms and image quality. We’re excited to announce our next generation auto-squelch and bottom detection algorithms.

Better Auto-Squelch Processing for In-Water Targets

As with previous releases, the seafloor is automatically detected out to the bottom mapping limit of the sonar and displayed as a smoothed 3D contour. Targets on the sea floor beyond the bottom mapping limit or in the water column are detected as In-Water Targets. In order to detect In-Water Targets, a detection threshold based on the signal level of the In-Water Targets must be specified. This threshold is called the In-Water Target Squelch level, and it operates similar to the squelch level on a radio: with a high squelch level, small target echoes are rejected but little noise is accepted; with a low squelch level, small target reflections are accepted but more noise is detected too.

In the past, best results were achieved by setting this threshold manually. As of SonaSoft 3.0, we recommend that users go “hands free” in regards to processor settings. The automatic detection algorithms have been completely re-engineered and Auto Squelch is now the recommended operating mode for detection of in-water targets. With Auto Squelch, the software automatically determines the best detection threshold using adaptive thresholding methods.

Occasionally, the Auto Squelch may allow too much clutter to pass through the filters. In this case, the user can still increase the squelch level manually (though they can never reduce the squelch lower than the automatically detected level). The lowest signal level detected with Auto Squelch is indicated next to the manual squelch slider and can be used as a reference when switching to manual squelch mode.

Ping-to-Ping Integration of Seafloor Detections

Because FarSounder’s products generate a 3D image with a single ping, our sonars have a huge overlap in coverage zone from one ping to the next. If the information in the overlap area can be integrated from ping-to-ping, it becomes possible to reduce false detections and improve image stability. We’ve been doing that for a while to some extent. With our latest release, this approach is raised to the next level.

In-water target stabilization was initially introduced in SonaSoft 2.6. Since then, the FarSounder development team has completely re-architected our ping-to-ping processing algorithms. As of SonaSoft 3.0, the software now applies ping-to-ping processing to both in-water targets and the seafloor. Coupled with the improved adaptive threshold methods mentioned previously, the new processing architecture results in fuller in-water targets that fluctuate very little with each ping and a more sophisticated bottom image. The effect of these improvements is a more stable image that is easier and quicker to interpret. We’ve produced a ton of new videos to highlight our next generation processing algorithms.