Additional Modules

Open architecture tactical systems that provide single interface mission planning, situational awareness, and fast, assured access for forward deployed forces are in increasing demand in order to decrease risk and avoid unacceptable delays due to non-interoperability.

The following SeeTrack Military modules have been found to make both military and security operations quicker, safer and more cost-effective for military missions:

Automatic Target Recognition (ATR)

SeeByte's ATR system for forward-looking and side-scan sonar aims at detecting and labelling targets which are observed on a sequence of sonar frames. This ATR, comprised of a Computer-Aided Detection (CAD) and Computer-Aided Classification (CAC) components, relies on a modular architecture.

CAD is used to detect regions within the sensor data which could correspond to a possible object. SeeByte uses a suite of CAD detectors, which enables multiple CAD models to pick up the same object, thereby increasing the likelihood of the detection being real.

CAC is used to take the detections highlighted by the CAD algorithm and carrying out further processing to remove obvious false alarms in order to classify them. The ATR outputs object meta-data, such as position and size.

Change Detection

Our change detection module allows operators in the field to rapidly identify changes without the need for laborious side-by-side image comparison of current data versus historic. This technology has the capability to automatically group detections together into individual targets. This uses both the detection's absolute position and the spatial organization of the detection to ensure a correct grouping.

Sonar Mosaicing

Mosaicing produces real-time image maps of the seabed using images. SeeByte can produce accurate sonar mosaics. The sensor data is processed using the available sensor navigation data.

Seabed Classification

This module can be used to classify sonar seabed data into different regions, as well as create classified mosaics of the seabed. The classification routines in SeeByte currently use four features, seafloor height, shadow elongation and two Fourier features, to initially classify the seafloor. The segmentation result is then smoothed using a simple Markov model.