Environmental Data Fusion for Mine Warfare

"Effective countermeasures against enemy mines in the littoral require comprehensive knowledge of the undersea environment. To compile an accurate and comprehensive model of the undersea environment these data must be fused effectively. There are manychallenges associated with these data fusion tasks.In order to fuse images or maps the images and/or maps must be registered against each other. Without an accurate registration successful fusion cannot take place. Often the metadata may not hold all the measurements required for registration, or there maybe unknown biases present in the metadata. In addition the images and maps may contain nonlinear distortions that make registration a challenging task.Our proposal is to use graph matching techniques in order to achieve the required registrations between imagery and/or maps. Graph matching has been the subject of considerable interest in the image processing literature over the last five years and hasbeen shown to be an effective method for image registration. Graph matching techniques are unaffected by bias in position, orientation and scale. In addition the techniques are robust against non-linear bias and measurement error. The first application ofgraph matching would be in the registration of sea mine survey maps. Success in registration of sea mine survey maps would lead to a number of benefits that can also be generalized to success in the registration of undersea imagery. Primaril