Submarine port egress and ingress navigation using video processing

The U.S. Navy is interested in video-based navigation to ingress and egress harbors for their submarines. The implementation of such a navigation solution requires several components of image processing. One of these components is image tracking when the image being tracked is changing orientation and size. To implement this tracking, an image correlation technique that adjusts the template has been investigated as part of an undergraduate research effort. The ability to track an object, or target, while it is moving in the field of view allows for a navigation solution to be computed. The approach used is based on an image-correlation technique known as a phase-only filter. Standard image correlation utilizes the frequency domain by use of two-dimensional complex Fourier transforms. In phase-only filtering, the magnitude is normalized and results use only the phase information, which has shown to provide clearer identification than the standard fast-Fourier transform approach. In this research, phase-only filtering correlation is expanded to video imaging and used to track objects through a series of frames. As the image in the scene changes, the template is updated. In this application, the target tracking in the video image is performed in real time. The solution can be implemented to provide real-time results while accurately measuring and tracking the path of an object through a scene.