Synthetic Aperture Radar Image Based Navigation Using Siamese Neural Networks

Due to its ability to capture precise topological features in obstructive meteorological environments, synthetic aperture radar (SAR) technology offers a multitude of novel applications including the possibility of developing self-reliant navigational techniques for global positioning system denied settings. To this effect, the broader aim of this chapter is to utilize image data generated by SAR to determine the location of a given system that is navigating over a specified geographical area of interest. We propose an image retrieval technique that leverages on the concept of Siamese neural network, which is an artificial neural network (ANN) often used for signature verification and face recognition. As a backbone, the network architecture is constructed based on SqueezeNet, which is a compact deep neural network that offers greater scalability compared to other popular architectures. Numerical experiments are performed and demonstrate that the proposed method can be used effectively and holds promise for navigational tasks.