Superresolution for ultrasonic imaging in air using neural networks

Ultrasonic imaging in air using an array of transducers is studied. The authors describe a superresolution technique that uses the fact that most surfaces act as perfect reflectors to ultrasonic pulses in air to generate accurate maps for object identification. The technique involves the minimization of a quadratic objective function subject to a nonlinear equality constraint. The authors show that this minimization can be accomplished by a two-step penalty function method, which, although not practical on a general-purpose computer, can operate in real time on a pair of neural networks. Results demonstrate that the technique generates accurate surface maps even with low receive signal-to-noise ratios.<>