ABSTRACT Acoustic images have been widely used across diverse disciplines, ranging from medical to industry to military applications. Existing mechanisms always require multiple signals, which are either single‐measured by a complicated phased array in real‐time or multiple‐measured by a single sensor with a time delay. To overcome the long‐standing contradiction between the spatial and temporal dimensions in imaging systems, here we propose a mechanism that enables acoustic imaging from a single signal, based on a synthesized frequency codec achieved through the combination of artificial material and artificial intelligence (AI). The artificial material, with its inherent randomness and dispersion, mimics multiple masks in the spectral domain to produce scattered waves at synthesized frequencies. These waves, as analytically proven, carry sufficient geometric information about the object to enable complete acoustic imaging. Such spectral fingerprints are then decoded in real‐time by the AI, with no prior knowledge of material parameters. Thanks to this symbiosis, we demonstrate experimentally high‐precision reconstruction of complex objects based on a single signal captured by a single fixed sensor, which breaks the limitations in both spatial and temporal dimension simultaneously. Our mechanism may revolutionize the imaging techniques for various scenarios, such as medical diagnosis, and beyond.
Wang et al. (Thu,) studied this question.