Bat-G2 Net: Bat-Inspired Graphical Visualization Network Guided by Radiated Ultrasonic Call

Abstract

In this paper, a noise-immune Bat-inspired Graphical visualization network Guided by the radiated ultrasonic call (Bat-G2 net) that can reconstruct 3D shapes of a target from ultrasonic echoes is presented. The Bat-G2 net achieves noise-resiliency by emulating bat's auditory system that processes echoes along with the highly correlated radiated ultrasonic call (RUC). In order to extract the information contained in the echoes robustly and effectively, two implementation ideas have been applied to the Bat-G2 net: (1) RUC-guided attention, and (2) non-local attention. The Bat-G2 net is trained with ECHO-4CH dataset acquired by a custom-made Bat-I sensor. Noise-resistant property of the Bat-G2 net is demonstrated by comparing the reconstructed images with those from current state-of-the-art ultrasonic image reconstruction network under low SNR conditions. This study clearly demonstrates the implementation feasibility of the new modality of 'seeing by hearing' in practical environments.

Keywords

• 3D reconstruction
• biologically inspired vision
• deep learning: applications
• methodology and theory
• graphics
• vision applications and systems