Operation speed tuning for infrared computer vision testing with pixel integration structures of carbon nanotube photo-thermoelectric imagers

This work demonstrates real-time non-destructive monitoring satisfying material identification and structure reconstruction for multi-layered three-dimensional objects. This system conducts visual-hull (a representative computer vision) under broadband transmissive infrared-irradiation with a large-area-view camera. Visual-hull acquires two-dimensional silhouette of targets in different viewing-angles for reverse-projection, and hollows overlapped bodies out from the reference voxel into three-dimensional reconstruction. The camera comprises sensitive photo-thermoelectric carbon-nanotube pixels and aggregates broadband infrared information from targets. This work non-destructively visualizes constituents (glass, metal, semiconductor, and polymer) of opaque multi-layered three-dimensional objects at a 976 nm–10.3 µm wavelength range in a 16-seconds-speed and a 6 %-error ratio.