Artificial vision system based on porosity heterojunction for in-sensor processing and dynamic trace extraction

Semiconductor heterostructures can be used to explore band structures for creating optoelectronic devices. In this work, we demonstrate an all-optical-driven artificial vision system based on a porous silicon heterojunction with in-sensor processing capabilities. The abundant surface states of porous silicon enable light trapping and photoelectric conversion, which allow the sensor to exhibit high sensitivity and detect weak light signals. By utilizing differential optoelectronic signals, the integrated device can perform multiple functions within a single unit, such as in-sensor encoding and trajectory tracking of dynamic optical signals. The neuromorphic vision systems based on this photoelectric retina can perform optoelectronic modulation, trajectory reconstruction, and handwritten text recognition and can emulate functions of the human retina with temporal and spatial resolution. The design strategy here can be used to implement in-sensor processing of dynamic optical signals in machine vision systems.