A new in-sensor reservoir computing system for latent fingerprint recognition with a memristor array and deep ultraviolet photo-synapses has been created by a research team guided by Prof. Long Shibing from the University of Science and Technology of China (USTC) of the Chinese Academy of Sciences, working together with Prof. Qi Liu from Fudan University.
Image Credit: vchal/Shutterstock.com
This study has been published in the journal Nature Communications.
Deep ultraviolet (DUV) photodetectors have a crucial role to play in environmental monitoring, deep space exploration, and bio-information identification. The traditional ex-situ DUV fingerprint recognition platforms use a separate sensor, processor, and memory, which considerably raises the latency in decision-making and, therefore, the total computing power.
The research team, inspired by the human visual perception system, built a DUV in-sensor RC system comprising optical synapses as the input layer of the reservoir and the memristor device array as the readout network, which can perceive and process in parallel to guarantee high efficacy and minimal power consumption.
The researchers employed the Ga-rich component design and developed amorphous GaOx (a-GaOx) photo-synapses with improved persistent photoconductivity (PPC) effects. A non-linear mapping relationship for the DUV in-sensor RC platform was built by entering 4-bit equivalent light pulses for simulation so that the image pixel sequence data could be tested for feature values.
Eventually, the training of the reservoir outputs was realized via the memristor device array’s stable polymorphic modulation properties, allowing small-scale DUV fingerprint recognition. The superior recognition accuracy of DUV fingerprint images when using a twin-feature approach and this hardware system is nearly undistinguishable from the simulated outcomes.
The system accomplishes a recognition accuracy of 100% after 100 training epochs and maintains an accuracy of 90% even in the presence of 15% background noise, in keeping with the anti-noise features of DUV light.
This completely-hardware DUV in-sensor RC system offers a good reference prototype for excellent recognition and secure applications of latent fingerprints. It is also an important reference for creating smart optoelectronic devices in the DUV waveband.
“This prototype system…will provide more insight into emerging in-sensor reservoir computing. Overall, the topic of this work is truly interesting.” said one referee for Nature Communications.
Zhang, Z., et al. (2022) In-sensor reservoir computing system for latent fingerprint recognition with deep ultraviolet photo-synapses and memristor array. Nature Communications. doi.org/10.1038/s41467-022-34230-8