In situ cryptography in a neuromorphic vision sensor based on light-driven memristors

Lingxiang Hu,Jiale Shao,Jingrui Wang,Peihong Cheng,Li Zhang,Yang Chai,Zhizhen Ye,Fei Zhuge

DOI: https://doi.org/10.1063/5.0185502

IF: 15

2024-01-29

Applied Physics Reviews

Abstract:Vision sensors are becoming increasingly ubiquitous, and they continuously collect, store, communicate, and process vast amount of sensitive data that are vulnerable to being stolen and misused. Existing cryptosystems based on complex cipher algorithms generally require extensive computational resources, making them difficult to use in vision sensors that have limited processing capabilities. Here, we propose and experimentally demonstrate a novel in situ image cryptography scheme based on a neuromorphic vision sensor comprising all-optically controlled (AOC) memristors. Due to the unique light wavelength and irradiation history-dependent bidirectional persistent photoconductivity of AOC memristors, a visual image can be stored, encrypted, decrypted, denoised, and destroyed within a vision sensor. A decrypted image can be encoded in situ and then accurately recognized through a memristive neural network. Encrypted and destroyed images are capable of withstanding hacking attacks even with trained neural networks. Our cryptography scheme enables complete cryptographic operations entirely on a sensor and, therefore, effectively safeguards visual information. This work provides a simple yet efficient solution to the security challenges faced by vision sensors.

physics, applied