Bioinspired in-sensor visual adaptation for accurate perception

Charge trapping mechanisms in molybdenum-disulfide-based transistors can be used to mimic the adaptive behaviour of human eyes, allowing vision sensors to be created with high dynamic range. Machine vision systems that capture images for visual inspection and identification tasks have to be able to perceive a scene under a range of illumination conditions. To achieve this, current systems use circuitry and algorithms that compromise efficiency and increase complexity. Here we report bioinspired vision sensors that are based on molybdenum disulfide phototransistors and exhibit time-varying activation and inhibition characteristics. Charge trap states are intentionally introduced into the surface of molybdenum disulfide, enabling the dynamic modulation of the photosensitivity of the devices under different lighting conditions. The light-intensity-dependent characteristics of the sensors match Weber's law in which the perceived change in stimuli is proportional to the light stimuli. The approach offers visual adaptation with highly localized and dynamic modulation of photosensitivity under different lighting conditions at the pixel level, creating an effective perception range of up to 199 dB. The phototransistor arrays exhibit image contrast enhancement for both scotopic and photopic adaptation.