Machine vision systems face significant challenges in accurately extracting critical features from dim objects under complex scenarios. Here, we demonstrate a ferroelectric-configured weight-reconfigurable photovoltaic device array for in-sensor dynamic computing, enabling robust recognition of dim objects. A series of 2D perovskite ferroelectric nanoplates with controllable size, high crystallinity, and excellent yield are directly synthesized. Reconfigurable and nonvolatile photovoltaics in a graphene/ferroelectric/graphene heterostructure are modulated through switchable ferroelectric polarization. Leveraging the ferroelectric-configured photoresponsivity, a convolution kernel optoelectronic sensor array with dynamic correlation of adjacent units is designed for in-sensor dynamic computing. Compared with traditional static optoelectronic convolution processing, our approach selectively amplifies subtle differences of local image pixels, enabling effective edge feature extraction even in low-contrast scenes. Integrated with a convolutional neural network, the system significantly enhances the robustness and accuracy of dim object detection, offering a promising platform for advanced machine vision applications.
Chenhao Xu,Wen Wang,Zewen Li,Zongbo Feng,Taiyuan Chai,Yongxu Hu,Lei Zheng,Yuchan Wang,Wenping Hu
The tetrachromatic-inspired neuromorphic visual memristor (NVM) provides an opportunity to enhance human color perception and discrimination capabilities. However, current researches predominantly focus on UV sensing under high-power conditions and memory functionalities, often neglecting practical capabilities such as image perception enabled by weak UV signals and spatial-temporal pre-processing. Here, to address it, an all-in-one NVM based on a novel C 6 -DPA:Cs 3 Bi 2 I 9 /PMMA (C6-CP) organic/inorganic perovskite semiconductor bulk heterostructure is developed, fabricated using a scalable and low-temperature solution process. At a 1:6 volume ratio, it exhibits unique UV-modulated multi-channel conductive filaments, resulting in excellent neuromorphic synaptic performance, as well as an ultralow UV detection limit of ≈0.41 µW cm −2 and a high specific detectivity ( D * ) of 4.2 × 10 14 Jones. This represents a 1000-fold improvement over current UV-excited memristors. Moreover, the device uses its differential response to UV intensity variations to mimic the butterflyfish bio-retina, enabling grayscale detection and image extraction even with 50% noise interference. Subsequently, additional integration of multi-level edge-enhanced convolutional kernels allow C 6 -CP-based array devices to precisely control spatio-temporal synchronization between event detection and grayscale sensing, enabling accurate extraction of fish movement trajectories under low-light conditions. This work improves multi-dimensional image processing under complex lighting conditions for intelligent NVM systems.
Yang Yaqian,Ran Wenhao,Li Ying,Chen Yancheng,Chen Di,Shen Guozhen
Biological vision system-inspired optoelectronic synapses integrate sensing, memory, and processing for external information perception. However, most efforts focus on spatial expansion while overlooking critical dimensions like polarization and temporal evolution, which are critical for information extraction in complex environments. Inspired by the polarization-sensitive properties of kingfisher vision, we develop a polarization-sensitive optoelectronic synapse array device based on PEA 2 SnI 4 microwires array. Their anisotropic properties ensure polarization recognition, achieving a dichroic ratio of 1.38. And the asymmetric electrode designs create differentiated contact barriers, facilitating efficient charge storage and erasure under low power consumption. By employing four polarization-state-dependent convolutional kernels, the device demonstrates edge extraction capabilities even under 50% salt pepper noise. Furthermore, it enables high-precision in-sensor reservoir computing, with 100% accuracy in extracting fish trajectories under complex light environments. This work demonstrates motion perception in complex environments and provides a foundation for developing multi-dimensional, time-resolved visual systems for intelligent sensing and recognition.
危险品化学品经营许可证（带存储）