Logic‐Native Ferroelectric Transistor Arrays for Neuromorphic Edge Vision
Open Access
- 15 September 2026
- journal article
- research article
- Published by Wiley in Advanced Functional Materials
- Vol. 36 (86) , e78461
Abstract
Neuromorphic edge vision seeks to reduce data movement by extracting compact and decision‐relevant features close to the sensor. Most in‐sensor and in‐memory hardware, however, has been designed around synaptic weighting and multiply‐accumulate operations, whereas early visual perception also relies on local comparison, thresholding, and Boolean decisions. Here, we present a topology‐reconfigurable ferroelectric field‐effect transistor array for logic‐native front‐end visual computing. Remanent ferroelectric polarization programs reconfigurable in‐plane junctions in a van der Waals heterostructure, enabling dual‐mode operation as programmable transistors or non‐volatile logic‐memory elements. This device‐level programmability supports functionally complete Boolean operations in a compact two‐cell unit, while transistor‐mode cells act as reconfigurable interconnects for cascaded and parallel logic execution. Using this architecture, we implement hardware logic convolution that maps local spatial correlations into Boolean operations for in situ feature extraction. The logic‐processed outputs are intrinsically binarized, reducing data bit‐width and computational complexity while preserving discriminative visual information. The system achieves ultralow energy consumption (≈2.2 fJ per operation), improves CIFAR‐10 classification accuracy from 93% to 98% under the tested pipeline, and reduces computational cost by more than 50 times. This work establishes a complementary neuromorphic hardware primitive for edge vision, extending device‐level intelligence beyond synaptic MAC toward reconfigurable logic‐based visual preprocessing.Keywords
Funding Information
- National Key Research and Development Program of China (2025YFC3608500)
- National Natural Science Foundation of China (62274119)
This publication has 19 references indexed in Scilit:
- Toward Image Processing and Reservoir Computing: Neuromorphic Visual System Enabled by Nanotube‐Confined TelluriumAdvanced Functional Materials, 2025
- Flexible Neuromorphic Electronics for Wearable Near‐Sensor and In‐Sensor Computing SystemsAdvanced Materials, 2025
- Brain tumor feature extraction and edge enhancement algorithm based on U-Net networkHeliyon, 2023
- Different Thresholding Techniques in Image Processing : A Review2023
- A Morphological Image Preprocessing Method Based on the Geometrical Shape of Lesions to Improve the Lesion Recognition Performance of Convolutional Neural NetworksIEEE Access, 2022
- Non‐Volatile Electrolyte‐Gated Transistors Based on Graphdiyne/MoS2 with Robust Stability for Low‐Power Neuromorphic Computing and Logic‐In‐MemoryAdvanced Functional Materials, 2021
- A 1/2.8-inch 24Mpixel CMOS image sensor with 0.9μm unit pixels separated by full-depth deep-trench isolation2018
- Memory leads the way to better computingNature Nanotechnology, 2015
- 6.1 A 1/1.7-inch 20Mpixel Back-illuminated stacked CMOS image sensor for new imaging applications2015
- A Detailed Review of Feature Extraction in Image Processing Systems2014