基于忆阻器的运动信息提取与快速光流计算系统
By embedding spatiotemporal information into a memristor-based neuromorphic synaptic array for optical flow calculation, the latency and accuracy problems of existing optical flow calculation methods in real-time systems are solved, enabling fast and accurate motion information extraction, which is applicable to scenarios such as autonomous driving and drones.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- BEIHANG UNIV
- Filing Date
- 2025-06-23
- Publication Date
- 2026-07-17
AI Technical Summary
Existing optical flow calculation methods suffer from high latency, limitations of pure spatial analysis, and hardware compatibility issues in real-time systems, making it difficult to meet the real-time requirements and accuracy demands of scenarios such as autonomous driving.
A memristor-based neuromorphic synaptic array is used to embed spatiotemporal information at the hardware level to achieve rapid extraction of motion information and optical flow calculation. This includes front-end imaging, voltage conversion, neuromorphic modules, and signal processing modules. Regions of interest are selected using temporal motion cues for optical flow calculation.
It achieves ultra-low latency and improved accuracy, reducing system processing time from 0.6 seconds to 150ms, a 4-fold increase in speed, and improving accuracy by 213.5%-740.9% in some scenarios, meeting real-time requirements and reducing algorithm complexity.
Smart Images
Figure CN120707601B_ABST