一种基于图像识别的马铃薯叶片病害区域定位系统
By fusing data from event cameras and RGB cameras, and utilizing graph neural networks to calculate lesion probability and compensate for system latency, the problem of decreased lesion detection performance and spraying position deviation of traditional RGB cameras in high-speed and high-vibration environments is solved, achieving high-precision, low-latency lesion location and accurate spraying.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- 张北翼德农业发展有限公司
- Filing Date
- 2025-11-03
- Publication Date
- 2026-07-17
AI Technical Summary
Traditional RGB cameras suffer from reduced performance in detecting potato leaf diseases due to motion blur and low frame rate under high-speed and high-vibration conditions, and system delays cause the spraying position to deviate from the lesion area.
A heterogeneous data acquisition unit is used to fuse data from event cameras and RGB cameras. The probability of lesions is calculated through spatiotemporal feature fusion and graph neural networks. Combined with the predictive spray control unit to compensate for system delay, a precise spray valve control signal is generated.
It achieves high-precision, low-delay disease location and accurate spraying in high-speed, high-vibration environments, improving pesticide utilization and reducing environmental pollution.
Smart Images
Figure CN121121498B_ABST