一种融合空间多尺度感知与环境解耦的作物病害诊断方法
By combining large-kernel deep convolution and small-kernel convolution, along with channel attention and adversarial training, environmental information is extracted from disease features, solving the problem of insufficient generalization performance of UAV remote sensing crop disease identification technology under multiple environmental conditions, and achieving stable disease diagnosis and environmental monitoring.
CN122176583BActive Publication Date: 2026-07-17SICHUAN SHUSHENG INTELLIGENT TECHNOLOGY CO LTD
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- SICHUAN SHUSHENG INTELLIGENT TECHNOLOGY CO LTD
- Filing Date
- 2026-05-11
- Publication Date
- 2026-07-17
Smart Images
Figure CN122176583B_ABST
Abstract
本发明公开了一种融合空间多尺度感知与环境解耦的作物病害诊断方法，属于农业遥感图像处理技术领域。通过大核深度卷积对遥感影像进行广域空间感知，生成编码全局环境背景信息的空间权重图；随后将空间权重图重塑为动态卷积核，驱动小核分组卷积在局部邻域内执行上下文自适应的细粒度病害纹理提取，实现跨尺度特征协同建模。将深层特征分解为病害特征与环境特征，通过梯度反转层对抗训练迫使病害特征剥离环境干扰，并结合互信息最小化与反事实增强学习消除环境伪相关性，最终获得对环境变化鲁棒的病害诊断结果。本发明有效解决了现有技术在变环境条件下泛化性能不足的问题，在多时相、多地块农业遥感监测场景中具有显著应用价值。
Need to check novelty before this filing date? Find Prior Art