《计算机应用》唯一官方网站 ›› 2026, Vol. 46 ›› Issue (10): 3342-3353.DOI: 10.11772/j.issn.1001-9081.2025091184

• 多媒体计算与计算机仿真 • 上一篇    

基于多尺度特征融合与注意力增强的小麦条锈病识别方法

黄会敏1,2, 汪鹏飞1, 赵应1, 袁爱红1,2, 胡瑾1,2, 王建锋3,4()   

  1. 1.西北农林科技大学 信息工程学院,陕西 杨凌 712100
    2.农业农村部农业物联网重点实验室(西北农林科技大学),陕西 杨凌 712100
    3.作物抗逆与高效生产全国重点实验室(西北农林科技大学),陕西 杨凌 712100
    4.西北农林科技大学 植物保护学院,陕西 杨凌 712100
  • 收稿日期:2025-10-11 修回日期:2025-12-11 接受日期:2025-12-12 发布日期:2026-01-04 出版日期:2026-10-10
  • 通讯作者: 王建锋
  • 作者简介:黄会敏(1991—),女,湖北孝感人,讲师,博士,主要研究方向:图像处理、点云配准、小麦条锈病检测
    汪鹏飞(2002—),男,陕西铜川人,硕士研究生,主要研究方向:图像处理、点云配准、小麦条锈病检测
    赵应(1996—),男,河南南阳人,硕士研究生,主要研究方向:图像处理、点云配准、小麦条锈病检测
    袁爱红(1988—),男,湖北十堰人,副教授,博士,主要研究方向:机器学习、图像处理、人工智能、无监督特征选择
    胡瑾(1980—),女,陕西西安人,教授,博士,主要研究方向:农业传感器、智能决策、农业人工智能
  • 基金资助:
    国家重点研发计划项目(2024YFD1400403);国家小麦产业技术体系(CARS-03-31);杨凌示范区科技计划项目(2024NY-06)

Wheat stripe rust recognition method based on multi-scale feature fusion and attention enhancement

Huimin HUANG1,2, Pengfei WANG1, Ying ZHAO1, Aihong YUAN1,2, Jin HU1,2, Jianfeng WANG3,4()   

  1. 1.College of Information Engineering,Northwest A&F University,Yangling Shaanxi 712100,China
    2.Key Laboratory of Agricultural Internet of Things,Ministry of Agriculture and Rural Affairs (Northwest A&F University),Yangling Shaanxi 712100,China
    3.State Key Laboratory for Crop Stress Resistance and High-Efficiency Production (Northwest A&F University),Yangling Shaanxi 712100,China
    4.College of Plant Protection,Northwest A&F University,Yangling Shaanxi 712100,China
  • Received:2025-10-11 Revised:2025-12-11 Accepted:2025-12-12 Online:2026-01-04 Published:2026-10-10
  • Contact: Jianfeng WANG
  • About author:HUANG Huimin, born in 1991, Ph. D., lecturer. Her research interests include image processing, point cloud registration, wheat stripe rust detection.
    WANG Pengfei, born in 2002, M. S. candidate. His research interests include image processing, point cloud registration, wheat stripe rust detection.
    ZHAO Ying, born in 1996, M. S. candidate. His research interests include image processing, point cloud registration, wheat stripe rust detection.
    YUAN Aihong, born in 1988, Ph. D., associate professor. His research interests include machine learning, image processing, artificial intelligence, unsupervised feature selection.
    HU Jin, born in 1980, Ph. D., professor. Her research interests include agricultural sensors, intelligent decision-making, agricultural artificial intelligence.
  • Supported by:
    National Key Research and Development Program of China(2024YFD1400403);National Wheat Industry Technology System(CARS-03-31);Science and Technology Program of Yangling Demonstration Zone(2024NY-06)

摘要:

针对小麦条锈病多等级识别中背景复杂、病斑尺度差异显著以及相邻类别特征差异细微等问题,提出一种多尺度特征融合与注意力增强残差网络(MSA-ResNet)。首先,利用残差连接缓解深层网络训练中的梯度消失问题,提高模型的特征表达能力;其次,采用自顶向下的多尺度特征融合策略,通过语义对齐增强不同尺度特征的一致性,提高模型对病斑尺寸变化和形态复杂性的适应能力;最后,引入融合通道注意力与空间注意力的卷积块注意力模块(CBAM),增强病斑关键区域的特征响应,并抑制杂草、土壤等背景信息的干扰。在公开数据集Yellow-Rust-19和自建数据集上的实验结果表明,MSA-ResNet的准确率分别达到95.67%和97.17%,相较于EfficientNet-CBAM分别提高了1.67和0.84个百分点。结果表明,MSA-ResNet能够更好地满足实际应用需求,为田间小麦条锈病的多级识别提供更优的解决方案。

关键词: 小麦条锈病, 多等级识别, 多尺度特征融合, 注意力机制, 残差网络

Abstract:

To address the problems of complex backgrounds, significant variations in lesion scale, and subtle differences between adjacent categories in multi-level recognition of wheat stripe rust, a Multi-Scale feature fusion and Attention-enhanced Residual Network (MSA-ResNet) was proposed. First, residual connections were employed to alleviate the gradient vanishing during deep network training and improve the feature representation capability of the model. Second, a top-down multi-scale feature fusion strategy was adopted to enhance the consistency of features at different scales through semantic alignment, thereby improving the model’s adaptability to lesion size variations and morphological complexity. Finally, a Convolutional Block Attention Module (CBAM), integrating channel and spatial attention mechanisms, was introduced to enhance feature responses in key lesion regions and suppress interference from background elements such as weeds and soil. Experimental results on the Yellow-Rust-19 public dataset and a self-built dataset show that MSA-ResNet achieves the accuracies of 95.67% and 97.17%, respectively, outperforming EfficientNet-CBAM by 1.67 and 0.84 percentage points, respectively. These results indicate that the proposed MSA-ResNet better meets practical application requirements and provides a superior solution for multi-level recognition of wheat stripe rust under field conditions.

Key words: wheat stripe rust, multi-level recognition, multi-scale feature fusion, attention mechanism, Residual Network (ResNet)

中图分类号: