欢迎访问《中国农学通报》，
中国农学通报

中国农学通报 ›› 2025, Vol. 41 ›› Issue (19): 151-158.doi: 10.11924/j.issn.1000-6850.casb2025-0245

• 农业信息·科技教育 • 上一篇    下一篇

基于U-Net骨架变体的内陆盐碱地信息遥感提取

令世豪1(), 杨粉莉2, 黄博涵1, 李瑶1, 杨联安1(), 闫琳悦1, 郝贝贝1   

  1. 1 西北大学城市与环境学院，西安 710127
    2 咸阳市农业科学研究院，陕西咸阳 712100
  • 收稿日期:2025-03-31 修回日期:2025-06-05 出版日期:2025-07-05 发布日期:2025-07-10
  • 通讯作者:
    杨联安，男，1968年出生，陕西咸阳人，副教授，博士，主要从事农业遥感和智慧农业方面的研究。通信地址：710127 陕西省西安市长安区学府大道1号 西北大学长安校区，Tel：029-88308427，E-mail：。
  • 作者简介:

    令世豪，男，2004年出生，陕西宝鸡人，本科在读，主要从事地理信息系统方面的研究。通信地址：710127 陕西省西安市长安区学府大道1号 西北大学长安校区，E-mail：。

  • 基金资助:
    陕西省大学生创新创业训练计划项目“混合U-Net模型下高分卫星影像盐碱地精准识别提取”(S202410697432)

Remote Sensing Extraction of Inland Saline Land Information Based on U-Net Skeleton Variant

LING Shihao1(), YANG Fenli2, HUANG Bohan1, LI Yao1, YANG Lian’an1(), YAN Linyue1, HAO Beibei1   

  1. 1 College of Urban and Environmental Sciences, Northwest University, Xi’an 710127
    2 Academy of Agriculture Sciences of Xianyang, Xianyang, Shaanxi 712100
  • Received:2025-03-31 Revised:2025-06-05 Published:2025-07-05 Online:2025-07-10

RichHTML

7

PDF (PC)

26

摘要：

针对传统遥感分类方法依赖人工特征设计、泛化能力不足的问题，本研究应用深度学习模型进行盐碱地信息精准提取。利用Landsat 8 OLI遥感影像，采用基于U-Net深度学习模型的盐碱地信息提取方法，系统对比ResNet34、MobileNetV2_100和TF_MobileNetV3_Small_100 3种骨架在冻结与不冻结训练策略下的性能差异。实验表明，ResNet34的收敛速度、分割精度与泛化能力总体优于轻量化模型(MobileNetV2_100、TF_MobileNetV3_Small_100)，尤其是不冻结的ResNet34模型综合表现最好，盐碱地类别的分类精度为0.880、召回率为0.708、F1分数为0.785，均优于其他模型。轻量化模型在资源受限场景下表现尚可，可在计算资源有限和分割精度要求不高的情况下使用，但在复杂场景下仍需高性能骨干网络支持。不冻结模型的表现普遍优于冻结模型，在深度学习模型训练过程中调整全部参数对于提高精度和泛化能力具有重要作用。研究验证深度学习在盐碱地遥感监测中的有效性，可为盐碱地智能识别监测提供模型选型依据。

关键词: 盐碱地, U-Net, ResNet34, MobileNet, 语义分割, 迁移学习, 深度学习模型

Abstract:

To address the limitations of conventional remote sensing classification methods, such as reliance on manual feature design and poor generalization ability, this study systematically evaluated three backbone architectures (ResNet34, MobileNetV2_100 and TF_MobileNetV3_Small_100) under frozen and non-frozen transfer learning strategies, utilizing Landsat 8 Operational Land Imager (OLI) multispectral imagery and employing a saline-alkali land information extraction method based on U-Net deep learning model. The experimental findings demonstrated that ResNet34 generally showed superior convergence speed, segmentation accuracy and generalization ability in comparison to the lightweight models (MobileNetV2_100, TF_MobileNetV3_Small_100). Specifically, the non-frozen ResNet34 model achieved the optimal overall performance, with a classification precision of 0.880, recall of 0.708, and F1-score of 0.785, all of which exceeded those of other models. The lightweight model demonstrated efficacy in scenarios characterized by limited resources, which could be employed in cases where computational resources were limited and segmentation precision required was low but still necessitates high-performance backbones for complex environments. Notably, non-frozen training consistently exceeded frozen strategies, emphasizing the importance of full-parameter optimization for enhancing accuracy and generalization ability. The research not only validates the effectiveness of deep learning in the remote sensing monitoring of saline-alkali land but also provides a model selection framework for intelligent identification and monitoring of saline and saline-alkali land, offering practical guidance for ecological governance and precision agriculture.

Key words: saline-alkali land, U-Net, ResNet34, MobileNet, semantic segmentation, transfer learning, deep learning model