摘 要
城市扩张是全球城市化进程中的重要现象，其对土地利用、生态环境及社会经济发展具有深远影响。遥感技术凭借其高时空分辨率和大范围覆盖能力，已成为监测城市扩张的关键手段。本研究以遥感影像分类技术为核心，结合多源遥感数据与先进的机器学习算法，旨在探索一种高效、精准的城市扩张监测方法。研究选取典型快速城市化区域为实验区，基于时间序列遥感影像，采用深度学习框架下的卷积神经网络（CNN）进行地物分类，并通过对比分析不同时期的土地覆盖变化特征，揭示城市扩张的空间格局与动态趋势。此外，研究创新性地引入了迁移学习策略，以解决遥感影像样本标注不足的问题，显著提升了模型的泛化能力和分类精度。结果表明，所提出的方法在复杂地表条件下仍能保持较高的分类准确率，有效捕捉城市扩张的细微变化。研究表明，遥感影像分类技术结合深度学习算法能够为城市规划与管理提供科学依据，其在动态监测和预测城市扩张方面的潜力值得进一步挖掘。该研究的主要贡献在于提出了适用于大规模、长时间序列城市扩张监测的技术框架，为相关领域的理论发展与实践应用提供了新思路。
关键词：城市扩张；遥感影像分类；深度学习
ABSTRACT
Urban expansion is a significant phenomenon in the global urbanization process, exerting profound impacts on land use, ecological environments, and socio-economic development. Remote sensing technology, with its high spatiotemporal resolution and extensive coverage capabilities, has become a critical tool for monitoring urban expansion. This study focuses on remote sensing image classification techniques, integrating multi-source remote sensing data with advanced machine learning algorithms to explore an efficient and precise method for urban expansion monitoring. A typical rapid urbanization area was selected as the experimental zone, where time-series remote sensing images were analyzed using convolutional neural networks (CNN) within a deep learning fr amework for ob ject classification. By comparing and analyzing the characteristics of land cover changes across different periods, the spatial patterns and dynamic trends of urban expansion were revealed. Additionally, this study innovatively incorporated transfer learning strategies to address the issue of insufficient labeled samples in remote sensing images, significantly enhancing the model's generalization ability and classification accuracy. The results indicate that the proposed method maintains a high classification accuracy even under complex surface conditions, effectively capturing subtle changes in urban expansion. This research demonstrates that the combination of remote sensing image classification techniques and deep learning algorithms can provide a scientific basis for urban planning and management, with considerable potential for further exploration in dynamic monitoring and predicting urban expansion. The primary contribution of this study lies in proposing a technical fr amework suitable for large-scale, long-time series urban expansion monitoring, offering new insights for theoretical development and practical applications in related fields.
Keywords: Urban Expansion; Remote Sensing Image Classification; Deep Learning
目 录
摘 要 I
ABSTRACT II
第1章 绪论 1
1.1 城市扩张监测的研究背景与意义 1
1.2 遥感影像分类技术的发展现状 1
1.3 本文研究方法与技术路线 2
第2章 遥感影像分类技术基础 3
2.1 遥感影像的基本特征与类型 3
2.2 分类技术的原理与算法概述 3
2.3 常用遥感影像分类方法比较 4
2.4 遥感影像预处理关键技术 5
第3章 城市扩张监测中的技术应用 6
3.1 城市扩张的遥感特征提取 6
3.2 分类技术在土地覆盖变化中的应用 6
3.3 时间序列分析在城市扩展监测中的作用 7
3.4 精度评估与误差分析方法 7
第4章 实验设计与案例分析 9
4.1 实验区域与数据选取 9
4.2 分类算法的选择与实现 9
4.3 城市扩张动态变化结果分析 10
4.4 应用效果评价与改进建议 10
结论 12
参考文献 13
致 谢 14
