摘 要
随着遥感技术的快速发展，高分辨率遥感影像在土地利用分类中的应用日益广泛，然而单一传感器获取的影像往往难以同时满足空间、光谱和时间分辨率的需求，这限制了分类精度的进一步提升为此，本研究以提高土地利用分类精度为目标，深入探讨了遥感影像融合技术在多源数据整合中的关键作用通过引入先进的融合算法，如基于小波变换和稀疏表示的方法，并结合传统与深度学习框架，构建了一套完整的影像融合与分类流程具体而言，研究首先对多源遥感影像进行预处理，包括辐射校正、几何配准等步骤，随后采用多种融合策略提取影像的空间与光谱特征，并将其输入至优化后的分类模型中经实验验证，所提出的方法在典型研究区域的土地利用分类任务中表现出显著优势，分类精度较传统方法平均提升了8%以上此外，该方法在复杂地物类型识别方面展现了较强的鲁棒性，尤其对于光谱特性相似的地类区分效果更为明显本研究的主要创新点在于将深度学习与传统融合技术有机结合，实现了多源遥感数据的优势互补，为高精度土地利用分类提供了新思路研究成果不仅有助于推动遥感影像融合技术的发展，还为自然资源管理、生态环境监测等领域提供了重要的技术支持
关键词：遥感影像融合；土地利用分类；深度学习
ABSTRACT
With the rapid development of remote sensing technology, high-resolution remote sensing images are increasingly applied in land use classification. However, images acquired by a single sensor often fail to simultaneously meet the requirements of spatial, spectral, and temporal resolutions, which restricts further improvement in classification accuracy. To address this issue, this study aims to enhance land use classification accuracy by exploring the critical role of remote sensing image fusion technology in multi-source data integration. By introducing advanced fusion algorithms such as wavelet transform-based and sparse representation-based methods, and combining traditional approaches with deep learning fr ameworks, a comprehensive workflow for image fusion and classification was constructed. Specifically, the study first conducted preprocessing on multi-source remote sensing images, including radiometric correction and geometric registration, followed by the adoption of various fusion strategies to extract spatial and spectral features from the images, which were then fed into an optimized classification model. Experimental results demonstrated that the proposed method exhibited significant advantages in land use classification tasks in typical study areas, with an average improvement in classification accuracy exceeding 8% compared to traditional methods. Moreover, the method showed strong robustness in identifying complex land cover types, particularly excelling in distinguishing land classes with similar spectral characteristics. The primary innovation of this study lies in the organic combination of deep learning and traditional fusion techniques, achieving complementary advantages of multi-source remote sensing data and providing new insights for high-precision land use classification. The research findings not only contribute to the advancement of remote sensing image fusion technology but also offer crucial technical support for natural resource management, ecological environment monitoring, and related fields.
Keywords: Remote Sensing Image Fusion; Land Use Classification; Deep Learning
目 录
摘 要 I
ABSTRACT II
第1章 绪论 1
1.1 土地利用分类的研究背景与意义 1
1.2 遥感影像融合技术的发展现状 1
1.3 提升分类精度的关键方法综述 1
1.4 本文研究方法与技术路线 2
第2章 遥感影像融合技术基础 3
2.1 遥感影像融合的基本原理 3
2.2 常见遥感影像融合算法分析 3
2.3 融合技术对分类精度的影响机制 4
2.4 不同融合方法的适用性评估 4
2.5 融合技术在土地利用中的应用现状 5
第3章 分类精度提升的技术路径 6
3.1 数据预处理对分类精度的作用 6
3.2 特征提取与优化方法研究 6
3.3 融合后影像的质量评价标准 7
3.4 精度提升的关键影响因素分析 7
3.5 技术路径的验证与改进策略 8
第4章 实验设计与结果分析 9
4.1 实验数据选取与处理方法 9
4.2 融合算法的对比实验设计 9
4.3 土地利用分类精度评估指标 10
4.4 实验结果分析与讨论 10
4.5 精度提升效果的总结与反思 11
结论 12
参考文献 13
致 谢 14
