基于高分辨率遥感数据传统民居建筑提取

Automatic Extraction and Interpretation of Ancient Village Buildings Based on High-Resolution Remote Sensing Data

摘要: 传统民居建筑是一种重要的人工地物，是适应地方环境以及地方条件的有机产物，它的自动提取以及更新对传统民居建筑的保护和传承具有重要的意义。由于不同分辨率下传统民居建筑呈现了不同的形态和特征，因此，深入探索在高分辨率影像环境中提取传统民居建筑的方法，不仅丰富了理论研究的内容，也对实际应用具有重要的指导意义，本论文聚焦于传统民居建筑所独有的光谱属性与几何形态特征，通过创新性地应用一种建筑物分割算法，旨在精确地从复杂背景中识别并提取出目标建筑物，然后对分割后的影像进行非监督分类，分类精度达到82%，取得了比较好的分类效果，提取出建筑物不同的纹理信息，对获得的纹理信息导入ArcGIS中进行矢量化分析，并且对获得的数据导出表进行分析，得出结论，并且提出一些关于保护传统民居建筑的建议与方案。

Abstract: As a testament to local culture and ecological wisdom, traditional residential architecture embodies unique social and historical values. This paper focuses on the application of high-resolution remote sensing imagery technology in the automatic extraction and analysis of traditional dwellings, aiming to foster their preservation and legacy. Recognizing the disparities in morphology and features of traditional houses under varying resolution images, this study specifically explores the potential of high-resolution remote sensing data. Through an in-depth analysis of the spectral and geometric characteristics of traditional dwellings, we devised and implemented an effective building segmentation algorithm, successfully achieving precise extraction of targeted structures. Following this, unsupervised classification techniques were employed to process the segmented images, resulting in a classification accuracy of 82%, significantly enhancing the recognition rate of architectural textural information. The texture data thus obtained were analyzed through vectorization on the ArcGIS platform, revealing spatial distribution patterns and evolutionary trends of traditional residences. Based on these findings, we not only substantiated the effectiveness of our methodology but also put forward concrete strategies for the conservation and sustainable development of traditional dwellings. This research introduces a new technological perspective to the field of cultural heritage protection, demonstrating the vast potential of high-resolution remote sensing imagery in the study of traditional residential architecture.