论文部分内容阅读
建筑物提取是高分辨率光学遥感图像理解和目标识别的重要研究方向。实现自动化、智能化、可靠准确的遥感图像建筑物提取对基础地理数据获取和更新具有重要的应用价值和现实意义。在概述高分辨率光学遥感图像建筑物提取研究现状的基础上,综述了当前建筑物提取的主要思想和方法,将主流的建筑物提取代表性方法分为自底向上数据驱动方法(Data-driven)和自顶向下模型驱动方法(Model-driven)两大类,在综合比较评述各方法特性的基础上,对该领域仍然存在的问题和研究方向进行了分析和展望。
Building extraction is an important research direction for high resolution optical remote sensing image understanding and target recognition. Realization of automated, intelligent, reliable and accurate remote sensing image building extraction has important application value and practical significance for obtaining and updating basic geographic data. This paper summarizes the current research status of building extraction in high resolution optical remote sensing images, summarizes the main ideas and methods of current building extraction, and divides the representative methods of mainstream building extraction into data-driven ) And top-down model-driven (Model-driven) two categories, a comprehensive comparison and review of the characteristics of each method based on the field of the existing problems and research directions are analyzed and prospects.