摘要
遥感图像分割技术旨在将包含复杂地物空间分布信息的遥感图像划分为具有特定语义标签的不同区域,在遥感影像的分析和解译环节中有着重要应用。文章参照人类视觉系统的感知和视觉信息逐层稀疏化加工的特性,从基于浅层特征的分割方法、基于中层特征的分割方法和基于深层特征的分割方法三个层面,对常见的光学遥感图像分割技术进行了分类综述,陈述了各种主要算法的基本特点,并对比了它们的优势和局限性,同时也对未来的改进和发展方向进行了展望,为遥感图像分割的研究提供了一定的参考借鉴价值。
Remote sensing image segmentation aims to divide the remote sensing image that contains spatial distribution information of complex ground objects into various regions with specific semantic labels,playing a pivotal role in remote sensing image analysis and interpretation.Referring to the characteristics of human visual system perception and visual information layer-by-layer sparse processing,this paper systematically classifies and reviews the common optical remote sensing image segmentation technologies from three perspectives:shallow-feature-based image segmentation,middle-feature-based image segmentation and deep-feature-based image segmentation.In addition,it states the basic characteristics of various common algorithms,compares their advantages and limitations,and prospects the future direction of improvement and development,providing a certain reference value for the research of optical remote sensing image segmentation.
作者
闵蕾
高昆
李维
王红
李婷
吴穹
焦建超
MIN Lei;GAO Kun;LI Wei;WANG Hong;LI Ting;WU Qiong;JIAO Jianchao(Key Laboratory of Photoelectronic Imaging Technology and System,Ministry of Education,Beijing Institute of Technology,Beijing 100081,China;Beijing Institute of Space Mechanics&Electricity,Beijing 100094,China;Institute of Spacecraft System Engineering,CAST,Beijing 100094,China)
出处
《航天返回与遥感》
CSCD
2020年第6期1-13,共13页
Spacecraft Recovery & Remote Sensing
基金
国家自然科学基金(61875013)
装备预研航天科技联合基金(6141B061004)。
关键词
图像分割
分类
图像特征层次
深度学习
光学遥感
image segmentation
classification
image feature level
deep learning
optical remote sensing
