摘要
支持向量机是一种基于统计学习理论的新型机器学习算法,它通过解算最优化问题,在高维特征空间中寻找最优分类超平面,从而解决复杂数据的分类及回归问题。随着应用面的不断扩大,支持向量机在遥感领域也得到了广泛关注。该算法已经成功的应用于遥感数据的土地覆盖、土地利用分类,多时相遥感数据的变化检测,多源遥感数据信息融合等,并且在高光谱遥感数据处理中得到了广泛应用。综述了支持向量机算法在遥感数据分类中的应用。首先对支持向量机的理论进行简要介绍,进而综述了该算法在不同遥感问题中的应用进展,最后阐述了新型支持向量机算法的发展以及在遥感中的应用。
Support Vector Machine （SVM） is a state-of-the-art machine learning algorithm based on statistical learning theory. It tries to find the optimal classification hyperplane in high dimensional feature space to handle complicated classification and regression problems by solving optimization problems. With the development of the theory and its applications, SVM has been used in remote sensing community successfully. SVM has been applied to land cover/land use classification for remotely sensed data, change detection for multi-temporal remote sensing data, and information fusion for multiple source data. Moreover, it has become a standard technique for hyperspectral data process. In this paper, the applications of SVM in remote sensing are reviewed. First, we introduced the basic theory of the SVM briefly. Then we reviewed the state of the art in different remote sensing applications. At last, we stated the development of several new SVM algorithms, which were derived from the SVM theory, and applications in remote sensing community.
出处
《地球科学进展》
CAS
CSCD
北大核心
2009年第5期555-562,共8页
Advances in Earth Science
基金
国家重点基础研究发展计划项目"陆表生态环境要素主被动遥感协同反演理论与方法"(编号:2007CB714406)
中国高技术研究发展计划项目"卫星遥感SAR与光学影像自动配准与融合技术系统研究"(编号:2007AA12Z157)
中国科学院知识创新工程青年人才领域前沿专项项目"多种地表覆盖条件下遥感邻近效应测量与校正方法研究"(编号:08S01100CX)资助
关键词
支持向量机
遥感数据分类
Support vector
Classification for remotely sensed data
Review.
