摘要
为了解决由于云层遮挡所引起的数据利用率低等问题,提出了一种新的基于支持向量机(SVM)与无监督聚类算法相结合的分类算法,实现可见光遥感图像快速高效地自动云判别。该算法首先使用ISODATA进行聚类,再利用聚类结果为SVM挑选训练集,从而大大减少SVM的训练时间,融合了SVM准确率高与ISODATA聚类速度快的优势。结果表明:该算法使得SVM的训练时间降低至单独使用SVM算法所需训练时间的2%,基本满足实时性需求,并保证分类正确率达90%以上。
Cloud shelter in the optical remote sensing image may cause low data utilization rate and affect the subsequent process of remote sensing image such as target identification,so the research of real time and efficient cloud detection method is very important.We proposed a high speed and high accuracy classification algorithm for the cloud classification based on the combination of Support Vector Machine（SVM） and unsupervised clustering algorithm.This method uses the ISODATA clustering results to select the training set for SVM in order to reduce the training time of SVM.It takes advantage of the high accuracy capability of SVM and the fast clustering speed of ISODATA.The experiment shows that the SVM training time in the proposed method is greatly lower than it in the method using SVM alone,and the proposed method can improve the detection rate.
出处
《遥感技术与应用》
CSCD
北大核心
2012年第1期106-110,共5页
Remote Sensing Technology and Application
关键词
云判别
SVM
聚类
Cloud classification
SVM
Clustering
