论文部分内容阅读
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。
In order to solve the problem of low utilization rate of data due to cloud cover, a new classification algorithm based on Support Vector Machine (SVM) and unsupervised clustering algorithm is proposed to realize fast and efficient automatic cloud discrimination of visible remote sensing images . The algorithm firstly uses ISODATA to cluster, and then uses the clustering result to select the training set for SVM, which greatly reduces the training time of SVM, and combines the advantages of high SVM accuracy and ISODATA clustering speed. The results show that the proposed algorithm can reduce the training time of SVM to 2% of the training time required by SVM algorithm, which can meet the real-time requirement and guarantee the correct classification rate of more than 90%.