论文部分内容阅读
近年来随着小卫星数量与传感器类型的快速增加,急需研究和发展快速可靠的小卫星遥感影像分类方法。针对分类方法各具局限性、具体应用中最优分类器选取困难等问题,本文基于多分类器集成学习的思路,引入随机森林(Random Forests)方法用于小卫星遥感影像分类。采用灾害监测预报小卫星(HJ-1)、北京1号小卫星(BJ-1)两种国产小卫星多光谱遥感影像进行试验,并与传统分类方法进行比较,结果表明,随机森林比最大似然分类器(MLC)、支持向量机分类器(SVM)等具有更好的稳定性、更高的分类精度和更快的运算速度,具有很好的适用性。
In recent years, with the rapid increase of the number of small satellites and sensor types, it is urgent to research and develop fast and reliable small satellite remote sensing image classification methods. Aiming at the limitations of the classification methods and the difficulties in the selection of the optimal classifiers in the concrete application, this paper introduces Random Forests method for the classification of small satellite remote sensing images based on the idea of multi-classifier integrated learning. Two domestically small satellite multi-spectral remote sensing images of HJ-1 and BJ-1 were tested and compared with the traditional classification methods. The results showed that the ratio of random forest to maximum likelihood However, MLC, SVM and so on have better stability, higher classification accuracy and faster computing speed, and have good applicability.