论文部分内容阅读
本文综述了国际遥感分类研究,使用Landsat7ETM+遥感数据和地理辅助数据,应用BP神经网络方法,将莽汉山林场作为研究区进行了遥感影像的分类研究。比较了BP神经网络分类与最大似然、简单和复杂非监督分类法之间的类型与数量精度。BP神经网络分类的总类型精度是70.5%,总数量精度为84.65%,KAPPA系数是0.6455。结果说明BP神经网络的分类质量优于其他方法,其总的类型精度与其他三种分类方法相比分别增加了10.5%、32%和33%,总的质量精度增加了5.3%。因此,辅以地理参考数据的BP神经网络分类可以作为一种有效的分类方法。
This paper summarizes the research on international remote sensing classification, uses Landsat7ETM + remote sensing data and geography aided data, and applies BP neural network method to study the classification of remote sensing images in Manghan Mountain Forest Farm. The type and quantity accuracy between BP neural network classification and maximum likelihood, simple and complex unsupervised classification methods are compared. The total type accuracy of BP neural network classification is 70.5%, the total number accuracy is 84.65% and the KAPPA coefficient is 0.6455. The results show that the classification quality of BP neural network is superior to other methods, the total type accuracy of which is increased by 10.5%, 32% and 33% respectively compared with the other three classification methods, and the total mass accuracy is increased by 5.3%. Therefore, BP neural network with georeferenced data can be used as an effective classification method.