应用Landsat-8卫星遥感图像的青海地区土壤分类研究  

Research on Soil Classification of Qinghai Region Using Landsat-8 Remote Sensing Images

作　　者：刘志勇 黄宇飞[1] 刘佳欣 金丽花[1] 徐嘉[1] LIU Zhiyong;HUANG Yufei;LIU Jiaxin;JIN Lihua;XU Jia(Beijing Institute of Spacecraft System Engineering,Beijing 100094,China;College of Computer Science and Technology,Harbin Engineering University,Harbin 150001,China)

机构地区：[1]北京空间飞行器总体设计部,北京100094 [2]哈尔滨工程大学计算机科学与技术学院,哈尔滨150001

出　　处：《航天器工程》2020年第1期105-109,共5页Spacecraft Engineering

摘　　要：遥感技术的迅速发展和应用,为土壤分类提供了新的技术与数据支持,文章提出一种利用陆地卫星-8(Landsat-8)遥感图像的土壤分类方法,用于青海地区土壤分类,所提取的特征包括4种,即基于遥感图像的变换特征、指数特征、纹理特征和基于高程图像的地形特征,通过随机森林方法实现土壤分类。文章首先提出2个基于遥感图像的纹理特征--熵与平滑度,从中国土壤数据集中自动提取了16 444个数据点,并在试验中采用,充分验证了此分类方法的有效性。The rapid development and application of remote sensing technology provides new technical and data support for soil classification. In this paper a soil classification method using Landsat-8 remote sensing images is proposed and applied in soil classification in a region of Qinghai. Four types of features are extracted, which are transform features, index features and texture features based on remote sensing images and terrain features based on elevation images. Two new texture features based on remote sensing image are proposed in this paper, which are entropy and smoothness. 16444 data points are automatically extracted from the Chinese soil dataset. More data points are used in the experiment, which verifies the effectiveness of the proposed method.

关 键 词：土壤分类 陆地卫星-8遥感图像 纹理特征 随机森林 

分 类 号：V19[航空宇航科学与技术—人机与环境工程]