基于纹理信息的Landsat-8影像植被分类初探  

A preliminary study on vegetation classifi cation based on texture information of Landsat-8 images

作　　者：宁亮亮[1,2] 张晓丽[1,2] 

机构地区：[1]北京林业大学林学院,北京100083 [2]北京林业大学北京市森林培育与保护省部共建重点实验室,北京100083

出　　处：《中南林业科技大学学报》2014年第9期60-64,共5页Journal of Central South University of Forestry & Technology

基　　金：国家高技术发展研究计划(863)课题"数字化森林资源监测关键技术研究"(2012AA102001-5)

摘　　要：遥感图像植被分类一直为遥感领域的热点,对于中低分辨率的影像,传统的分类方法主要是利用影像的光谱信息,对于影像的空间信息利用较少,而事实证明遥感影像的空间信息也十分丰富。为了提高遥感影像的空间信息利用率,提取了最新的Landsat-8的空间纹理信息,结合空间纹理信息与光谱信息对遥感影像进行植被的分类。实验证明:辅以纹理的分类总体精度为84.68%和83.87%,光谱分类总体精度为82.26%,结合了空间纹理信息后的分类精度比传统的方法有明显的提高。Vegetation classification by remote sensing images has been as the hotspot in the field of remote sensing, for the low resolution images, the traditional classiifcation method mainly used spectral information of image, the images of the spatial information were used less, but the facts have proved that spatial information of remote sensing image is also very rich. In order to improve the utilization of spatial information of remote sensing image, the latest Landsat-8 spatial texture information were extracted, by combining with the spatial information and spectral information of remote sensing image texture information, the vegetation the classification in remote sensing image was carried out. The experimental results veriifed that with the texture information classiifcation, the overall accuracy were 84.68% and 83.87%, with the spectral information classification the overall accuracy was 82.26%, the classification accuracy after combining the spatial texture information was more obvious enhancement than that of traditional method.

关 键 词：纹理信息 Landsat-8 植被分类 分类精度 

分 类 号：S771.8[农业科学—森林工程]