基于高分一号影像的土地覆被分类方法初探

发布时间：2018-05-11 00:02

本文选题：高分一号 + 土地覆被分类　；参考：《干旱区地理》2016年01期

【摘要】：高分一号是我国发射的第一颗高分辨率卫星,其包含地物信息较为丰富,较多的应用于土地覆被分类中,但高分影像普遍存在基于像元分类精度稍低的问题,为了提高遥感影像的分类精度,基于高分一号影像,以新疆艾比湖湿地保护区为研究样区进行土地覆被分类研究。利用灰度共生矩阵方法提取图像的纹理信息,并将结果作为参数量输入到支持向量机(SVM)分类器中,将研究结果与传统的SVM分类及最大似然分类法作对比分析可得:辅以纹理特征的SVM分类方法可更好的区分地物信息,分类精度高达93.64%;传统的SVM分类精度为92.27%;最大似然分类为87.90%;因地制宜的开展辅以纹理特征的SVM分类方法是提高土地覆被监测精度的有效手段。
[Abstract]:Gaofan-1 is the first high-resolution satellite launched in China, which contains abundant information of ground objects and is widely used in land cover classification. However, the accuracy of classification based on pixel is generally lower in high-score images. In order to improve the classification accuracy of remote sensing images, the land cover classification of Ebinur Wetland Reserve in Xinjiang was studied on the basis of Gaofen No. 1 image. The grayscale co-occurrence matrix method is used to extract the texture information of the image, and the result is input into the support vector machine (SVM) classifier as a parameter. Comparing the results with the traditional SVM classification and maximum likelihood classification, we can get that the SVM classification method with texture features can better distinguish the feature information. The classification accuracy is as high as 93.64; the traditional SVM classification accuracy is 92.27; the maximum likelihood classification is 87.90; according to local conditions, it is an effective means to improve the accuracy of land cover monitoring by developing the SVM classification method supplemented with texture features.
【作者单位】：新疆大学资源与环境科学学院绿洲生态教育部重点实验室;
【基金】：自治区科技支疆项目(201504051064) 高分辨率对地观测重大专项(民用部分)(95-Y40B02-9001-13/15-03-01) 自治区重点实验室专项基金(2014KL005)
【分类号】：P237

【参考文献】