论文部分内容阅读
获取了川西高原区多时相Sentinel-1星载合成孔径雷达(SAR)影像,引入了地形辐射校正技术,发展了一种阴影双视向补偿算法,综合分析后向散射系数、纹理特征、相干系数、数字高程的统计特征,采用支持向量机进行了土地覆盖分类.结果表明,对地形起伏大的高原山区,该方法能够有效提取常绿针叶林、混交林、灌木林、天然草地、水体等地物信息,分类总精度为83.62%.
A multi-temporal Sentinel-1 spaceborne synthetic aperture radar (SAR) image was acquired in the western Sichuan Plateau. A topographic radiation correction technique was introduced and a shaded double vision compensation algorithm was developed. The backscattering coefficients, texture features, coherence coefficients , The digital elevation of the statistical characteristics of the use of support vector machine for land cover classification.The results show that this method can effectively extract the evergreen coniferous forest, mixed forest, shrub forest, natural grassland, water bodies, etc. Geographical information, classification accuracy of 83.62%.